Gen III Battle Frontier Discussion and Records

Congratulations on the streak and for reaching that amazing record! Getting more than 100 wins is a massive feat on the Arena considering how difficult and unforgiving it can be. Your streak is added to the leaderboard for the years to come and hope to see more of you!!
Thank you very much! I love Battle Frontier and will definitely keep playing. Will try Open Level at some point.
The streak is done in cartridge btw so I think the notation is wrong :psynervous:
 
Thank you very much! I love Battle Frontier and will definitely keep playing. Will try Open Level at some point.
The streak is done in cartridge btw so I think the notation is wrong :psynervous:
Oops, I didn't realized before making your entry. It's fixed now!
 
Joe's Declassified Factory Survival Guide


Become a knowledgeable gen 3 battler before entering. I recommend doing this facility last. Are you familiar with the gen 3 battle mechanics and the particularly dangerous pokemon sets? You should know that thunder always lands in the rain, even if the enemy is at +6 evasion. You should know that you can revenge kill pokemon on a sliver of health just by sending out Tyranitar to summon sand before the turn ends. You should know electric types can be paralyzed and that paralysis lowers speed by 75% in this gen. If you don't fully recognize all the tools at your disposal, you'll never make it far in this facility. You are going to lose battles that you seemed guaranteed to win, and win battles you were sure you'd lose (so never throw in the towel until the battle is completely over!). Familiarize yourself with the resources that remove guesswork: spreadsheet of pokemon sets, gen 3 damage calc, and rules of the Factory.
Learn as much as you can about how the AI will act in any given situation. This will cause you to alter how to make decisions for each bracket of AI.

Rounds 1 and 2: Dumb AI: opponent randomly chooses a 1 of their 4 moves. The answer? Stay in majority of the time basing your decisions on how likely they'll randomly select moves. Switch based on how scared you are of opponent's moves. They might have that super effective move that can OHKO, but there's only a 1 in 4 chance they'll select it. Switch if you have a strong counter, but remember that even counters can get statused. However, you should never risk your wincon (unless playing passively lets them set up and beat your wincon).

Rounds 3 and 4: Moderate AI: opponent won't status twice. And the AI usually doesn't fall for type immunities (e.g. an opponent won't try to thunderbolt a ground type). So if you're already confused, you can switch in another pokemon knowing that they won't try use confuse ray again. Damaging moves are still randomly selected. The answer? Play similarly to the previous rounds, bearing in mind the AI's new aversion to certain moves. For example, if you have a Fearow out against a Golem, you could reasonably safely switch in your Lanturn, as the opponent knows not to go for earthquake.

Rounds 5+: Smart AI: the opponent will prioritize going for the KO if they know a move that guarantees the KO. They always choose the smartest damaging move, but still may go for status, weather, or (emphasis on) stat boosts. If the opponent starts setting up offensive boosts, it will *almost certainly* continue setting up until it can knock out your pokemon in front of it in one hit. I've seen a "Smart AI" Starmie not go for a surf on my Moltres because it would only do 99% (56.3% chance to OHKO). This AI seemingly favors speed control over the guaranteed KO; for example, an enemy Marowak used icy wind six times as my Tyranitar dragon danced six times. The AI also misjudges with curse. Twice now, I've been able to 3HKO a pokemon with curse AND rest because they curse twice to be able to OHKO, not considering that they will be knocked out in the process of setting up. The answer? Battle your heart out. Use every single move, ability and item available if it can help you win. Predict the AI's moves by comparing damage outputs on the damage calculator. Go slowly and think about every inkling of information on the enemy that you've received and map out the battle. Plug in every possibility, including hax. Pro Tip: play with animations to slow the game down! You do not want to miss seeing if that Arcanine had intimidate or if a water pulse confused or not, and this is a potential consequence of playing with animations off.

Noland AI: acts like Round 5+ AI, except he more heavily favors OHKO moves and has an aversion to using, or else just plain misuses, his own endure/flail strategies. Noland values set up moves differently and does not necessarily set up until he can OHKO. For example, his Dragonite1 used Brick Break to 2HKO my Tyranitar despite having the option to dragon dance and make it an OHKO. He does not always go for the highest damaging move and I suspect it's because he favors moves that have added effects. For example, turn one his Sceptile once used crush claw, instead of the stronger leaf blade, on my Zapdos (an instance where neither move would have killed me).

Every Level of Opponent AI: regardless of the round, the opponent will favor sending out a pokemon with a super effective move. Now, dumb AI might never even click it, but you can make detailed inferences if you're paying enough attention. For example, say I lead with my Moltres and kill their steel type, and they then send out Ludicolo4 (toxic, leech seed, double team and rain dance). Lets say I beat it down with aerial ace and then the opponent sends out Tyranitar third. The fact that the opponent didn't send out Tyranitar sooner tells me that it isn't running rock slide, thus allowing us to disregard 7 of the possible 10 sets it could be running. But don't get ahead of yourself; if the Ludicolo is running surf, you cannot draw any subsequent conclusions about Tyranitar. In other words, if the second enemy pokemon comes out and does not have super effective coverage to hit your pokemon, neither will the third pokemon. Always remind yourself that they chose to send out this second pokemon. The enemy AI guesses your ability if there are multiple possibilities, but it will remember your ability once it is revealed. For example, they won't thunderbolt your low health Lanturn with volt absorb if they've already seen it activate.
You must decide between Level 50 and Open Level. I think Level 50 is easier because of the static sets--who doesn't have an aversion to scouting unknown sets? Furthermore, in theory the Level 50 Round 8+ metagame is safer than Open Level without Tyranitar and Dragonite, who often sport hax items. Level 50 also excludes sets 5 and 6 of the bird and dog trio pokemon which, considering Raikou6 is the god of the gen 3 factory, is yet another factor that makes Level 50 easier. A good counterargument to this "level 50 is easier" argument lies in Level 50 Round 5, when battling with the brutal meta of the set 2 pokemon. I've somewhat generalized the Open level sets as such:

Set 1: One STAB move, status move, stat lowering move, counter users.

Set 2: Double team and other stat boosting move users, many running resto-chesto. Suboptimal sets.

Set 3: Diverse coverage moves. Ie lots of elemental punch users (although many of which are physical attackers), Earthquake Rock Slide combo, and Surf Ice Beam combo. Some explosion and OHKO users.

Set 4: "Ideal" movesets with coverage. Set up moves, OHKO moves, and only special attackers run elemental punches. Reminiscent of their ADV OU sets.

Factory Ubers Sets: Some limited to STAB moves, some run "ideal" sets, and plenty of hax/gimmick sets.

Combining the AI's intelligence with the following generalizations of every set's meta will help show the general difficulty for each round. Ignoring the all important factor of luck, the Level 50 rounds probably look something like this:

Round 1: Dumb AI is piloting the "factory little cup"'s putrid movesets with many bad moves to choose. This is an easy round.

(Tips: Charmeleon1 can mow down its counters thanks to bad AI; this also makes for a face paced round. Grass types packing toxic are pretty strong. Seek out pokemon with (4) usuable moves and items; for example, Pidgeotto1 and Masquerain1. For a sense of power creep, remember that Plusle1 is pretty good this round)

Round 2: Dumb AI has the exact same bad movesets with a few new ones thrown in. "Middle cup" should be an easy round too.

(Tips: The Charmeleon1 lead is still good. But by now I consider Plusle1 a pretty bad pokemon, considering the other evolved pokemon's bulk. So look for evolved pokemon of your own. Psychic types are great this round as long as you are conscious of the dark types (Sableye1, Cacturne1, Sharpedo1). Remember that Future Sight affects dark types, which is very relevant this round. Fighting type leads are strong and make for quick paced battles. Normal types are nice when paired with a rock and steel answer.)

Round 3: It feels like the Moderate AI has movesets tailor made for it, with many status options and solid coverage. A tricky metagame and a Noland fight? I consider this a potentially challenging round.

(Tips: Rain teams can be strong (Omastar2, Gorebyss2, Magneton2). Jumpluff2 is a run ender so familiarize yourself with it. The fast fighting lead is still strong. Chansey2 is not an auto win, but close. Grumpig2, Scyther2, and Gligar2 are some of my favorites.)

Round 4: Moderate AI and limited damaging moves means powerful STAB will be selected often, and with status options plaguing the round, it's moderately difficult. The starting draft having 12 IVs could now be enough for them to speed creep, i.e. outspeed opponents with a higher base speed stat than they have.

(Tips: Maintaining synergy is always important, but you'll likely be switching into status. Coincidentally, a lot of viable pokemon run Lum this round (Gyrados1, Espeon1, Milotic1))

Round 5: In this round, smart AI is pulling the strings of the most hated sets in the Frontier. Set 2 is home to some of the most dangerous sets (e.g. Gardevoir2 and Snorlax2) and yet the worst sets too (see Salamence2, Crobat2, Flygon2 and Porygon2-2). Consequently, it is a real race to see who can get the better set up sets.

(Tips: Toxic users can shut down a lot of pokemon and pesky double teamers, but beware of rest users and Immunity Snorlax2. I like to have a flying type/levitator and a Rock Slide resist just to PP stall Snorlax2. Don't even consider Salamence2. It's a trap.)

Round 6: Noland round where Smart AI makes use of the OHKO moves (so sturdy pokemon get an extra nod here). Ideally you'll have the defensive synergy necessary to dance around the coverage moves. This can sometimes can be a tough round. Starmie3 is lurking.

(Tips: Steelix2 and Forretress2 are powerful Explosion users and act as a get-out-of-jail-free card against anything you don't like. If either have sturdy you can obliterate Dewgong3. Ludicolo3 is a streak ender, so familiarize yourself with the set. Gengar3 is now good. Be on the lookout for specially defensive or thick fat pokemon to make good pivots against the plethora of elemental punches flying around this round.)

Round 7: Smart AI has all the coverage and set up it needs. Build a solid team quickly. Expect the most traditional coverage moves, for example, earthquake/rock slide and surf/ice beam. This is a tough round.

(Tips: All of Round 6's tips still apply. Traditionally "good" pokemon are particularly good this round, and most "bad" pokemon actually have redeeming movesets. It goes without saying that Blissey is good every round, but Blissey4 sets up so easily (if there are no thick fat water/ice types) that it should be a crime.)

Round 8+: Smart AI piloting unknown sets means this is as hard as the Factory gets. Scouting finally becomes a factor.

(Tips: Protect, Endure Substitute users gain viability (other than beating Slaking) in the form of scouting. Most electric types can handle the birds, whereas most fighting types can handle the Regi's. Lati@s walls/counters are worth holding onto; ideally you have the Lati wall/counter in the back, so you can swap for Lati@s in the lead.)

And then Open Level rounds look something like this:

Round 1: Dumb AI combined with many pointless moves for opponents to select makes this round easy.

Round 2: Dumb AI won't properly use set up moves and plenty of bad sets exist. I consider this round easy-ish (Side note: I'd recommend attacking chesto resto users immediately, even when in a bad matchup, so hopefully they rest, eat their chesto, then put themselves to sleep again).

Round 3: Moderate AI has many coverage moves at its disposal but is still randomly firing them off. This round could be difficult if you can't get a good team before Noland or if you get exploded upon/OHKO'd in the process.

Round 4: Moderate AI is still in effect and only smart enough not to status twice; it still chooses damaging moves randomly. These are typically the best sets, so hopefully you'll patch together offensive synergy (defensive synergy is less important because that implies switching, which is still a bad idea). The 12IV is starting to cause a nice IV gap between the opponent and as such this round is traditionally easier than Round 3.

Round 5 and beyond: Smart AI has unknown sets and won't waste time setting up. In Round 6 you have to challenge Noland's 31IV team with your 18IV and 3IV pokemon. This is as hard as it gets.

If your goal is just to get the gold symbol, you'll find that Open Level goes way faster than Level 50. You can probably challenge Noland2 twice by the time it takes to get to Noland2 once in Level 50. I attribute this solely to how slow-paced Level 50 Rounds 1-3 are. But if your goal is to get an even higher streak, you are studying the spreadsheet of pokemon and you have the patience to spend many, many hours playing, I'd recommend Level 50, because having a smaller pool of pokemon to sift through will help you make more informed decisions, at least up until Round 8.
Team-building Archetypes

A) Concrete team-building notes:

-PP stalling is wonderful and opens doors for other teammates (Blissey, recover Milotic and Starmies 7 and 8), but know your crit range and avoid it!

-Enfeeble then set up (think Latias4). Know your set up pokemon's limitations because there is no truly safe battle (e.g. +6 calm mind Lati@s can fall to quick claw Scizor, +6 Gyarados can't beat Skarmory, etc.).

-Explosion users are very useful and sometimes having explosion even allows them to transcend their weaknesses. Who cares if water is super effective on rock if Regirock tanks any Suicune's surf and revenge kills it with explosion? They usually end up being my answer to double teamers. Just avoid leaving them as your last pokemon, because your last pokemon fainting on the same turn your opponent's last pokemon will result in a loss no matter what.

-Destiny Bond is similar to explosion in that it is a tool to get you out of problematic pokemon for your team. Sometimes you'll want to awkwardly hard switch your ghost in during a 2v2 to avoid keeping them last, for the same reason as above.

-A Lati@s lead followed by a special wall counter (oftentimes an explosion user, e.g. Steelix) is an effective strategy, though you should fear Umbreon.

-A Raikou lead partnered with a ground type switch in/counter is another effective strategy. Beware of Marowak with swords dance and max speed investment, however. Raikou6 doesn't even need two other teammates, as he literally sets up on any pokemon without earthquake, while also PP stalling, with little effort.

-Suicune is good, but it's only B-tier in my opinion. Only two sets have rest: Suicune1 is walled by water absorb, and Suicune6 is only in Open Level. Scope lens leaf blade Sceptile is always scary to Suicune.

-Protect or fast substitute can be used as a method of scouting the enemy pokemon's set, as long as you remember that they could set up during that passive turn.

-Toxic/will-o-wisp are nice back-up strategies for troublesome pokemon. If your team has good synergy/immunities, you can switch around while their HP is whittled away.

B) Abstract team-building: First and foremost, you want a solid team that minimizes the number of threatening pokemon or pokemon that will end your streak. Refer to every pokemon by its set name (Tauros1, Starmie8, etc). There is no need for ambiguity when you can type in any/every match up in the damage calculator. Say you have 18 swaps and have a "powerful" (31 IV) pokemon in your draft, a Houndoom. Now, a 31 IV Houndoom4 (95 base speed with 252 EVs) is pretty amazing because it outspeeds every Lati@s and is guaranteed to tank one of their earthquakes, whereas 3IV Houndoom4 gets outsped and OHKO'd in the same scenario. I think deeply about what 3IV enemy pokemon I'll encounter and how my team will fare. Think in the "right now" and reassess your situation every single turn. Never lose sight of your path to victory. Give preference to water types, because waters check a lot of types (primarily steel, fire, ground types) that they then partner with well. Give preference to fast pokemon, which I consider 161+ speed in Open Level and 83+ speed in Level 50. This is to outspeed Marowak2-4, who can take a hit setting up, and then demolish slower teams. If you can outspeed the opponent, that's one less turn to get attracted, confused, or exploded upon. One of my best streaks ended because my three pokemon couldn't stop a Crobat4 using confuse ray. Fast pokemon are prone to missing moves (as they often have powerful but low-accuracy moves) and are vulnerable to quick claw, while slow pokemon are prone to getting statused and critically hit.

C) Speed creeps for starting draft pokemon. These are IV boosts that cause the pokemon to outspeed when they would otherwise get outsped

Round 4 - 12IV Blastoise1 ouspeeds 3IV Glalie1, Meganium1, base 80s with 0 speed EV's.
12 IV Magmar1 ouspeeds 3IV Nidoking1, Heracross1, Kingdra1.
12IV Feraligatr1 outspeeds 3IV Venusaur1.

Round 5 - 15IV Houndoom2, 15IV Heracross2 and 15IV Arcanine2 ouspeed base speed 100's with 252 speed EV's.
15IV Marowak2 outspeeds Milotic2.

Round 6 - 21IV base speed 100's (Ninetails3, Raichu3, Typhlosion3, Fearow3 etc.) outspeed the 3IV base speed 105's (Manectric3 and Electabuzz3).
21IV Rapidash3 outspeeds 3IV base 100's with 252 speed EV's.
21IV Starmie3/4 outspeeds Dugtrio1/2/3/4.

Round 7 - 31IV base speed 100's with 252 speed EV's (Charizard4, Dodrio4, Zapdos1 etc.) outspeed base speed base 110's with 252 speed EV's (Gengar4 and Espeon4).
31IV 252 speed EV Manectric outspeeds 3IV Starmie3/4 with 252 speed EV's.
All 31IV Alakazam sets outspeed 3IV Base100's with 252 speed EV's.
31IV 0 speed EV Armaldo4 outspeeds all 3IV 0 speed Blisseys and Exeggutors.
One day I noticed green_typhlosion's Level 50 Factory streak, where he swapped nearly every battle and won 64 battles (this record was actually never added to the leaderboard for some reason). Up until then, I'd always swapped conservatively and tried to build a balanced team for any opponent. But when the assistant tells you what type of team you'll be facing, a lot of times it pays off to build a lopsided team to counter the next opponent (especially in the lower rounds). It would only be scary if the opponent could exploit a weakness of your team, but this can be patched by poor AI move choice and mitigating risk. So my new playstyle involves, for the first 4 rounds, swapping as constantly as possible, learning as much as possible about all pokemon sets by trying them out, and boosting my elevations. The exception to this is in Round 3, when you have a Noland battle at the end of the round. I will stop swapping early in the round if I'm confident in the team. The elevations are quite nifty down the line, as they will potentially let you speed creep with offensive pokemon and/or stall more effectively with defensive pokemon. Not to mention you're almost always battling any Noland at a huge IV disadvantage, so I do everything possible to even the playing field. These elevations are actually different based on what round you're on, and are the same for Level 50 and Open Level.

Round 3 Elevations are 15IVs.

Round 4 Elevations are 18IVs.

Round 5 Elevations are 21IVs.

Round 6+ Elevations are 31IVs.
When I say identify your current team threats, I don't mean saying "I have one water weakness and no resists", I mean what tangible set will you run into that will wreck you? Think about dual typings and common coverage moves, as these will sneak up on you. Load the most threatening possible enemy pokemon of that typing into the damage calc and see how your team would fare. Swap for pokemon that can patch that hole. If your team has three primarily physical attackers, imagine how they would fare against Regirocks, Forretresses or Skarmorys. If yout team has three primarily special attackers, imagine how they would fare against Blisseys, Regices, and Umbreons. If absolutely no synergy is possible, pick fast pokemon. Fast pokemon get flinched less often, can usually guarantee chip damage, and can avoid one turn of infatuation, status, or confusion if they outspeed.

If you encounter one of your threats, play carefully. You might have to sack whatever pokemon is currently in just to get a pokemon to 2HKO in return. It may seem dumb to sack your Snorlax2, but say you lead Fearow2 against Nidoking2 and you know the opponent uses fighting types. It wouldn't pan out to try to save Snorlax for late battle where he'd likely get cross chopped. Try to plan out each pokemon's role in every battle. Don't lose sight of the path to victory.
Scout as much as possible. In battle, scout the enemy team by paying attention to the next trainer's favored type and playstyle, what pokemon is sent out first and second, and what items are in play. With this information, map out each individual battle accordingly. I hope to prove that sorting through all of the data is worthwhile. You'll always have a definitive list of opposing pokemon to possibly encounter, and you can then create situations where you're least likely to lose.

Now, in two regards the Factory is easier than any other facility. What? How?! Firstly, the IV's are going to be in your favor before you swap (barring Noland battles). In what other facility would you stay in with a Moltres on a max special attack Starmie? The IV boost can also be a godsend if one of your starting draft is at a crowded speed tier. A slower example is 31IV Marowak2-4 getting the jump on base 50's up through uninvested base 85's. Faster examples are 31IV max speed base 90s (Kangaskhan and Moltres), base 95's, and 100 base speed outspeeding the 252 speed base 100 crowd. Furthermore, the Factory assistant tells you if there is a "clear favored typing" from the next opponent. The list of pokemon you can encounter is finite. The opponent can't have your own pokemon (barring Noland, who has had my own pokemon from my current team on two occasions). If you are outrageously thorough and patient, you can map out your battles by plugging in every possibility. It's more useful when there is a favored typing, because the list of possible opponents becomes much more manageable. The player can break item clause but the AI (and starting draft) does not. So you will only face a maximum of leftovers and one chesto berry per battle.

Let's say your oppenent favors water types and they send out a Gyarados first. This tells you that they don't have any other flying types, or else the assistant would have reported that the opponent has no clear favorites. Upon every new turn, you should absolutely reassess what the AI will do, and confirm that your current path to victory is the best possible one. Even if there is no clear favored typing, remember that mid-battle when you see the second pokemon. Unless it's the unlikely instance of 2 double types, like Swampert/Whiscash or two grass/poison types, in which case you already beat the first of said dual type pokemon, so chances are you can beat the second as well. Most of the time when I hear no clear typing and I'm mid-battle, I think "there probably isn't another one of these types in the back."

As an example, say you've heard "no clear favored typing" and your lead Raikou1 (whose only attacking move is thunderbolt) takes out the enemy's lead Weezing. Next your opponent throws out Flygon. Now is big brain time, because you can't be 100% certain that the opponent won't have a Nidoking in the back. So you'll have to play with at least one answer to the Nidos. Using the same example with no clear favored typing, let's say after your lead Raikou kills their lead Weezing, they instead send out Nidoking. Now you know something's up, or else the assistant would have told you the opponent prefers poison types. You must now prepare yourself to encounter another ground type pokemon in the last slot.

As soon as you hear a clear favored type, you should imagine how your team would fare against a legendary of that type. Definitely swap if that hypothetical legendary pokemon of that type would beat your team in practice.

The only time I can find usefulness in the battle style of the trainer is when it's "depends on the battles flow", which refers to there being two or more weather inducing moves in the next battle; And since there's no pokemon with two weather moves on one moveset, you can start only looking at pokemon that round that have weather movesets and seeing what makes sense combined with any type favoring or items already presented.
Now put it all together! I always have the damage calculator with have both pokemon loaded up in one hand and spreadsheet pulled up in Excel in the other.

To start open the attached spreadsheet below in Microsoft Excel and you will see all of the pokemon sets with my addition of the Type column. Firstly, you should click on "Set #" and filter by using "number filter" and "equals" and typing the correct set number. This will make sure you are only looking at the potential pokemon this round.

Second, if the Factory assistant gives you a favored type, you'll want to refresh yourself on what possible pokemon of that type you may face. Filter by clicking "Type" and filtering using "text filters" then "contains..." and type the favored type.

Next, after seeing the opponent's first pokemon, and possibly after scouting, you'll know what the first pokemon's item is. So before attacking any further you'll want to click the "Item" cell and then "text filter" but now hit "does not contain..." whatever item the first enemy pokemon was holding. You can only do this filter once, even though you will find out the second item eventually.

That is the best I can narrow down the potential sets so far. My next project was to add another column entitled "Battle Style" and for every pokemon's every move type out the corresponding battle style category. This would look something like: 1,3,6,1. With that new column, you could run a separate filter for the corresponding battle style when the assistant tells you. It would also be nice to update the speed stats of all of the pokemon to reflect their speeds with 3IVs. This would also be nice to order the results in descending order and be hyper aware of your speed tiers that round. Alternatively I could rearrange the results in descending order of base stat total, if I was to go through and make a new column for base stat totals.
While playing the Battle Factory exclusively for over year, I lost a lot. But what I started doing early on was writing a short report on how I lost. I detailed my team, the team I lost to and the type of loss I had. There are four ways I lost:

Hax - happens when I choose to map my battle out a certain way and it was undone by an attack of mine missing or a hax item activating.

Counter-team - happens when the opponent's team has great answers to my pokemon and there was nothing I could do.

Bad Draft - happens when you lose the first battle of a round due to geting a draft of bad pokemon/bad sets. This could also be a lack of synergy. I once had 6 psychic types to choose from. You can also get bad duplicates due to an elevation (Wailord1 and Wailord2).

Misplays - happens when I lost when there was something I could have done to win.

The purpose of writing reports was to minimize misplays and to grow from every loss. You will certainly lose in unfair ways and when you lose to hax, counter-teams, and bad drafts you cannot beat yourself up. You can only move on and start again.
There are still mysteries within the Factory that I can't figure out. Either Bulbapedia or my Emerald cartridge is wrong. I'm namely referring to the starting draft pokemon IV values. Bulbapedia would have you believe:

Round 1 = 3IV starting picks
Round 2 = 6IV starting picks
Round 3 = 9IV starting picks
Round 4 = 12IV starting picks
Round 5 = 15IV starting picks
Round 6 = 21IV starting picks
Round 7+ = 31IV starting picks
However, I have hit Round 9 multiple times and, about half of the time, I encounter utterly random IVs for every single stat on every single starting draft pokemon (with the exception of one or two 31IV pokemon, depending on my swaps). It's mostly a gripe because of how much more time it takes to load up every random IV for a pokemon multiple times in a battle. But this is not absolute, because sometimes Round 9 will give me random IV draft picks and then by Round 10 the draft goes back to all being 31IV.

I've also been in a situation where I swapped 38 times and expected 4 elevation pokemon, yet only the last 2 pokemon of the starting draft had 31IV's. After winning the first battle, I swapped for the enemy's pokemon in the third party slot. I wish I would have instead swapped for the first or second pokemon slot, because I suspect these pokemon may have received the 31IVs instead of me! I base this off of a breakdown of the Emerald Factory coding that juxtaposes the starting draft and the first enemy trainer.

***Update: All starting drafts in Round 9 will have random IV's instead of the normal 31 IV's, and since Round 8's elevations are pulled from Round 9's starting draft, Round 8's elevations are random IV's instead of 31 IV's. So this clears up my confusion and becomes yet another quirk of the Factory.

Original Spreadsheet:
https://docs.google.com/spreadsheet...ouid=101302679580129866296&rtpof=true&sd=true


Optimized Spreadsheet:
https://1drv.ms/x/s!ApUbJTvfV1pWgxTGms737BxoljQ4

Final Spreadsheet (with Battle Style Column): https://1drv.ms/x/s!ApUbJTvfV1pWgxhk92MOXhXEWgry
 
Last edited:

QuentinQuonce

formerly green_typhlosion
Joe's Declassified Factory Survival Guide


Become a knowledgeable gen 3 battler before entering. I recommend doing this facility last. Are you familiar with the gen 3 battle mechanics and the particularly dangerous pokemon sets? You should know that thunder always lands in the rain, even if the enemy is at +6 evasion. You should know that you can revenge kill pokemon on a sliver of health just by sending out Tyranitar to summon sand before the turn ends. You should know electric types can be paralyzed and that paralysis lowers speed by 75% in this gen. If you don't fully recognize all the tools at your disposal, you'll never make it far in this facility. You are going to lose battles that you seemed guaranteed to win, and win battles you were sure you'd lose (so never throw in the towel until the battle is completely over!). Familiarize yourself with the resources that remove guesswork: spreadsheet of pokemon sets, gen 3 damage calc, and rules of the Factory.
Learn as much as you can about how the AI will act in any given situation. This will cause you to alter how to make decisions for each bracket of AI.

Rounds 1 and 2: Dumb AI: opponent randomly chooses a 1 of their 4 moves. The answer? Stay in majority of the time basing your decisions on how likely they'll randomly select moves. Switch based on how scared you are of opponent's moves. They might have that super effective move that can OHKO, but there's only a 1 in 4 chance they'll select it. Switch if you have a strong counter, but remember that even counters can get statused. However, you should never risk your wincon (unless playing passively lets them set up and beat your wincon).

Rounds 3 and 4: Moderate AI: opponent won't status twice. And the AI usually doesn't fall for type immunities (e.g. an opponent won't try to thunderbolt a ground type). So if you're already confused, you can switch in another pokemon knowing that they won't try use confuse ray again. Damaging moves are still randomly selected. The answer? Play similarly to the previous rounds, bearing in mind the AI's new aversion to certain moves. For example, if you have a Fearow out against a Golem, you could reasonably safely switch in your Lanturn, as the opponent knows not to go for earthquake.

Rounds 5+: Smart AI: the opponent will prioritize going for the KO if they know a move that guarantees the KO. They always choose the smartest damaging move, but still may go for status, weather, or (emphasis on) stat boosts. If the opponent starts setting up offensive boosts, it will *almost certainly* continue setting up until it can knock out your pokemon in front of it in one hit. I've seen a "Smart AI" Starmie not go for a surf on my Moltres because it would only do 99% (56.3% chance to OHKO). This AI seemingly favors speed control over the guaranteed KO; for example, an enemy Marowak used icy wind six times as my Tyranitar dragon danced six times. The AI also misjudges with curse. Twice now, I've been able to 3HKO a pokemon with curse AND rest because they curse twice to be able to OHKO, not considering that they will be knocked out in the process of setting up. The answer? Battle your heart out. Use every single move, ability and item available if it can help you win. Predict the AI's moves by comparing damage outputs on the damage calculator. Go slowly and think about every inkling of information on the enemy that you've received and map out the battle. Plug in every possibility, including hax. Pro Tip: play with animations to slow the game down! You do not want to miss seeing if that Arcanine had intimidate or if a water pulse confused or not, and this is a potential consequence of playing with animations off.

Noland AI: acts like Round 5+ AI, except he more heavily favors OHKO moves and has an aversion to using, or else just plain misuses, his own endure/flail strategies. Noland values set up moves differently and does not necessarily set up until he can OHKO. For example, his Dragonite1 used Brick Break to 2HKO my Tyranitar despite having the option to dragon dance and make it an OHKO. He does not always go for the highest damaging move and I suspect it's because he favors moves that have added effects. For example, turn one his Sceptile once used crush claw, instead of the stronger leaf blade, on my Zapdos (an instance where neither move would have killed me).

Every Level of Opponent AI: regardless of the round, the opponent will favor sending out a pokemon with a super effective move. Now, dumb AI might never even click it, but you can make detailed inferences if you're paying enough attention. For example, say I lead with my Moltres and kill their steel type, and they then send out Ludicolo4 (toxic, leech seed, double team and rain dance). Lets say I beat it down with aerial ace and then the opponent sends out Tyranitar third. The fact that the opponent didn't send out Tyranitar sooner tells me that it isn't running rock slide, thus allowing us to disregard 7 of the possible 10 sets it could be running. But don't get ahead of yourself; if the Ludicolo is running surf, you cannot draw any subsequent conclusions about Tyranitar. In other words, if the second enemy pokemon comes out and does not have super effective coverage to hit your pokemon, neither will the third pokemon. Always remind yourself that they chose to send out this second pokemon. The enemy AI guesses your ability if there are multiple possibilities, but it will remember your ability once it is revealed. For example, they won't thunderbolt your low health Lanturn with volt absorb if they've already seen it activate.
You must decide between Level 50 and Open Level. I think Level 50 is easier because of the static sets--who doesn't have an aversion to scouting unknown sets? Furthermore, in theory the Level 50 Round 8+ metagame is safer than Open Level without Tyranitar and Dragonite, who often sport hax items. Level 50 also excludes sets 5 and 6 of the bird and dog trio pokemon which, considering Raikou6 is the god of the gen 3 factory, is yet another factor that makes Level 50 easier. A good counterargument to this "level 50 is easier" argument lies in Level 50 Round 5, when battling with the brutal meta of the set 2 pokemon. I've somewhat generalized the Open level sets as such:

Set 1: One STAB move, status move, stat lowering move, counter users.

Set 2: Double team and other stat boosting move users, many running resto-chesto. Suboptimal sets.

Set 3: Diverse coverage moves. Ie lots of elemental punch users (although many of which are physical attackers), Earthquake Rock Slide combo, and Surf Ice Beam combo. Some explosion and OHKO users.

Set 4: "Ideal" movesets with coverage. Set up moves, OHKO moves, and only special attackers run elemental punches. Reminiscent of their ADV OU sets.

Factory Ubers Sets: Some limited to STAB moves, some run "ideal" sets, and plenty of hax/gimmick sets.

Combining the AI's intelligence with the following generalizations of every set's meta will help show the general difficulty for each round. Ignoring the all important factor of luck, the Level 50 rounds probably look something like this:

Round 1: Dumb AI is piloting the "factory little cup"'s putrid movesets with many bad moves to choose. This is an easy round.

(Tips: Charmeleon1 can mow down its counters thanks to bad AI; this also makes for a face paced round. Grass types packing toxic are pretty strong. Seek out pokemon with (4) usuable moves and items; for example, Pidgeotto1 and Masquerain1. For a sense of power creep, remember that Plusle1 is pretty good this round)

Round 2: Dumb AI has the exact same bad movesets with a few new ones thrown in. "Middle cup" should be an easy round too.

(Tips: The Charmeleon1 lead is still good. But by now I consider Plusle1 a pretty bad pokemon, considering the other evolved pokemon's bulk. So look for evolved pokemon of your own. Psychic types are great this round as long as you are conscious of the dark types (Sableye1, Cacturne1, Sharpedo1). Remember that Future Sight affects dark types, which is very relevant this round. Fighting type leads are strong and make for quick paced battles. Normal types are nice when paired with a rock and steel answer.)

Round 3: It feels like the Moderate AI has movesets tailor made for it, with many status options and solid coverage. A tricky metagame and a Noland fight? I consider this a potentially challenging round.

(Tips: Rain teams can be strong (Omastar2, Gorebyss2, Magneton2). Jumpluff2 is a run ender so familiarize yourself with it. The fast fighting lead is still strong. Chansey2 is not an auto win, but close. Grumpig2, Scyther2, and Gligar2 are some of my favorites.)

Round 4: Moderate AI and limited damaging moves means powerful STAB will be selected often, and with status options plaguing the round, it's moderately difficult. The starting draft having 12 IVs could now be enough for them to speed creep, i.e. outspeed opponents with a higher base speed stat than they have.

(Tips: Maintaining synergy is always important, but you'll likely be switching into status. Coincidentally, a lot of viable pokemon run Lum this round (Gyrados1, Espeon1, Milotic1))

Round 5: In this round, smart AI is pulling the strings of the most hated sets in the Frontier. Set 2 is home to some of the most dangerous sets (e.g. Gardevoir2 and Snorlax2) and yet the worst sets too (see Salamence2, Crobat2, Flygon2 and Porygon2-2). Consequently, it is a real race to see who can get the better set up sets.

(Tips: Toxic users can shut down a lot of pokemon and pesky double teamers, but beware of rest users and Immunity Snorlax2. I like to have a flying type/levitator and a Rock Slide resist just to PP stall Snorlax2. Don't even consider Salamence2. It's a trap.)

Round 6: Noland round where Smart AI makes use of the OHKO moves (so sturdy pokemon get an extra nod here). Ideally you'll have the defensive synergy necessary to dance around the coverage moves. This can sometimes can be a tough round. Starmie3 is lurking.

(Tips: Steelix2 and Forretress2 are powerful Explosion users and act as a get-out-of-jail-free card against anything you don't like. If either have sturdy you can obliterate Dewgong3. Ludicolo3 is a streak ender, so familiarize yourself with the set. Gengar3 is now good. Be on the lookout for specially defensive or thick fat pokemon to make good pivots against the plethora of elemental punches flying around this round.)

Round 7: Smart AI has all the coverage and set up it needs. Build a solid team quickly. Expect the most traditional coverage moves, for example, earthquake/rock slide and surf/ice beam. This is a tough round.

(Tips: All of Round 6's tips still apply. Traditionally "good" pokemon are particularly good this round, and most "bad" pokemon actually have redeeming movesets. It goes without saying that Blissey is good every round, but Blissey4 sets up so easily (if there are no thick fat water/ice types) that it should be a crime.)

Round 8+: Smart AI piloting unknown sets means this is as hard as the Factory gets. Scouting finally becomes a factor.

(Tips: Protect, Endure Substitute users gain viability (other than beating Slaking) in the form of scouting. Most electric types can handle the birds, whereas most fighting types can handle the Regi's. Lati@s walls/counters are worth holding onto; ideally you have the Lati wall/counter in the back, so you can swap for Lati@s in the lead.)

And then Open Level rounds look something like this:

Round 1: Dumb AI combined with many pointless moves for opponents to select makes this round easy.

Round 2: Dumb AI won't properly use set up moves and plenty of bad sets exist. I consider this round easy-ish (Side note: I'd recommend attacking chesto resto users immediately, even when in a bad matchup, so hopefully they rest, eat their chesto, then put themselves to sleep again).

Round 3: Moderate AI has many coverage moves at its disposal but is still randomly firing them off. This round could be difficult if you can't get a good team before Noland or if you get exploded upon/OHKO'd in the process.

Round 4: Moderate AI is still in effect and only smart enough not to status twice; it still chooses damaging moves randomly. These are typically the best sets, so hopefully you'll patch together offensive synergy (defensive synergy is less important because that implies switching, which is still a bad idea). The 12IV is starting to cause a nice IV gap between the opponent and as such this round is traditionally easier than Round 3.

Round 5 and beyond: Smart AI has unknown sets and won't waste time setting up. In Round 6 you have to challenge Noland's 31IV team with your 18IV and 3IV pokemon. This is as hard as it gets.

If your goal is just to get the gold symbol, you'll find that Open Level goes way faster than Level 50. You can probably challenge Noland2 twice by the time it takes to get to Noland2 once in Level 50. I attribute this solely to how slow-paced Level 50 Rounds 1-3 are. But if your goal is to get an even higher streak, you are studying the spreadsheet of pokemon and you have the patience to spend many, many hours playing, I'd recommend Level 50, because having a smaller pool of pokemon to sift through will help you make more informed decisions, at least up until Round 8.
Team-building Archetypes

A) Concrete team-building notes:

-PP stalling is wonderful and opens doors for other teammates (Blissey, recover Milotic and Starmies 7 and 8), but know your crit range and avoid it!

-Enfeeble then set up (think Latias4). Know your set up pokemon's limitations because there is no truly safe battle (e.g. +6 calm mind Lati@s can fall to quick claw Scizor, +6 Gyarados can't beat Skarmory, etc.).

-Explosion users are very useful and sometimes having explosion even allows them to transcend their weaknesses. Who cares if water is super effective on rock if Regirock tanks any Suicune's surf and revenge kills it with explosion? They usually end up being my answer to double teamers. Just avoid leaving them as your last pokemon, because your last pokemon fainting on the same turn your opponent's last pokemon will result in a loss no matter what.

-Destiny Bond is similar to explosion in that it is a tool to get you out of problematic pokemon for your team. Sometimes you'll want to awkwardly hard switch your ghost in during a 2v2 to avoid keeping them last, for the same reason as above.

-A Lati@s lead followed by a special wall counter (oftentimes an explosion user, e.g. Steelix) is an effective strategy, though you should fear Umbreon.

-A Raikou lead partnered with a ground type switch in/counter is another effective strategy. Beware of Marowak with swords dance and max speed investment, however. Raikou6 doesn't even need two other teammates, as he literally sets up on any pokemon without earthquake, while also PP stalling, with little effort.

-Suicune is good, but it's only B-tier in my opinion. Only two sets have rest: Suicune1 is walled by water absorb, and Suicune6 is only in Open Level. Scope lens leaf blade Sceptile is always scary to Suicune.

-Protect or fast substitute can be used as a method of scouting the enemy pokemon's set, as long as you remember that they could set up during that passive turn.

-Toxic/will-o-wisp are nice back-up strategies for troublesome pokemon. If your team has good synergy/immunities, you can switch around while their HP is whittled away.

B) Abstract team-building: First and foremost, you want a solid team that minimizes the number of threatening pokemon or pokemon that will end your streak. Refer to every pokemon by its set name (Tauros1, Starmie8, etc). There is no need for ambiguity when you can type in any/every match up in the damage calculator. Say you have 18 swaps and have a "powerful" (31 IV) pokemon in your draft, a Houndoom. Now, a 31 IV Houndoom4 (95 base speed with 252 EVs) is pretty amazing because it outspeeds every Lati@s and is guaranteed to tank one of their earthquakes, whereas 3IV Houndoom4 gets outsped and OHKO'd in the same scenario. I think deeply about what 3IV enemy pokemon I'll encounter and how my team will fare. Think in the "right now" and reassess your situation every single turn. Never lose sight of your path to victory. Give preference to water types, because waters check a lot of types (primarily steel, fire, ground types) that they then partner with well. Give preference to fast pokemon, which I consider 161+ speed in Open Level and 83+ speed in Level 50. This is to outspeed Marowak2-4, who can take a hit setting up, and then demolish slower teams. If you can outspeed the opponent, that's one less turn to get attracted, confused, or exploded upon. One of my best streaks ended because my three pokemon couldn't stop a Crobat4 using confuse ray. Fast pokemon are prone to missing moves (as they often have powerful but low-accuracy moves) and are vulnerable to quick claw, while slow pokemon are prone to getting statused and critically hit.

C) Speed creeps for starting draft pokemon. These are IV boosts that cause the pokemon to outspeed when they would otherwise get outsped

Round 4 - 12IV Blastoise1 ouspeeds 3IV Glalie1, Meganium1, base 80s with 0 speed EV's.
12 IV Magmar1 ouspeeds 3IV Nidoking1, Heracross1, Kingdra1.
12IV Feraligatr1 outspeeds 3IV Venusaur1.

Round 5 - 15IV Houndoom2, 15IV Heracross2 and 15IV Arcanine2 ouspeed base speed 100's with 252 speed EV's.

Round 6 - 21IV base speed 100's (Ninetails3, Raichu3, Typhlosion3, Fearow3 etc.) outspeed the 3IV base speed 105's (Manectric3 and Electabuzz3).
21IV Rapidash3 outspeeds 3IV base 100's with 252 speed EV's.

Round 7 - 31IV base speed 100's with 252 speed EV's (Charizard4, Dodrio4, Zapdos1 etc.) outspeed base speed base 110's with 252 speed EV's (Gengar4 and Espeon4).
One day I noticed green_typhlosion's Level 50 Factory streak, where he swapped nearly every battle and won 64 battles (this record was actually never added to the leaderboard for some reason). Up until then, I'd always swapped conservatively and tried to build a balanced team for any opponent. But when the assistant tells you what type of team you'll be facing, a lot of times it pays off to build a lopsided team to counter the next opponent (especially in the lower rounds). It would only be scary if the opponent could exploit a weakness of your team, but this can be patched by poor AI move choice and mitigating risk. So my new playstyle involves, for the first 4 rounds, swapping as constantly as possible, learning as much as possible about all pokemon sets by trying them out, and boosting my elevations. The exception to this is in Round 3, when you have a Noland battle at the end of the round. I will stop swapping early in the round if I'm confident in the team. The elevations are quite nifty down the line, as they will potentially let you speed creep with offensive pokemon and/or stall more effectively with defensive pokemon. Not to mention you're almost always battling any Noland at a huge IV disadvantage, so I do everything possible to even the playing field. These elevations are actually different based on what round you're on, and are the same for Level 50 and Open Level.

Round 3 Elevations are 15IVs.

Round 4 Elevations are 18IVs.

Round 5 Elevations are 21IVs.

Round 6+ Elevations are 31IVs.
When I say identify your current team threats, I don't mean saying "I have one water weakness and no resists", I mean what tangible set will you run into that will wreck you? Think about dual typings and common coverage moves, as these will sneak up on you. Load the most threatening possible enemy pokemon of that typing into the damage calc and see how your team would fare. Swap for pokemon that can patch that hole. If your team has three primarily physical attackers, imagine how they would fare against Regirocks, Forretresses or Skarmorys. If yout team has three primarily special attackers, imagine how they would fare against Blisseys, Regices, and Umbreons. If absolutely no synergy is possible, pick fast pokemon. Fast pokemon get flinched less often, can usually guarantee chip damage, and can avoid one turn of infatuation, status, or confusion if they outspeed.

If you encounter one of your threats, play carefully. You might have to sack whatever pokemon is currently in just to get a pokemon to 2HKO in return. It may seem dumb to sack your Snorlax2, but say you lead Fearow2 against Nidoking2 and you know the opponent uses fighting types. It wouldn't pan out to try to save Snorlax for late battle where he'd likely get cross chopped. Try to plan out each pokemon's role in every battle. Don't lose sight of the path to victory.
Scout as much as possible. In battle, scout the enemy team by paying attention to the next trainer's favored type and playstyle, what pokemon is sent out first and second, and what items are in play. With this information, map out each individual battle accordingly. I hope to prove that sorting through all of the data is worthwhile. You'll always have a definitive list of opposing pokemon to possibly encounter, and you can then create situations where you're least likely to lose.

Now, in two regards the Factory is easier than any other facility. What? How?! Firstly, the IV's are going to be in your favor before you swap (barring Noland battles). In what other facility would you stay in with a Moltres on a max special attack Starmie? The IV boost can also be a godsend if one of your starting draft is at a crowded speed tier. A slower example is 31IV Marowak2-4 getting the jump on base 50's up through uninvested base 85's. Faster examples are 31IV max speed base 90s (Kangaskhan and Moltres), base 95's, and 100 base speed outspeeding the 252 speed base 100 crowd. Furthermore, the Factory assistant tells you if there is a "clear favored typing" from the next opponent. The list of pokemon you can encounter is finite. The opponent can't have your own pokemon (barring Noland, who has had my own pokemon from my current team on two occasions). If you are outrageously thorough and patient, you can map out your battles by plugging in every possibility. It's more useful when there is a favored typing, because the list of possible opponents becomes much more manageable. The player can break item clause but the AI (and starting draft) does not. So you will only face a maximum of leftovers and one chesto berry per battle.

Let's say your oppenent favors water types and they send out a Gyarados first. This tells you that they don't have any other flying types, or else the assistant would have reported that the opponent has no clear favorites. Upon every new turn, you should absolutely reassess what the AI will do, and confirm that your current path to victory is the best possible one. Even if there is no clear favored typing, remember that mid-battle when you see the second pokemon. Unless it's the unlikely instance of 2 double types, like Swampert/Whiscash or two grass/poison types, in which case you already beat the first of said dual type pokemon, so chances are you can beat the second as well. Most of the time when I hear no clear typing and I'm mid-battle, I think "there probably isn't another one of these types in the back."

As an example, say you've heard "no clear favored typing" and your lead Raikou1 (whose only attacking move is thunderbolt) takes out the enemy's lead Weezing. Next your opponent throws out Flygon. Now is big brain time, because you can't be 100% certain that the opponent won't have a Nidoking in the back. So you'll have to play with at least one answer to the Nidos. Using the same example with no clear favored typing, let's say after your lead Raikou kills their lead Weezing, they instead send out Nidoking. Now you know something's up, or else the assistant would have told you the opponent prefers poison types. You must now prepare yourself to encounter another ground type pokemon in the last slot.

As soon as you hear a clear favored type, you should imagine how your team would fare against a legendary of that type. Definitely swap if that hypothetical legendary pokemon of that type would beat your team in practice.

The only time I can find usefulness in the battle style of the trainer is when it's "depends on the battles flow", which refers to there being two or more weather inducing moves in the next battle; And since there's no pokemon with two weather moves on one moveset, you can start only looking at pokemon that round that have weather movesets and seeing what makes sense combined with any type favoring or items already presented.
Now put it all together! I always have the damage calculator with have both pokemon loaded up in one hand and spreadsheet pulled up in Excel in the other.

To start open the attached spreadsheet below in Microsoft Excel and you will see all of the pokemon sets with my addition of the Type column. Firstly, you should click on "Set #" and filter by using "number filter" and "equals" and typing the correct set number. This will make sure you are only looking at the potential pokemon this round.

Second, if the Factory assistant gives you a favored type, you'll want to refresh yourself on what possible pokemon of that type you may face. Filter by clicking "Type" and filtering using "text filters" then "contains..." and type the favored type.

Next, after seeing the opponent's first pokemon, and possibly after scouting, you'll know what the first pokemon's item is. So before attacking any further you'll want to click the "Item" cell and then "text filter" but now hit "does not contain..." whatever item the first enemy pokemon was holding. You can only do this filter once, even though you will find out the second item eventually.

That is the best I can narrow down the potential sets so far. My next project was to add another column entitled "Battle Style" and for every pokemon's every move type out the corresponding battle style category. This would look something like: 1,3,6,1. With that new column, you could run a separate filter for the corresponding battle style when the assistant tells you. It would also be nice to update the speed stats of all of the pokemon to reflect their speeds with 3IVs. This would also be nice to order the results in descending order and be hyper aware of your speed tiers that round. Alternatively I could rearrange the results in descending order of base stat total, if I was to go through and make a new column for base stat totals.
While playing the Battle Factory exclusively for over year, I lost a lot. But what I started doing early on was writing a short report on how I lost. I detailed my team, the team I lost to and the type of loss I had. There are four ways I lost:

Hax - happens when I choose to map my battle out a certain way and it was undone by an attack of mine missing or a hax item activating.

Counter-team - happens when the opponent's team has great answers to my pokemon and there was nothing I could do.

Bad Draft - happens when you lose the first battle of a round due to geting a draft of bad pokemon/bad sets. This could also be a lack of synergy. I once had 6 psychic types to choose from. You can also get bad duplicates due to an elevation (Wailord1 and Wailord2).

Misplays - happens when I lost when there was something I could have done to win.

The purpose of writing reports was to minimize misplays and to grow from every loss. You will certainly lose in unfair ways and when you lose to hax, counter-teams, and bad drafts you cannot beat yourself up. You can only move on and start again.
There are still mysteries within the Factory that I can't figure out. Either Bulbapedia or my Emerald cartridge is wrong. I'm namely referring to the starting draft pokemon IV values. Bulbapedia would have you believe:

Round 1 = 3IV starting picks
Round 2 = 6IV starting picks
Round 3 = 9IV starting picks
Round 4 = 12IV starting picks
Round 5 = 15IV starting picks
Round 6 = 21IV starting picks
Round 7+ = 31IV starting picks
However, I have hit Round 9 multiple times and, about half of the time, I encounter utterly random IVs for every single stat on every single starting draft pokemon (with the exception of one or two 31IV pokemon, depending on my swaps). It's mostly a gripe because of how much more time it takes to load up every random IV for a pokemon multiple times in a battle. But this is not absolute, because sometimes Round 9 will give me random IV draft picks and then by Round 10 the draft goes back to all being 31IV.

I've also been in a situation where I swapped 38 times and expected 4 elevation pokemon, yet only the last 2 pokemon of the starting draft had 31IV's. After winning the first battle, I swapped for the enemy's pokemon in the third party slot. I wish I would have instead swapped for the first or second pokemon slot, because I suspect these pokemon may have received the 31IVs instead of me! I base this off of a breakdown of the Emerald Factory coding that juxtaposes the starting draft and the first enemy trainer.

https://docs.google.com/spreadsheet...ouid=101302679580129866296&rtpof=true&sd=true
This is an incredible post and needs to be pinned. Also thanks for the shoutout!

I've gotten the Gold symbol in the Factory several times now, but I wish I'd had this years ago. You hit on several things I'd never even really considered, like the point about the order in which the AI sends out their Pokemon.

I also agree, from long experience, that level 50 is the way to go. You give yourself so much more range to swap in the early rounds that you get an advantage by the time you get to round 4. It's painstaking but well worth it.

Really, really great guide and I hope this is read by lots of people.
 
Joe's Declassified Factory Survival Guide


Become a knowledgeable gen 3 battler before entering. I recommend doing this facility last. Are you familiar with the gen 3 battle mechanics and the particularly dangerous pokemon sets? You should know that thunder always lands in the rain, even if the enemy is at +6 evasion. You should know that you can revenge kill pokemon on a sliver of health just by sending out Tyranitar to summon sand before the turn ends. You should know electric types can be paralyzed and that paralysis lowers speed by 75% in this gen. If you don't fully recognize all the tools at your disposal, you'll never make it far in this facility. You are going to lose battles that you seemed guaranteed to win, and win battles you were sure you'd lose (so never throw in the towel until the battle is completely over!). Familiarize yourself with the resources that remove guesswork: spreadsheet of pokemon sets, gen 3 damage calc, and rules of the Factory.
Learn as much as you can about how the AI will act in any given situation. This will cause you to alter how to make decisions for each bracket of AI.

Rounds 1 and 2: Dumb AI: opponent randomly chooses a 1 of their 4 moves. The answer? Stay in majority of the time basing your decisions on how likely they'll randomly select moves. Switch based on how scared you are of opponent's moves. They might have that super effective move that can OHKO, but there's only a 1 in 4 chance they'll select it. Switch if you have a strong counter, but remember that even counters can get statused. However, you should never risk your wincon (unless playing passively lets them set up and beat your wincon).

Rounds 3 and 4: Moderate AI: opponent won't status twice. And the AI usually doesn't fall for type immunities (e.g. an opponent won't try to thunderbolt a ground type). So if you're already confused, you can switch in another pokemon knowing that they won't try use confuse ray again. Damaging moves are still randomly selected. The answer? Play similarly to the previous rounds, bearing in mind the AI's new aversion to certain moves. For example, if you have a Fearow out against a Golem, you could reasonably safely switch in your Lanturn, as the opponent knows not to go for earthquake.

Rounds 5+: Smart AI: the opponent will prioritize going for the KO if they know a move that guarantees the KO. They always choose the smartest damaging move, but still may go for status, weather, or (emphasis on) stat boosts. If the opponent starts setting up offensive boosts, it will *almost certainly* continue setting up until it can knock out your pokemon in front of it in one hit. I've seen a "Smart AI" Starmie not go for a surf on my Moltres because it would only do 99% (56.3% chance to OHKO). This AI seemingly favors speed control over the guaranteed KO; for example, an enemy Marowak used icy wind six times as my Tyranitar dragon danced six times. The AI also misjudges with curse. Twice now, I've been able to 3HKO a pokemon with curse AND rest because they curse twice to be able to OHKO, not considering that they will be knocked out in the process of setting up. The answer? Battle your heart out. Use every single move, ability and item available if it can help you win. Predict the AI's moves by comparing damage outputs on the damage calculator. Go slowly and think about every inkling of information on the enemy that you've received and map out the battle. Plug in every possibility, including hax. Pro Tip: play with animations to slow the game down! You do not want to miss seeing if that Arcanine had intimidate or if a water pulse confused or not, and this is a potential consequence of playing with animations off.

Noland AI: acts like Round 5+ AI, except he more heavily favors OHKO moves and has an aversion to using, or else just plain misuses, his own endure/flail strategies. Noland values set up moves differently and does not necessarily set up until he can OHKO. For example, his Dragonite1 used Brick Break to 2HKO my Tyranitar despite having the option to dragon dance and make it an OHKO. He does not always go for the highest damaging move and I suspect it's because he favors moves that have added effects. For example, turn one his Sceptile once used crush claw, instead of the stronger leaf blade, on my Zapdos (an instance where neither move would have killed me).

Every Level of Opponent AI: regardless of the round, the opponent will favor sending out a pokemon with a super effective move. Now, dumb AI might never even click it, but you can make detailed inferences if you're paying enough attention. For example, say I lead with my Moltres and kill their steel type, and they then send out Ludicolo4 (toxic, leech seed, double team and rain dance). Lets say I beat it down with aerial ace and then the opponent sends out Tyranitar third. The fact that the opponent didn't send out Tyranitar sooner tells me that it isn't running rock slide, thus allowing us to disregard 7 of the possible 10 sets it could be running. But don't get ahead of yourself; if the Ludicolo is running surf, you cannot draw any subsequent conclusions about Tyranitar. In other words, if the second enemy pokemon comes out and does not have super effective coverage to hit your pokemon, neither will the third pokemon. Always remind yourself that they chose to send out this second pokemon. The enemy AI guesses your ability if there are multiple possibilities, but it will remember your ability once it is revealed. For example, they won't thunderbolt your low health Lanturn with volt absorb if they've already seen it activate.
You must decide between Level 50 and Open Level. I think Level 50 is easier because of the static sets--who doesn't have an aversion to scouting unknown sets? Furthermore, in theory the Level 50 Round 8+ metagame is safer than Open Level without Tyranitar and Dragonite, who often sport hax items. Level 50 also excludes sets 5 and 6 of the bird and dog trio pokemon which, considering Raikou6 is the god of the gen 3 factory, is yet another factor that makes Level 50 easier. A good counterargument to this "level 50 is easier" argument lies in Level 50 Round 5, when battling with the brutal meta of the set 2 pokemon. I've somewhat generalized the Open level sets as such:

Set 1: One STAB move, status move, stat lowering move, counter users.

Set 2: Double team and other stat boosting move users, many running resto-chesto. Suboptimal sets.

Set 3: Diverse coverage moves. Ie lots of elemental punch users (although many of which are physical attackers), Earthquake Rock Slide combo, and Surf Ice Beam combo. Some explosion and OHKO users.

Set 4: "Ideal" movesets with coverage. Set up moves, OHKO moves, and only special attackers run elemental punches. Reminiscent of their ADV OU sets.

Factory Ubers Sets: Some limited to STAB moves, some run "ideal" sets, and plenty of hax/gimmick sets.

Combining the AI's intelligence with the following generalizations of every set's meta will help show the general difficulty for each round. Ignoring the all important factor of luck, the Level 50 rounds probably look something like this:

Round 1: Dumb AI is piloting the "factory little cup"'s putrid movesets with many bad moves to choose. This is an easy round.

(Tips: Charmeleon1 can mow down its counters thanks to bad AI; this also makes for a face paced round. Grass types packing toxic are pretty strong. Seek out pokemon with (4) usuable moves and items; for example, Pidgeotto1 and Masquerain1. For a sense of power creep, remember that Plusle1 is pretty good this round)

Round 2: Dumb AI has the exact same bad movesets with a few new ones thrown in. "Middle cup" should be an easy round too.

(Tips: The Charmeleon1 lead is still good. But by now I consider Plusle1 a pretty bad pokemon, considering the other evolved pokemon's bulk. So look for evolved pokemon of your own. Psychic types are great this round as long as you are conscious of the dark types (Sableye1, Cacturne1, Sharpedo1). Remember that Future Sight affects dark types, which is very relevant this round. Fighting type leads are strong and make for quick paced battles. Normal types are nice when paired with a rock and steel answer.)

Round 3: It feels like the Moderate AI has movesets tailor made for it, with many status options and solid coverage. A tricky metagame and a Noland fight? I consider this a potentially challenging round.

(Tips: Rain teams can be strong (Omastar2, Gorebyss2, Magneton2). Jumpluff2 is a run ender so familiarize yourself with it. The fast fighting lead is still strong. Chansey2 is not an auto win, but close. Grumpig2, Scyther2, and Gligar2 are some of my favorites.)

Round 4: Moderate AI and limited damaging moves means powerful STAB will be selected often, and with status options plaguing the round, it's moderately difficult. The starting draft having 12 IVs could now be enough for them to speed creep, i.e. outspeed opponents with a higher base speed stat than they have.

(Tips: Maintaining synergy is always important, but you'll likely be switching into status. Coincidentally, a lot of viable pokemon run Lum this round (Gyrados1, Espeon1, Milotic1))

Round 5: In this round, smart AI is pulling the strings of the most hated sets in the Frontier. Set 2 is home to some of the most dangerous sets (e.g. Gardevoir2 and Snorlax2) and yet the worst sets too (see Salamence2, Crobat2, Flygon2 and Porygon2-2). Consequently, it is a real race to see who can get the better set up sets.

(Tips: Toxic users can shut down a lot of pokemon and pesky double teamers, but beware of rest users and Immunity Snorlax2. I like to have a flying type/levitator and a Rock Slide resist just to PP stall Snorlax2. Don't even consider Salamence2. It's a trap.)

Round 6: Noland round where Smart AI makes use of the OHKO moves (so sturdy pokemon get an extra nod here). Ideally you'll have the defensive synergy necessary to dance around the coverage moves. This can sometimes can be a tough round. Starmie3 is lurking.

(Tips: Steelix2 and Forretress2 are powerful Explosion users and act as a get-out-of-jail-free card against anything you don't like. If either have sturdy you can obliterate Dewgong3. Ludicolo3 is a streak ender, so familiarize yourself with the set. Gengar3 is now good. Be on the lookout for specially defensive or thick fat pokemon to make good pivots against the plethora of elemental punches flying around this round.)

Round 7: Smart AI has all the coverage and set up it needs. Build a solid team quickly. Expect the most traditional coverage moves, for example, earthquake/rock slide and surf/ice beam. This is a tough round.

(Tips: All of Round 6's tips still apply. Traditionally "good" pokemon are particularly good this round, and most "bad" pokemon actually have redeeming movesets. It goes without saying that Blissey is good every round, but Blissey4 sets up so easily (if there are no thick fat water/ice types) that it should be a crime.)

Round 8+: Smart AI piloting unknown sets means this is as hard as the Factory gets. Scouting finally becomes a factor.

(Tips: Protect, Endure Substitute users gain viability (other than beating Slaking) in the form of scouting. Most electric types can handle the birds, whereas most fighting types can handle the Regi's. Lati@s walls/counters are worth holding onto; ideally you have the Lati wall/counter in the back, so you can swap for Lati@s in the lead.)

And then Open Level rounds look something like this:

Round 1: Dumb AI combined with many pointless moves for opponents to select makes this round easy.

Round 2: Dumb AI won't properly use set up moves and plenty of bad sets exist. I consider this round easy-ish (Side note: I'd recommend attacking chesto resto users immediately, even when in a bad matchup, so hopefully they rest, eat their chesto, then put themselves to sleep again).

Round 3: Moderate AI has many coverage moves at its disposal but is still randomly firing them off. This round could be difficult if you can't get a good team before Noland or if you get exploded upon/OHKO'd in the process.

Round 4: Moderate AI is still in effect and only smart enough not to status twice; it still chooses damaging moves randomly. These are typically the best sets, so hopefully you'll patch together offensive synergy (defensive synergy is less important because that implies switching, which is still a bad idea). The 12IV is starting to cause a nice IV gap between the opponent and as such this round is traditionally easier than Round 3.

Round 5 and beyond: Smart AI has unknown sets and won't waste time setting up. In Round 6 you have to challenge Noland's 31IV team with your 18IV and 3IV pokemon. This is as hard as it gets.

If your goal is just to get the gold symbol, you'll find that Open Level goes way faster than Level 50. You can probably challenge Noland2 twice by the time it takes to get to Noland2 once in Level 50. I attribute this solely to how slow-paced Level 50 Rounds 1-3 are. But if your goal is to get an even higher streak, you are studying the spreadsheet of pokemon and you have the patience to spend many, many hours playing, I'd recommend Level 50, because having a smaller pool of pokemon to sift through will help you make more informed decisions, at least up until Round 8.
Team-building Archetypes

A) Concrete team-building notes:

-PP stalling is wonderful and opens doors for other teammates (Blissey, recover Milotic and Starmies 7 and 8), but know your crit range and avoid it!

-Enfeeble then set up (think Latias4). Know your set up pokemon's limitations because there is no truly safe battle (e.g. +6 calm mind Lati@s can fall to quick claw Scizor, +6 Gyarados can't beat Skarmory, etc.).

-Explosion users are very useful and sometimes having explosion even allows them to transcend their weaknesses. Who cares if water is super effective on rock if Regirock tanks any Suicune's surf and revenge kills it with explosion? They usually end up being my answer to double teamers. Just avoid leaving them as your last pokemon, because your last pokemon fainting on the same turn your opponent's last pokemon will result in a loss no matter what.

-Destiny Bond is similar to explosion in that it is a tool to get you out of problematic pokemon for your team. Sometimes you'll want to awkwardly hard switch your ghost in during a 2v2 to avoid keeping them last, for the same reason as above.

-A Lati@s lead followed by a special wall counter (oftentimes an explosion user, e.g. Steelix) is an effective strategy, though you should fear Umbreon.

-A Raikou lead partnered with a ground type switch in/counter is another effective strategy. Beware of Marowak with swords dance and max speed investment, however. Raikou6 doesn't even need two other teammates, as he literally sets up on any pokemon without earthquake, while also PP stalling, with little effort.

-Suicune is good, but it's only B-tier in my opinion. Only two sets have rest: Suicune1 is walled by water absorb, and Suicune6 is only in Open Level. Scope lens leaf blade Sceptile is always scary to Suicune.

-Protect or fast substitute can be used as a method of scouting the enemy pokemon's set, as long as you remember that they could set up during that passive turn.

-Toxic/will-o-wisp are nice back-up strategies for troublesome pokemon. If your team has good synergy/immunities, you can switch around while their HP is whittled away.

B) Abstract team-building: First and foremost, you want a solid team that minimizes the number of threatening pokemon or pokemon that will end your streak. Refer to every pokemon by its set name (Tauros1, Starmie8, etc). There is no need for ambiguity when you can type in any/every match up in the damage calculator. Say you have 18 swaps and have a "powerful" (31 IV) pokemon in your draft, a Houndoom. Now, a 31 IV Houndoom4 (95 base speed with 252 EVs) is pretty amazing because it outspeeds every Lati@s and is guaranteed to tank one of their earthquakes, whereas 3IV Houndoom4 gets outsped and OHKO'd in the same scenario. I think deeply about what 3IV enemy pokemon I'll encounter and how my team will fare. Think in the "right now" and reassess your situation every single turn. Never lose sight of your path to victory. Give preference to water types, because waters check a lot of types (primarily steel, fire, ground types) that they then partner with well. Give preference to fast pokemon, which I consider 161+ speed in Open Level and 83+ speed in Level 50. This is to outspeed Marowak2-4, who can take a hit setting up, and then demolish slower teams. If you can outspeed the opponent, that's one less turn to get attracted, confused, or exploded upon. One of my best streaks ended because my three pokemon couldn't stop a Crobat4 using confuse ray. Fast pokemon are prone to missing moves (as they often have powerful but low-accuracy moves) and are vulnerable to quick claw, while slow pokemon are prone to getting statused and critically hit.

C) Speed creeps for starting draft pokemon. These are IV boosts that cause the pokemon to outspeed when they would otherwise get outsped

Round 4 - 12IV Blastoise1 ouspeeds 3IV Glalie1, Meganium1, base 80s with 0 speed EV's.
12 IV Magmar1 ouspeeds 3IV Nidoking1, Heracross1, Kingdra1.
12IV Feraligatr1 outspeeds 3IV Venusaur1.

Round 5 - 15IV Houndoom2, 15IV Heracross2 and 15IV Arcanine2 ouspeed base speed 100's with 252 speed EV's.

Round 6 - 21IV base speed 100's (Ninetails3, Raichu3, Typhlosion3, Fearow3 etc.) outspeed the 3IV base speed 105's (Manectric3 and Electabuzz3).
21IV Rapidash3 outspeeds 3IV base 100's with 252 speed EV's.

Round 7 - 31IV base speed 100's with 252 speed EV's (Charizard4, Dodrio4, Zapdos1 etc.) outspeed base speed base 110's with 252 speed EV's (Gengar4 and Espeon4).
One day I noticed green_typhlosion's Level 50 Factory streak, where he swapped nearly every battle and won 64 battles (this record was actually never added to the leaderboard for some reason). Up until then, I'd always swapped conservatively and tried to build a balanced team for any opponent. But when the assistant tells you what type of team you'll be facing, a lot of times it pays off to build a lopsided team to counter the next opponent (especially in the lower rounds). It would only be scary if the opponent could exploit a weakness of your team, but this can be patched by poor AI move choice and mitigating risk. So my new playstyle involves, for the first 4 rounds, swapping as constantly as possible, learning as much as possible about all pokemon sets by trying them out, and boosting my elevations. The exception to this is in Round 3, when you have a Noland battle at the end of the round. I will stop swapping early in the round if I'm confident in the team. The elevations are quite nifty down the line, as they will potentially let you speed creep with offensive pokemon and/or stall more effectively with defensive pokemon. Not to mention you're almost always battling any Noland at a huge IV disadvantage, so I do everything possible to even the playing field. These elevations are actually different based on what round you're on, and are the same for Level 50 and Open Level.

Round 3 Elevations are 15IVs.

Round 4 Elevations are 18IVs.

Round 5 Elevations are 21IVs.

Round 6+ Elevations are 31IVs.
When I say identify your current team threats, I don't mean saying "I have one water weakness and no resists", I mean what tangible set will you run into that will wreck you? Think about dual typings and common coverage moves, as these will sneak up on you. Load the most threatening possible enemy pokemon of that typing into the damage calc and see how your team would fare. Swap for pokemon that can patch that hole. If your team has three primarily physical attackers, imagine how they would fare against Regirocks, Forretresses or Skarmorys. If yout team has three primarily special attackers, imagine how they would fare against Blisseys, Regices, and Umbreons. If absolutely no synergy is possible, pick fast pokemon. Fast pokemon get flinched less often, can usually guarantee chip damage, and can avoid one turn of infatuation, status, or confusion if they outspeed.

If you encounter one of your threats, play carefully. You might have to sack whatever pokemon is currently in just to get a pokemon to 2HKO in return. It may seem dumb to sack your Snorlax2, but say you lead Fearow2 against Nidoking2 and you know the opponent uses fighting types. It wouldn't pan out to try to save Snorlax for late battle where he'd likely get cross chopped. Try to plan out each pokemon's role in every battle. Don't lose sight of the path to victory.
Scout as much as possible. In battle, scout the enemy team by paying attention to the next trainer's favored type and playstyle, what pokemon is sent out first and second, and what items are in play. With this information, map out each individual battle accordingly. I hope to prove that sorting through all of the data is worthwhile. You'll always have a definitive list of opposing pokemon to possibly encounter, and you can then create situations where you're least likely to lose.

Now, in two regards the Factory is easier than any other facility. What? How?! Firstly, the IV's are going to be in your favor before you swap (barring Noland battles). In what other facility would you stay in with a Moltres on a max special attack Starmie? The IV boost can also be a godsend if one of your starting draft is at a crowded speed tier. A slower example is 31IV Marowak2-4 getting the jump on base 50's up through uninvested base 85's. Faster examples are 31IV max speed base 90s (Kangaskhan and Moltres), base 95's, and 100 base speed outspeeding the 252 speed base 100 crowd. Furthermore, the Factory assistant tells you if there is a "clear favored typing" from the next opponent. The list of pokemon you can encounter is finite. The opponent can't have your own pokemon (barring Noland, who has had my own pokemon from my current team on two occasions). If you are outrageously thorough and patient, you can map out your battles by plugging in every possibility. It's more useful when there is a favored typing, because the list of possible opponents becomes much more manageable. The player can break item clause but the AI (and starting draft) does not. So you will only face a maximum of leftovers and one chesto berry per battle.

Let's say your oppenent favors water types and they send out a Gyarados first. This tells you that they don't have any other flying types, or else the assistant would have reported that the opponent has no clear favorites. Upon every new turn, you should absolutely reassess what the AI will do, and confirm that your current path to victory is the best possible one. Even if there is no clear favored typing, remember that mid-battle when you see the second pokemon. Unless it's the unlikely instance of 2 double types, like Swampert/Whiscash or two grass/poison types, in which case you already beat the first of said dual type pokemon, so chances are you can beat the second as well. Most of the time when I hear no clear typing and I'm mid-battle, I think "there probably isn't another one of these types in the back."

As an example, say you've heard "no clear favored typing" and your lead Raikou1 (whose only attacking move is thunderbolt) takes out the enemy's lead Weezing. Next your opponent throws out Flygon. Now is big brain time, because you can't be 100% certain that the opponent won't have a Nidoking in the back. So you'll have to play with at least one answer to the Nidos. Using the same example with no clear favored typing, let's say after your lead Raikou kills their lead Weezing, they instead send out Nidoking. Now you know something's up, or else the assistant would have told you the opponent prefers poison types. You must now prepare yourself to encounter another ground type pokemon in the last slot.

As soon as you hear a clear favored type, you should imagine how your team would fare against a legendary of that type. Definitely swap if that hypothetical legendary pokemon of that type would beat your team in practice.

The only time I can find usefulness in the battle style of the trainer is when it's "depends on the battles flow", which refers to there being two or more weather inducing moves in the next battle; And since there's no pokemon with two weather moves on one moveset, you can start only looking at pokemon that round that have weather movesets and seeing what makes sense combined with any type favoring or items already presented.
Now put it all together! I always have the damage calculator with have both pokemon loaded up in one hand and spreadsheet pulled up in Excel in the other.

To start open the attached spreadsheet below in Microsoft Excel and you will see all of the pokemon sets with my addition of the Type column. Firstly, you should click on "Set #" and filter by using "number filter" and "equals" and typing the correct set number. This will make sure you are only looking at the potential pokemon this round.

Second, if the Factory assistant gives you a favored type, you'll want to refresh yourself on what possible pokemon of that type you may face. Filter by clicking "Type" and filtering using "text filters" then "contains..." and type the favored type.

Next, after seeing the opponent's first pokemon, and possibly after scouting, you'll know what the first pokemon's item is. So before attacking any further you'll want to click the "Item" cell and then "text filter" but now hit "does not contain..." whatever item the first enemy pokemon was holding. You can only do this filter once, even though you will find out the second item eventually.

That is the best I can narrow down the potential sets so far. My next project was to add another column entitled "Battle Style" and for every pokemon's every move type out the corresponding battle style category. This would look something like: 1,3,6,1. With that new column, you could run a separate filter for the corresponding battle style when the assistant tells you. It would also be nice to update the speed stats of all of the pokemon to reflect their speeds with 3IVs. This would also be nice to order the results in descending order and be hyper aware of your speed tiers that round. Alternatively I could rearrange the results in descending order of base stat total, if I was to go through and make a new column for base stat totals.
While playing the Battle Factory exclusively for over year, I lost a lot. But what I started doing early on was writing a short report on how I lost. I detailed my team, the team I lost to and the type of loss I had. There are four ways I lost:

Hax - happens when I choose to map my battle out a certain way and it was undone by an attack of mine missing or a hax item activating.

Counter-team - happens when the opponent's team has great answers to my pokemon and there was nothing I could do.

Bad Draft - happens when you lose the first battle of a round due to geting a draft of bad pokemon/bad sets. This could also be a lack of synergy. I once had 6 psychic types to choose from. You can also get bad duplicates due to an elevation (Wailord1 and Wailord2).

Misplays - happens when I lost when there was something I could have done to win.

The purpose of writing reports was to minimize misplays and to grow from every loss. You will certainly lose in unfair ways and when you lose to hax, counter-teams, and bad drafts you cannot beat yourself up. You can only move on and start again.
There are still mysteries within the Factory that I can't figure out. Either Bulbapedia or my Emerald cartridge is wrong. I'm namely referring to the starting draft pokemon IV values. Bulbapedia would have you believe:

Round 1 = 3IV starting picks
Round 2 = 6IV starting picks
Round 3 = 9IV starting picks
Round 4 = 12IV starting picks
Round 5 = 15IV starting picks
Round 6 = 21IV starting picks
Round 7+ = 31IV starting picks
However, I have hit Round 9 multiple times and, about half of the time, I encounter utterly random IVs for every single stat on every single starting draft pokemon (with the exception of one or two 31IV pokemon, depending on my swaps). It's mostly a gripe because of how much more time it takes to load up every random IV for a pokemon multiple times in a battle. But this is not absolute, because sometimes Round 9 will give me random IV draft picks and then by Round 10 the draft goes back to all being 31IV.

I've also been in a situation where I swapped 38 times and expected 4 elevation pokemon, yet only the last 2 pokemon of the starting draft had 31IV's. After winning the first battle, I swapped for the enemy's pokemon in the third party slot. I wish I would have instead swapped for the first or second pokemon slot, because I suspect these pokemon may have received the 31IVs instead of me! I base this off of a breakdown of the Emerald Factory coding that juxtaposes the starting draft and the first enemy trainer.

https://docs.google.com/spreadsheet...ouid=101302679580129866296&rtpof=true&sd=true
I took my time to read all of this and thanks a lot for sharing a very extensive and thoughtful resource like this. I'll add this to the main body of the forum when I get the chance since I am sure this will help a lot of people who want to do Factory and complete it. Keep up the hard work! :)
 
This post is to summarize the immense work that has been done to improve Dome Assistant in the last couple of weeks. For clarity, I'll describe what Dome Assistant does and how the newest version, Dome Assistant 2.0, improves the older versions.

The program first uses the trainer card info and possible known moves via the tournament of your next opponent to eliminate possible teams they can bring. Both the battle style and the EV style are very important in this reduction; sometimes it's possible to reduce 370 team options to just a few.

All versions 1.x of DomeAssistant were already able to do this reduction quite consistently, aside from minor rounding mistakes that have now been corrected. The main improvement lies in the selection prediction, i.e. what Pokemon will the opponent select for the actual battle?

According to the information on Bulbapedia, this is done by looking at the effectiveness of their moves against your Pokemon. There are two scoring methods, offensive and defensive. The info was easy to implement, but time after time the program made inexplicable mistakes so eventually I gave up on it and deleted the entire prediction utility.

Revisiting this with Mow and The First Letter Of The Alphabet on the discord server, we had endless talks, tests, source code research and technical discussions about the matter.

As it turns out, most of the mistakes were explained by:
  • Levitate and Wonder Guard not working as expected;
  • The opponent's Pokemon selection depending on your selection;
  • The Pokemon you DON'T select is regarded as a "MissingNo", having Normal-type for scoring purposes.
After precisely implementing all new information, DomeAssistant now does the following:
  • It asks for the Trainer Name;
  • It asks for the three Pokemon species that are shown;
  • It asks for Battle Style and EV Style and if any info reduces the team options to just 1, it outputs that option.
  • If there is still more than one team option left, it asks for extra move info from the Tourney Tree;
  • Then it prints all of the possible teams together with items, moves and the actual Speed stats (taking into account the 3IV glitch of course).
  • For every Pokemon on their team, three occurrence probabilities are assigned that become very easy to interpret once you get used to the syntax:
    • Probability to see that Pokemon if you omit your first Pokemon;
    • Probability to see that Pokemon if you omit your second Pokemon;
    • Probability to see that Pokemon if you omit your third Pokemon.
  • If there are more than 8 team options left, it gives you some percentages about which Pokemon can have Quick Claw at least.
The Python script is to be executed from command line and requires the .py-file to be in the same folder as the accompanying .csv files. Your team and whether you play on Level 50 or Level 100, has to be written into the .py-file, before the "don't touch anything below" comment. Below is an example of how to interact with the program:

Code:
------------------------------------------------------------------------------------------------------------
WELCOME TO DOME ASSISTANT! The currently loaded team is Tauros, Latios, Medicham.
------------------------------------------------------------------------------------------------------------
Who is your next opponent? Fill in the trainer name.
Leon
------------------------------------------------------------------------------------------------------------
Fill in the different species you see, separated by Enters.
Ampharos
Claydol
Gengar
------------------------------------------------------------------------------------------------------------
There are 42 possible teams.
------------------------------------------------------------------------------------------------------------
Select the battle style from the following list:                                      
(0) Willing to risk total disaster at times      (14) Skilled at upsetting foes emotionally
(1) Skilled at enduring long battles             (15) Uses strong and straightforward moves
(2) Varies tactics to suit the opponent          (16) Aggressively uses strong moves
(3) Has a tough winning pattern                  (17) Battles while cleverly dodging attacks
(4) Occasionally uses a very rare move           (18) Skilled at using upsetting attacks
(5) Uses startling and disruptive moves          (19) Uses many popular moves        
(6) Constantly watches HP in battle              (20) Has moves for powerful combinations
(7) Good at storing then loosing power           (21) Uses high-probability attacks  
(8) Skilled at enfeebling foes                   (22) Aggressively uses spectacular moves
(9) Prefers tactics that rely on luck            (23) Emphasizes offense over defense
(10) Attacks with a regal atmosphere             (24) Emphasizes defense over offense
(11) Attacks with powerful, low-PP moves         (25) Attacks quickly with strong moves
(12) Skilled at enfeebling, then attacking       (26) Often uses moves with added effects
(13) Battles while enduring all attacks          (27) Uses a well-balanced mix of moves
0
------------------------------------------------------------------------------------------------------------
There are 17 possible teams left.
------------------------------------------------------------------------------------------------------------
Is there emphasis (E) or neglection (N) of any stats, or is the team "well-balanced" (B)?
E
------------------------------------------------------------------------------------------------------------
Which stat(s)? Type "none" if the second input is not necessary.
Atk
none
------------------------------------------------------------------------------------------------------------
There are 3 possible teams left.
------------------------------------------------------------------------------------------------------------
Finally, enter any moves you know via the Tourney Tree, or "none" if you don't know.
Focus Punch
Fire Punch
none
------------------------------------------------------------------------------------------------------------
TEAM 1
100%  50%  67%      Ampharos 2 BrightPowder   Hardy ThunderPunch   Fire Punch  Focus Punch Thunder Wave   61
100% 100% 100%       Claydol 4   Focus Band Adamant      Psychic   Earthquake  Shadow Ball    Explosion   81
  0%  50%  33%        Gengar 1    Leftovers   Timid  Dream Eater     Hypnosis  Confuse Ray      Attract  150

TEAM 2
50%  50%  50%       Ampharos 2 BrightPowder   Hardy ThunderPunch   Fire Punch  Focus Punch Thunder Wave   61
100% 100% 100%       Claydol 4   Focus Band Adamant      Psychic   Earthquake  Shadow Ball    Explosion   81
50%  50%  50%         Gengar 2    Leftovers Adamant  Sludge Bomb  Shadow Ball  Confuse Ray  Will-O-Wisp  148
For example, if it's Team 1 then you are certain to see Ampharos + Claydol if you take Latios and Medicham (omitting Tauros).

Edit: if you are interested in my Dome Doubles team, here it is: https://pokepast.es/687c1447444709a1. No interesting matches yet in 74 consecutive tournaments.
 
Last edited:
MERRY CHRISTMAS!

Since I've been spending a lot of time on Dome Assistant the last weeks, I figured I'd share my 2021 Christmas wishes to you all by sharing the Dome Christmas team I've been using and optimizing along the way:

untitled(1).png


Of course, it functions all the better with the new predictions Dome Assistant gives. It's currently at 23 tournament wins and quite fun to play, since Delibird sees some use occasionally:

1639929101525.png

JINGLEBELL (Stantler) @ [No Item]
Ability: Intimidate
EVs: 228 HP / 20 Atk / 44 Def / 204 SpD / 12 Spe
Careful Nature (+SpD, -SpA)
- Thief
- Thunder Wave
- Sand-Attack
- Earthquake


Most of the time, this is the crippling lead of choice. Works the same as on the Battle Tower Christmas challenge from last year (featuring a Suicune instead of the way cooler Articuno) and has important roles:
  • Make sure Articuno outspeeds dangerous threats;
  • Pre-crippling opponents so Articuno can fully setup;
  • Steal Quick Claws and sometimes Leftovers if it's something that can't be beaten, such as Leftovers Double Team Registeel;
  • Provide another basic attacker to beat outstalled opponents after Articuno has done the stalling part.
The last bullet explains Earthquake. Ideally I had wanted to use Swift to tear through opponents with Evasion boosts, but Earthquake turned out to be the best choice because it dents Metagross if needed and, in the first place, removes all Jolteon outspeeding Articuno with Dome IV advantages.

Here's some cool things this magical EV spread achieves (note that this is surprisingly only possible without maximum HP EVs because of jump points):

  • Can't be 2HKOed by Starmie Surf, Latios-2 Psychic, Jolteon/Raikou Thunderbolt (or OHKOed by their critical hits of course)
  • Can't be OHKOed by Intimidated Hariyama non-crit Cross Chop or Machamp-3 Cross Chop (who has Quick Claw).
  • Outspeeds even the fast Metagross, as well as base 100s without investment (therefore also Nidoking and Jynx etc, what have you)
  • Has about 90% to 2HKO non-physically bulky Metagross
  • Easily OHKOs Jolteon and Flareon without defensive investment.
  • Guaranteed 3HKO on 255 HP Registeel.

1639929163983.png

MAJESTIC (Articuno) @ Leftovers
Shiny: Yes
Ability: Pressure
EVs: 220 HP / 36 Def / 4 SpA / 4 SpD / 244 Spe
Timid Nature (+Spe, -Atk)
- Protect
- Substitute
- Double Team
- Ice Beam


The carrier of the team, of course, this majestic bird and my favourite Pokemon of all time. Sadly it's not an extremely good Pokemon, but with Dome IV advantages it gets some nice, new attributes. Its main role is simple; outstall opponents while setting up Double Teams. Then KO them with repeated Ice Beams, repeat for the next Pokemon. If Articuno gets stuck against something, it can proceed to make them Struggle and the other team members shall help to finish them off without wasting too much Articuno PP.

Articuno's Ice Beam is decently strong even with the minimal investment given. After an Intimidate, its Substitute easily outbulks stuff like Ursaring's Facade and other middle-strong moves. Articuno's SpD, its main selling point, allows the Substitute to stay intact against Starmie's Ice Beam and very likely Latios-2's Ice Beam as well. Against the more flexible, high-PP special attackers such as these, Mr. Mime, Gardevoir, etc, sometimes at least one Sand-Attack is needed to get the snowball rolling.

It's not Suicune, but it still performs awesomely. Oh yeah, it survives Metagross Meteor Mash at full health if absolutely necessary (hopefully not).


1639935993822.png

MERRY XMAS (Delibird) @ [No Item] / Chesto Berry / Sitrus Berry
Ability: Hustle
EVs: 252 HP / 52 Atk / 180 Def / 4 SpA / 20 Spe
Naive Nature (+Spe, -SpD)
- Aerial Ace
- Thief
- Hail

- Rest

There it is again, our lovely Christmas mascotte. It's always a pleasure to think of a way to make it useful. Unlike on the Tower challenge team, this Delibird has to do a little more sometimes than just kill off Struggling Water Absorbers. At first, I made it a Choice Band Hustler with Aerial Ace, Focus Punch, Blizzard (OHKOs Rhydon uninvested, haha) and PRESENT just for memes. It OHKOed a Kecleon with Focus Punch and also OHKOed some fun things like Houndoom, Fighting-types and a +6 Curselax eventually.

But I needed a more dependable insurance against last-poke stallers in case I can afford to NOT take Stantler. This Delibird is designed to take care of any Struggler that has no Attack boosts; it can give decently powerful stabs with Hustle Aerial Ace, and keeps damaging the foe while Restlooping, possibly stealing their Item in the process if necessary. Hail is also a useful effect if you have to switch back to Articuno later (or just sack Mr. Xmas in order to get it in). Think of Registeel, Suicune, Ludicolo, Blissey, et cetera; they all fall to this thing. Even Double Team Thick Fat Snorlax' Struggles are not enough to break this Santa's restloop without two crits.


xmas.png


I will be streaming this team on Discord until Christmas!
 
Last edited:
This thread is awesome! Some amazing strategies and teams in here, many of which I would never have the patience to actually use. Lots of fun reads and useful resources so major thanks to everyone involved!

Anyway, I suppose I should share my Battle Tower streak on retail. I've been trying to get Level 50 Battle Tower ribbons on a bunch of different Pokemon, so starting with Round 8 I was just going through rounds with Snorlax and two randos and praying it worked, and somehow I kept not losing despite bringing garbage like mostly not EVed Articuno/Zapdos (captured at lv50 from an XD RNG) and no STAB Entei (whoops) because Snorlax is just that much of a beast. Once I got into the high 100s I started bringing more consistent teams consisting of Snorlax, Suicune, and a physical attacker that could cover most of the issues. This obviously isn't an original team at all, but I've just gone with the classic:

Salamence @ Choice Band
Ability: Intimidate
EVs: 4 HP / 252 Atk / 252 Spd
IVs: 30 HP / 30 Atk / 30 Def / 26 SAtk / 30 SDef
Adamant Nature (+Atk, -SAtk)
- Hidden Power [Flying]
- Earthquake
- Rock Slide
- Brick Break

Suicune @ Lum Berry
Ability: Pressure
EVs: 252 HP / 252 Def / 4 Spd
IVs: 11 Atk
Bold Nature (+Def, -Atk)
- Surf
- Ice Beam
- Calm Mind
- Rest

Snorlax @ Leftovers
Ability: Thick Fat
EVs: 140 HP / 96 Atk / 56 Def / 216 SDef
Careful Nature (+SDef, -SAtk)
- Body Slam
- Earthquake
- Curse
- Rest

On Anabel rounds, I substitute out Salamence with Marowak so I can clean sweep her (barring Latios Psychic crit or Brightpowder miss, which thankfully hasn't happened yet):

Marowak @ Thick Club
Ability: Rock Head
EVs: 4 HP / 252 Atk / 252 Spd
Jolly Nature (+Spd, -SAtk)
- Earthquake
- Rock Slide
- Double-Edge
- Swords Dance

There have been a ton of other Pokemon involved: Kingdra, Medicham, Machamp, Gengar, Metagross, Slaking, Tauros, and decent-ish IVs Latios, among a couple others I'm forgetting, but a lot of those were one-offs to equip the ribbons.

Some notes: This team is comically weak to OHKO spam Lapras sets. I've faced each of them once. Lapras 7 was an encounter worth sharing. It came in on Suicune, so I started firing off Ice Beams to chip it hoping that Snorlax Body Slam + Salamence Brick Break would kill it. But the OHKO moves kept missing, and with Pressure I got optimistic. I switched Snorlax in to milk the final PP of Sheer Cold and then Leppa reset it on a miss. I got off one Body Slam before it got me, and when I switched Suicune back in, for the first time I can remember, I misclicked the fucking move and hit surf, restoring most of the chip damage from Water Absorb. Thankfully it missed Salamence twice but it would have seriously sucked to lose to that after it was in Brick Break range because I accidentally pressed A too soon and selected the wrong fucking move. You've got to love the Battle Frontier.

Anyway, it's otherwise decently consistent, and while it has its weak spots (Gyarados 4 isn't fun, you have to PP stall most Gengar sets and pray they don't double crit, etc.), usually you can set up Snorlax and Suicune early and steamroll otherwise troublesome Pokemon. Having Snorlax fully PP-boosted has been massively helpful, but I still haven't done it for Suicune yet because I can't decide if I want to make it a ribbon master and do contests with it or not yet. I agonized about Snorlax's EVs a lot and the huge SDef investment was a very, very good idea. I've considered Amnesia, but Earthquake has bailed me out of some sticky situations with Aggron, Electabuzz, Raikou...IDK, I'm just partial to it. The same goes for Ice Beam on Suicune, which is so massively helpful against Salamence and Lati@s that I'm cool without Substitute. I've noticed that the standard Suicune EVs seem to put it at 240 instead of 207, but I'm not sure I've ever missed the Speed. The Defense has been massively helpful though. I might RNG another and try out the alt set with a different lineup and see how it performs though. Finally, Rock Slide is basically useless on Marowak. I use Double Edge all the time, but I almost never want to risk a miss on RS. I think Perish Song is a better option if I could do it over again.

I've had quite a few close calls so I don't think this is equipped for a long streak, but we're up to 280 wins so I'm quite happy with that. To be honest, I'm mostly posting this to jinx the streak and lose because...

Has anyone ever had the game freeze on "Saving...don't turn off the power" on the confirmation to enter the Battle Tower? It happened to me, and not only did it stop my Tower streak, but it also ended my streak at every other facility. I don't know what the hell happened, but now I'm paranoid about it and do one facility at a time. (You can see in my proof pic that there is an otherwise inexplicable previous win streak of 0 on the Open Level record...that 35 had been ongoing and that was the result of the corruption for every facility...maddening.) I'm pretty sick of the Tower and am kind of hoping I lose immediately after my 300th win, but I'm not just gonna intentionally lose obviously.

Before I found this thread, I resorted to writing a Selenium script to scrape the Frontier Pokemon database from Bulbapedia which got converted into a readable .txt file -- actually two, one organized alphabetically and one by Speed tier. I hope they're useful to someone even though they're not breaking any new ground...I just hate spreadsheets.

Oh yeah -- I've noticed some Pokemon in streak posts that, to my knowledge, require crazy difficult or (again, just to my knowledge) impossible RNGs to pull off. Sorry if I missed these elsewhere (or if this is too off-topic I guess), but how on earth do you RNG XD Moltres? Perfect Emerald Method 1 spreads that are waaaaaay far from the static initial seed? Any links/resources would be appreciated!
 

Attachments

Set a new Open Level Arena record of 135 wins with Suicune, Metagross and Latios! Credits to Jellal for the Modest Suicune! He asked me to try Suicune in Arena a while back and I’ve finally found the right spread and combination.

Suicune in the Arena is the perfect combination of bulk and immediate power, fulfilling a different role than the slow CM user we are used to using. Suicune checks dangerous physical pokemon like Metagross, Regirock, Steelix and Tyranitar.

I chose to run this team at Open Level for 3 reasons.

1st- I wasn’t satisfied with “only” 98 wins in the Arena being the Open Level record.

2nd- Tyranitar and Dragonite appear in Open Level, which reduces Metagross appearances. Tyranitar and Dragonite are comparatively easier to deal with than Metagross.

3rd- Suicune and Metagross are bulkier at Level 100 compared to level 50.

The spread is as follows.
1640621645583.png

Suicune @ Salac Berry
Level: 100
Modest Nature
Ability: Pressure
EVs: 152 HP / 204 SpA / 16 SpD / 136 Spe
IVs: 30 Atk / 30 SpA

- Hidden Power Grass
- Surf
- Ice Beam
- Substitute

  • Avoids the cross chop 2 hit ko from Adamant Machamp
  • “usually” avoids the 2 hit ko from Starmie’s thunderbolt.
  • Faster than all Metagross and all variants of Breloom with Spore
  • Reaches 360 speed at +1, enough to outspeed everything besides Jolteon 4
  • Remainder goes into SpA
The first 3 moves are self-explanatory. Surf/Ice Beam/ Hp Grass hits everything neutrally and has solid power behind it with a Modest nature. Hp grass is necessary given that Mirror coat Swampert, Quagsire and QC whicash exists. Mantine is rare and Gyarados in the worst case scenario is less dangerous than Quagsire/QC whiscash.

The last move was harder to pin down. These were the moves I experimented: Icy Wind, Reflect and Rest and Mirror Coat. These moves were far too situational and felt like a waste of a move-slot most of the time. It was @Actaeon’s suggestion to consider Salac Berry with Substitute. I started using it and didn’t turn back. Suicune makes an excellent user of Substitute to block mirror coat, tank hits while leaving the substitute intact and allows you to “potentially” beat faster special threats like Gengar, Sceptile and Latios. Suicune tanks the super effective hit, slow substitutes and activates salac berry. Suicune can then 2 hit ko these targets and give you an advantage.

252+ Atk Level 100 Machamp Cross Chop vs. 152 HP / 0 Def Level 100 Suicune: 160-189 (42.2 - 49.8%)

252+ SpA Level 100 Starmie Thunderbolt vs. 152 HP / 16 SpD Level 100 Suicune: 166-196 (43.7 - 51.7%) -- 10.9% chance to 2HKO

252+ SpA Level 50 Starmie Thunderbolt vs. 152 HP / 16 SpD Level 50 Suicune: 88-104 (45.3 - 53.6%) -- 36.3% chance to 2HKO

1640621747831.png
Metagross @ Quick Claw
Level: 100
Adamant Nature
Ability: Clear Body
EVs: 152 HP / 140 Atk / 44 SpD / 172 Spe

- Aerial Ace
- Explosion
- Earthquake
- Shadow Ball

  • “Usually avoids the 2 hit ko from Starmie”
  • Enough speed to out-speed most metagross sets
  • Dumped the rest in attack.
Metagross is the glue of the team. It checks strong special threats, beats double team grass types like Ludicolo, Vileplume and Meganium and booms on fat specials walls. It has severe 4 moveslot syndrome, but it accomplishes what I need it too. Quick Claw having a 20% activation rate has been pretty clutch here. Allows a low HP metagross to cheese out one more win. Persim Berry, Chesto Berry and Cheri Berry are also viable.

252+ SpA Level 100 Starmie Surf vs. 152 HP / 44 SpD Level 100 Metagross: 148-175 (43.6 - 51.6%) -- 10.5% chance to 2HKO

252+ Atk Level 100 Rhydon Earthquake vs. 152 HP / 0 Def Level 100 Metagross: 287-338 (84.6 - 99.7%) -- guaranteed 2HKO

1640621837669.png
Latios @ Lum Berry
Level: 100
Timid Nature
Ability: Levitate
EVs: 40 HP / 252 SpA / 216 Spe

- Calm Mind
- Psychic
- Thunderbolt
- Ice Beam

Set Explanation
  • Outspeeds Sceptile and Aerodactyl
  • Rest in SpA and HP
Latios is a very safe third pokemon and can clutch out wins in some dire situations.

Overly Comprehensive List of Difficult matchups​

Porygon 2- Set 4 with Bright powder (Bolt beam coverage with recover). Usually beats Suicune and ties with Metagross unfortunately.

Latios/ Gengar/Starmie can paralyze or crit Suicune, instantly putting me at a disadvantage.

Regice leads that use Thunderbolt. Explosion Target.

Hilariously lost to Seismic toss Chansey in the early rounds.

Houndoom in slot 2 or 3 (if Suicune somehow loses)- It has happened 2x now, which I’ve clutched out. I also got this ridiculous roll against Houndoom. Should have lost at 130 actually.

252 SpA Latios Thunderbolt vs. 0 HP / 0 SpD Houndoom: 125-148 (42.9 - 50.8%) -- 3.9% chance to 2HKO

Heracross can crit Suicune and leave a large dent into Metagross (if there’s no QC activation)

Sunnyday Solar Beam fire types. Encountered some minor difficulties with fire types setting up sun, solar beaming Suicune and leaving Metagross a little vulnerable.

Tyranitar/Metagross in slot 2 or 3 (if Suicune is gone).

Loss at 135. https://vimeo.com/manage/videos/660498848
Just barely missed out on the damage roll against Regice and miscalculated the damage against Registeel. Shame, but this team did put in the work.

https://pokepast.es/a6b2a455346ce279

Open Level Arena Record.png
 
Last edited:
Set a new Open Level Arena record of 135 wins with Suicune, Metagross and Latios! Credits to Jellal for the Modest Suicune! He asked me to try Suicune in Arena a while back and I’ve finally found the right spread and combination.

Suicune in the Arena is the perfect combination of bulk and immediate power, fulfilling a different role than the slow CM user we are used to using. Suicune checks dangerous physical pokemon like Metagross, Regirock, Steelix and Tyranitar.

I chose to run this team at Open Level for 3 reasons.

1st- I wasn’t satisfied with “only” 98 wins in the Arena being the Open Level record.

2nd- Tyranitar and Dragonite appear in Open Level, which reduces Metagross appearances. Tyranitar and Dragonite are comparatively easier to deal with than Metagross.

3rd- Suicune and Metagross are bulkier at Level 100 compared to level 50.

The spread is as follows.
View attachment 394800

Suicune @ Salac Berry

Level: 100

Modest Nature

Ability: Pressure

EVs: 152 HP / 204 SpA / 16 SpD / 136 Spe

IVs: 30 Atk / 30 SpA

- Hidden Power Grass

- Surf

- Ice Beam

- Substitute

  • Avoids the cross chop 2 hit ko from Adamant Machamp
  • “usually” avoids the 2 hit ko from Starmie’s thunderbolt.
  • Faster than all Metagross and all variants of Breloom with Spore
  • Reaches 360 speed at +1, enough to outspeed everything besides Jolteon 4
  • Remainder goes into SpA
The first 3 moves are self-explanatory. Surf/Ice Beam/ Hp Grass hits everything neutrally and has solid power behind it with a Modest nature. Hp grass is necessary given that Mirror coat Swampert, Quagsire and QC whicash exists. Mantine is rare and Gyarados in the worst case scenario is less dangerous than Quagsire/QC whiscash.

The last move was harder to pin down. These were the moves I experimented: Icy Wind, Reflect and Rest and Mirror Coat. These moves were far too situational and felt like a waste of a move-slot most of the time. It was @Actaeon’s suggestion to consider Salac Berry with Substitute. I started using it and didn’t turn back. Suicune makes an excellent user of Substitute to block mirror coat, tank hits while leaving the substitute intact and allows you to “potentially” beat faster special threats like Gengar, Sceptile and Latios. Suicune tanks the super effective hit, slow substitutes and activates salac berry. Suicune can then 2 hit ko these targets and give you an advantage.

252+ Atk Level 100 Machamp Cross Chop vs. 152 HP / 0 Def Level 100 Suicune: 160-189 (42.2 - 49.8%)

252+ SpA Level 100 Starmie Thunderbolt vs. 152 HP / 16 SpD Level 100 Suicune: 166-196 (43.7 - 51.7%) -- 10.9% chance to 2HKO

252+ SpA Level 50 Starmie Thunderbolt vs. 152 HP / 16 SpD Level 50 Suicune: 88-104 (45.3 - 53.6%) -- 36.3% chance to 2HKO

View attachment 394801
Metagross @ Quick Claw

Level: 100

Adamant Nature

Ability: Clear Body

EVs: 152 HP / 140 Atk / 44 SpD / 172 Spe

- Aerial Ace

- Explosion

- Earthquake

- Shadow Ball

  • “Usually avoids the 2 hit ko from Starmie”
  • Enough speed to out-speed most metagross sets
  • Dumped the rest in attack.
Metagross is the glue of the team. It checks strong special threats, beats double team grass types like Ludicolo, Vileplume and Meganium and booms on fat specials walls. It has severe 4 moveslot syndrome, but it accomplishes what I need it too. Quick Claw having a 20% activation rate has been pretty clutch here. Allows a low HP metagross to cheese out one more win. Persim Berry, Chesto Berry and Cheri Berry are also viable.

252+ SpA Level 100 Starmie Surf vs. 152 HP / 44 SpD Level 100 Metagross: 148-175 (43.6 - 51.6%) -- 10.5% chance to 2HKO

252+ Atk Level 100 Rhydon Earthquake vs. 152 HP / 0 Def Level 100 Metagross: 287-338 (84.6 - 99.7%) -- guaranteed 2HKO

View attachment 394802
Latios @ Lum Berry

Level: 100

Timid Nature

Ability: Levitate

EVs: 40 HP / 252 SpA / 216 Spe

- Calm Mind

- Psychic

- Thunderbolt

- Ice Beam

Set Explanation
  • Outspeeds Sceptile and Aerodactyl
  • Rest in SpA and HP
Latios is a very safe third pokemon and can clutch out wins in some dire situations.

Overly Comprehensive List of Difficult matchups​

Porygon 2- Set 4 with Bright powder (Bolt beam coverage with recover). Usually beats Suicune and ties with Metagross unfortunately.

Latios/ Gengar/Starmie can paralyze or crit Suicune, instantly putting me at a disadvantage.

Regice leads that use Thunderbolt. Explosion Target.

Hilariously lost to Seismic toss Chansey in the early rounds.

Houndoom in slot 2 or 3 (if Suicune somehow loses)- It has happened 2x now, which I’ve clutched out. I also got this ridiculous roll against Houndoom. Should have lost at 130 actually.

252 SpA Latios Thunderbolt vs. 0 HP / 0 SpD Houndoom: 125-148 (42.9 - 50.8%) -- 3.9% chance to 2HKO

Heracross can crit Suicune and leave a large dent into Metagross (if there’s no QC activation)

Sunnyday Solar Beam fire types. Encountered some minor difficulties with fire types setting up sun, solar beaming Suicune and leaving Metagross a little vulnerable.

Tyranitar/Metagross in slot 2 or 3 (if Suicune is gone).

Loss at 135. https://vimeo.com/manage/videos/660498848
Just barely missed out on the damage roll against Regice and miscalculated the damage against Registeel. Shame, but this team did put in the work.
View attachment 394836
Congratulations on breaking the Arena record once again!!! Glad to know it has worked out for you and proud of your progress! Updated the Arena leader-board as well. You'll have a new challenger soon! ;)
 
Doing another post on the Arena as well. It's about the Open Level modification of my classic "GOD SPEED" Lax / Latios / Medicham team, and after much testing and brainstorming (mainly with my Arena comrade submenceisop ), I concluded it's still the best setup for Arena even with Tyranitar and Dragonite around:

1641129453593.png

GOD (Snorlax) @ Lum Berry
Ability: Thick Fat
EVs: 232 HP / 252 Def / 20 SpD / 4 Spe
Careful Nature (+SpD, -SpA)
- Yawn
- Brick Break
- Shadow Ball

- Rest

It might have new threats to deal with in Tyranitar and Dragonite, but Snorlax is bulkier than it has ever been in Level 100. The shift towards physical bulk is to be more certain of avoiding very powerful physical 2HKOs, such as Ursaring and Metagross (MM boost + MM; you have 91% chance to win against 170+ Atk MM-spamming Metagross as per 100.000 simulations). The physical bulk is immense; stuff like Flygon very likely won't even 2HKO you if they crit an Earthquake. If Snorlax starts a battle at non-full HP the crazy bulk helps to nail the Body category. Doesn't need much explanation if you know the Level 50 team, but Brick Break is essential for Tyranitar here and Shadow Ball gives it neutral coverage. Also dents Psychic-types greatly and wins Skill even if you use Rest Yawn-less. This compensates for the fact that Snorlax got a little less bulky on the special side. STAB is missed sometimes, e.g. this Snorlax usually doesn't even break a Zapdos Substitute.

1641129465323.png

ENDIKU (Latios) @ Cheri Berry
Ability: Levitate
EVs: 40 HP / 252 SpA / 216 Spe
Timid Nature (+Spe, -Atk)
- Calm Mind
- Psychic
- Dragon Claw

- Thunderbolt

Not much has changed; this doesn't need an overly complicated EV distribution anymore, since all SpA matters and the "free level 100 bulk" ensures it already lives through Ursaring Double-Edge and stuff like Glalie's STAB unboosted. Ice Beam was replaced by Dragon Claw to have a better matchup against most Psychics (e.g. Hypno/Gardevoir/Espeon), fellow Lati twins, and Houndoom. At +1 it does most things Ice Beam would do as well, except against some bulkier Ground-types. The Steelix matchup got a LOT worse, sadly, but most of the time you still power through it.

1641129483945.png

DALAI LAMA (Medicham) @ Salac Berry
Ability: Pure Power
EVs: 252 Atk / 4 Def / 252 Spe
Adamant Nature (+Atk, -SpA)
- Brick Break
- Shadow Ball
- Endure

- Reversal

Same old Medicham except for the nickname. The Level 100 bulk helps Medicham; it still OHKOs all Metagross with Reversal, something other Reversal-users like Heracross can only dream of. Basically we concluded that Medicham outclasses any Reversal Heracross. Shadow Ball's coverage is perfect. Ironically Lax now also has this coverage. Of course you can't Endure if Sand is already up, but usually Brick Break is already extremely helpful against teams containing Tyranitar.

https://pokepast.es/17ebaae6d88767c3

Edit: got to 154 and lost battle 155. Sacked Lax to get Latios to +2 safely against a Zapdos, but faced Regice in the middle who lived 2 Psychics by a sliver and boomed on Latios. Last was AA Moltres and I was sure it had Overheat too, so I had to play for an Overheat miss, but didnt get it.
 
Last edited:

QuentinQuonce

formerly green_typhlosion
After a long time, I've managed to climb back up the ranks of Battle Tower Multi and surpassed my old record to sit at 294 wins.

At this point, I'm running exclusively with my record-mixed NPC partner Orienne, since she so thoroughly outranks all the other trainers. Much as I wish it were otherwise, even a generic NPC partner with - say - Gyarados and Zapdos is still inferior to Orienne's team due to the substandard EV spreads and imperfect movesets Frontier mons have. All Lati in the Tower are too slow, and Dragonite is just never good enough.

It's been a long slog to get back to where I was. Took three attempts - the first broke at 253, the second at 188, and now we're here. I'm extremely pumped to finally get past 280 (where my last record broke) and excited to hopefully finally pass 300. I'll probably be continuing the streak tonight so will update later on.

After last time I actually went back and mixed records with another file, giving myself another preset partner who has my team of Tauros and Latios. My thinking was, why give the NPC the less powerful team of Latias and Aerodactyl when I could probably make better use of it. Aerodactyl is faster than everyone else (thus harder to play around) and when the random team order means that it's Tauros/Aerodactyl leading it often hesitates to use Earthquake because it damages Tauros, even if it would result in a double KO on the opponent's side.

However, it didn't end up being that effective. Tauros only having two moves (Double-Edge and Earthquake) means that the AI's capacity for stupidity is increased. A human player can play around that shortcoming; the AI can't. Sometimes it's more optimal to use Earthquake when only one foe is hit by it - but the AI doesn't see that.

Case in point:

Battle 164
I'm leading with Latias; NPC partner leads with Tauros.

The enemy trainer sends out Tyranitar. Cool, smack it with Earthquake. But the partner is Skarmory, who Earthquake can't hit. Never mind, Latias can deal with Skarmory easily.

But the AI, counting Earthquake's ineffectiveness against Skarmory as a big negative, instead opts to use Double-Edge.

...on Skarmory.

Lateral thinking also comes into play here. For instance, sometimes it's more optimal to leave Tauros in and ineffectively use Earthquake than it is to switch to Latios and take a hit in the process. But the AI might switch because it can't do damage. Though with that said I've found it often takes the AI a couple of turns to "realise" that they should switch.

So it was a fun experiment, but ultimately the NPC having the weaker team is the most optimal setup. Sadly species clause appears to be in effect, because the NPC partner with Tauros and Latios never showed up when I had both of those on my team.

(I can't say I'm not a little bit disappointed. How cool would it be to wreck the Tower with a team of Tauros, Latios, Tauros, and Latios?)



1641229651597.png




EDIT: Aaaaaaaand it's over. How ironic to finish on 300 exactly, given that that was pretty much where I wanted to get to.

The loss

Orienne had Aerodactyl up first and Latias second, which is the un-optimal arrangement, but needs must. Tauros can never be in the second position, eventually you will be stuck facing a Gengar or Misdreavus.

The enemy trainer(s) sent out Feraligatr and Blastoise, which I correctly divined to be Blastoise1. Having no wish to be hit by Counter, I struck Feraligatr with Double-Edge, bringing it to ~25% while dumbass Aerodactyl used Double-Edge on Blastoise. Feraligatr then OHKOed Aerodactyl with Hydro Pump, leaving Latias wide open to take the incoming Counter damage.

Just me on my own, then. I finish off Feraligatr while Blastoise ineffectually uses Counter again. Slaking comes out and uses Mega Kick on Tauros after Tauros does around 50% with Double-Edge, finishing it. Now it's just Latios left. Since Slaking now has to rest, I kill Blastoise first hoping the final mon will be doable and I can ignore it while I finish off Slaking. But of course it isn't - Lapras comes out. My only way to win now would be if Thunderbolt gets a critical hit and Slaking's Mega Kick misses. Of course, neither of those things happened.

I did not play that as well as I could have done. The smart thing to do would have been to gang up on Blastoise, but with no way of knowing for sure who Aerodactyl would target it was always a dicey decision.


1641242631234.png


And that's me done for good with Multi, I think. I've probably said that before but 300 really does feel like the natural place to jack it in. I'm so proud of having gotten this far and having achieved good results in nearly all of the facilities (even the less-played formats like Palace Double) I'm quite satisfied with my prowess on Gen III's Frontier. Of course, you're never really done with this place (this thread, after all, is living proof of that) but I'm going to devote some time on the Subway from now on. I hope to still see posts on here from everyone, though! Particularly enjoying submenceisop and Actaeon's Arena journey. Keep it up, guys!
 
Last edited:
Set a new Open Level Arena record with Milotic, Heracross and Latios at 148 wins! I'm incredibly proud of my Milotic set (with help from Actaeon optimizing evs). I wanted to grind for 200 wins, but I've been haxed too many times at inopportune moments to grind further. Definitely need to a take a break first.

Milotic is usually inferior to Suicune, but it shines in Arena as a Dive user and access to recover. Dive gives you guaranteed skill points (if the AI chooses to attack) and you almost always win body with the use of recover on turn 3. Milotic also showed me how a fast dive user can easily cheese Greta's Umbreon, which guarantees a win.

The general strategy is to dive on physical users and recover on turn 3 to win Skill points and Body points. Dive counts as an attack (wins Mind points and guarantees a dodge. You do get hit by surf under water, but Milo still wins skill points and shrugs off that like a champ. Dive also pairs incredibly well with leftovers, as you gain health under water(even in sandstorm), helping you win body and out damaging key targets like Metagross. Against special threats, you risk the MC, which is a high risk, high reward play.

As a side note : Milotic MUST be female. Jynx, Latias, Miltank, Blissey all have one set with attract and you want to avoid the infatuation shenanigans.

The spread is as follows.

1641241950795.png


Milotic (F) @ Leftovers
Level: 100
Calm Nature (+ SpD, -Atk)
Ability: Marvel Scale
EVs: 252 HP / 220 Def / 4 SpA / 24 SpD / 8 Spe
Level: 100
- Dive
- Recover
- Icy-wind
- Mirror Coat

Features of the set
  • Avoids the thunderbolt 2ko from Modest Latios
    • 252+ SpA Latios Thunderbolt vs. 252 HP / 24+ SpD Milotic: 168-198 (42.6 - 50.2%)
  • Outspeeds base 100s after a speed drop
  • Outspeeds Magmar 1 and opposing Milotic
  • 4 SpA to increase this Rhydon damage calc.
    • 4 SpA Milotic Dive vs. 0 HP / 252 SpD Rhydon: 329-388 (93.7 - 110.5%) -- 62.5% chance to OHKO
    • 0 SpA Milotic Dive vs. 0 HP / 252 SpD Rhydon: 326-384 (92.8 - 109.4%) -- 56.3% chance to OHKO
Milotic is the first pokemon I've used that can use leftovers effectively in Arena. With access to recover (rather than rest for pokemon like Suicune) and dive, Milotic is an incredibly resilient pokemon. It is also naturally decently fast, which helps in situations where you just dive 3x, to avoid taking any further damage or dodge Haxrein/ Whiscash. It's quite amazing at winning judgements and gets outright koes with mirror coat if you time it right.

-Dive versus physical attackers like Metagross, Aerodactyl
-Dive first against confusion haxxers like Missy, Lanturn. Need to win Mind at least!
-MC directly into Starmie, Gengar.
-Dive 3x into Heracross, fighters. If you dodge a crit and live the rolls, you tie at least.
- Aids Ludicolo and aids grass types. Icy wind first to scout. Be prepared to tie or lose to sleep powder, leech seed or miss timing MC.
-QC hits (BEFORE you dive). Messes up your dodge strategy a bit.
- Strong Normals like CB granbull, Ursaring. If you live (no crit), you tie at least
- Para, confusion hax against electrics like Ampharos on ampharos (which can prevent you from getting off mirror coat)
- Unfortunate status conditions like freeze from tri attack etc.

Water Absorb Users- awkward matchups
Lapras
  • Icy Wind turn 1, in case this is DD. If it is, congrats. You get a free ko.
  • If it reveals T-bolt, MC for the damage
  • If it shows Sheercold, dive to dodge (if you didn't get hit)
Quagsire
  • You win outright or tie at the worst case scenario. Spam icy wind, maybe recover or dive turn 3
Vaporeon
  • Ironically as tough as Lapras, because it has acid armor and you have to risk MC somewhere. If it's body slam/shadowball, you kinda lose Milotic completely.

Everything else, assuming decent luck, Milotic wins. Seriously, it's that good. It's a good member to go ahead of Heracross too, as it destroys MOST physical stuff, most fliers, never loses outright to fire types without attract or confuse ray luck and checks psychic types.

I needed to have a 3rd member in the middle that dealt with the things that Milotic can lose to. I tried Medicham, but it made my matchup against grass types and electric types far worse and didn't help losing to status conditions. Enter: bulky heracross. I tried a lot of sets and this is probably one of the few ways it's not outclassed by Medicham. Medicham does any endure strat better. Heracross with a somewhat fast substitute can block status conditions, gain an advantage, activate swarm and salac somewhat reliably.

1641244719930.png

Heracross @ Salac Berry
Level: 100
Adamant Nature (+ Atk, -SpA)
Ability: Swarm
EVs: 160 HP / 100 Atk / 248 Spe
IVs: 30 SpD / 30 Spe
- Hidden Power Bug
- Earthquake
- Brick Break
- Substitute

Set features
  • Lives a Starmie's psychic
  • Lives a fast 170 SpA Latios Psychic most of the time
  • Avoids a 2ko from Regirock
  • Enough speed to out-speed the Aids Starmie set with t-wave and confuse ray and most gardevoir/Latios sets.
  • Dumped the rest in attack.
Heracross is the glue of the team. It can revenge ko Walrein, Regice, Lapras, normal types and grass types that Milotic can lose to. It can sub on grass types and t-wave and gain a huge advantage. There are times I wish I had endure or rockslide, but I decided that I would keep Heracross's role simple and play it safe. Megahorn would make Heracross far more dangerous, but it's hard to risk because of the miss chance.

170 SpA Latios Psychic vs. 160 HP / 0 SpD Heracross: 292-344 (85.6 - 100.8%) -- 6.3% chance to OHKO
252+ SpA Starmie Psychic vs. 160 HP / 0 SpD Heracross: 285-336 (83.5 - 98.5%) -- guaranteed 2HKO
252+ Atk Regirock Rock Slide vs. 160 HP / 0 Def Heracross: 143-169 (41.9 - 49.5%) -- guaranteed 3HKO

1641129465323.png


ENDIKU (Latios) @ Lum Berry
Ability: Levitate
EVs: 40 HP / 252 SpA / 216 Spe
Timid Nature (+Spe, -Atk)
- Calm Mind
- Psychic
- Dragon Claw
- Thunderbolt


Not much has changed; this doesn't need an overly complicated EV distribution anymore, since all SpA matters and the "free level 100 bulk" ensures it already lives through Ursaring Double-Edge and stuff like Glalie's STAB unboosted. Ice Beam was replaced by Dragon Claw to have a better matchup against most Psychics (e.g. Hypno/Gardevoir/Espeon), fellow Lati twins, and Houndoom. At +1 it does most things Ice Beam would do as well, except against some bulkier Ground-types. The Steelix matchup got a LOT worse, sadly, but most of the time you still power through it. "Cited from Actaeon's post above.

Set Explanation

Latios is a very safe third pokemon and can clutch out wins in some dire situations. I usually run ice beam, but I found dragon claw was a necessity for psychic types and Houndoom. Without dragon-claw, there's a very good chance you just lose to these, which are too dangerous. Latios also going after Heracross is ideal for Greta and for poison types and ghost types (which Heracross can't hit).

Overly Comprehensive List of Difficult matchups

Bird Keepers and hex trainers can be dangerous because of
  • Fast Crobat
  • Zapdos- T-wave shenanigans, sub or t-bolt crit
  • Articuno against Latios
  • Status hax from tri attack, thunderbolt, icebeam
Team is weak to QC physicals if Milotic isn't around.
Psychic types and Houndoom are tough matchups if Milotic isn't there. Riskier for sure.
Electric trainers with status conditions and Jolteon.
This team can't hit shedinja. ROFL. Milotic ties or wins depending on whether it actually uses an attacking move or not.

If Milotic wins, usually I'm in a good spot.

Replay and Loss

I've been grinding for 3+ weeks trying to reach 200 wins. I'm super gutted only reaching 148. I actually got to 148 on my second attempt. Lost to Bite Espeon critting Latios (which forced me to run dragon claw. The next best attempt reached 125, which I will attach a replay below. It's super bullsh+t. Lost AGAIN to that Aerial Ace Registeel.

Rage inducing replay. I just had to avoid a crit there to win. Lost Body by like 1 or 2%.

https://vimeo.com/manage/videos/662076194

https://pokepast.es/c09285c2f28b58ad
 

Attachments

Last edited:
Set a new Open Level Arena record with Milotic, Heracross and Latios at 148 wins! I'm incredibly proud of my Milotic set (with help from Actaeon optimizing evs). I wanted to grind for 200 wins, but I've been haxed too many times at inopportune moments to grind further. Definitely need to a take a break first.

Milotic is usually inferior to Suicune, but it shines in Arena as a Dive user and access to recover. Dive gives you guaranteed skill points (if the AI chooses to attack) and you almost always win body with the use of recover on turn 3. Milotic also showed me how a fast dive user can easily cheese Greta's Umbreon, which guarantees a win.

The general strategy is to dive on physical users and recover on turn 3 to win Skill points and Body points. Against special threats, you risk the MC, which is a high risk, high reward play.

The spread is as follows.

View attachment 396271

Milotic @ Leftovers
Level: 100
Calm Nature (+ SpD, -Atk)
Ability: Marvel Scale
EVs: 252 HP / 220 Def / 4 SpA / 24 SpD / 8 Spe
Level: 100
- Dive
- Recover
- Icy-wind
- Mirror Coat

Features of the set
  • Avoids the thunderbolt 2ko from Modest Latios
    • 252+ SpA Latios Thunderbolt vs. 252 HP / 24+ SpD Milotic: 168-198 (42.6 - 50.2%)
  • Outspeeds base 100s after a speed drop
  • Outspeeds Magmar 1 and opposing Milotic
  • 4 SpA to increase this Rhydon damage calc.
    • 4 SpA Milotic Dive vs. 0 HP / 252 SpD Rhydon: 329-388 (93.7 - 110.5%) -- 62.5% chance to OHKO
    • 0 SpA Milotic Dive vs. 0 HP / 252 SpD Rhydon: 326-384 (92.8 - 109.4%) -- 56.3% chance to OHKO
Milotic is the first pokemon I've used that can use leftovers effectively in Arena. It's longevity, access to recover (rather than rest for pokemon like Suicune) and dive, Milotic is an incredibly resilient pokemon. It is also naturally decently fast, which helps in situations where you just dive 3x, to avoid taking any further damage or dodge Haxrein/ Whiscash. It's quite amazing at winning judgements and gets outright koes with mirror coat if you time it right.

Aids Ludicolo and aids grass types. Icy wind first to scout. Be prepared to tie or lose to sleep powder, leech seed or missing timing MC.
Heracross (if you get crit you faint) or you can faint in 2 hits with 2 medium high rolls.
Other fighters which you outspeed (guaranteed tie with dive 3x) or win if you dodge a cross chop OR get a crit with dive.
QC hits (BEFORE you dive). Messes up your dodge strategy a bit.
Strong Normals like CB granbull, Ursaring.

Water Absorb Users- awkward matchups
Lapras
  • Icy Wind turn 1, in case this is DD. If it is, congrats. You get a free ko.
  • If it reveals T-bolt, MC for the damage
  • If it shows Sheercold, dive to dodge (if you didn't get hit)
Quagsire
  • You win outright or tie at the worst case scenario. Spam icy wind, maybe recover or dive turn 3
Vaporeon
  • Ironically as tough as Lapras, because it has acid armor and you have to risk MC somewhere. If it's body slam/shadowball, you kinda lose Milotic completely.
Para, confusion hax against electrics like Lanturn, Ampharos (which can prevent you from getting off mirror coat)
Unfortunate status conditions like freeze from tri attack etc.

Everything else, assuming decent luck, Milotic wins. Seriously, it's that good. It's a good member to go ahead of Heracross too, as it destroys MOST physical stuff, most fliers, never loses outright to fire types without attract or confuse ray luck and checks psychic types.

I needed to have a 3rd member in the middle that dealt with the things that Milotic can lose to. I tried Medicham, but it made my matchup against grass types and electric types far worse and didn't help losing to status conditions. Enter: bulky heracross. I tried a lot of sets and this is probably one of the few ways it's not outclassed by Medicham. Medicham does any endure strat better. Heracross with a somewhat fast substitute can block status conditions, gain an advantage, activate swarm and salac somewhat reliably.

View attachment 396274
Heracross @ Salac Berry
Level: 100
Adamant Nature (+ Atk, -SpA)
Ability: Swarm
EVs: 160 HP / 100 Atk / 248 Spe
IVs: 30 SpD / 30 Spe
- Hidden Power Bug
- Earthquake
- Brick Break
- Substitute

Set features
  • Lives a Starmie's psychic
  • Lives a fast 170 SpA Latios Psychic most of the time
  • Avoids a 2ko from Regirock
  • Enough speed to out-speed the Aids Starmie set with t-wave and confuse ray and most gardevoir/Latios sets.
  • Dumped the rest in attack.
Heracross is the glue of the team. It can revenge ko Walrein, Regice, Lapras, normal types and grass types that Milotic can lose to. It can sub on grass types and t-wave and gain a huge advantage. There are times I wish I had endure or rockslide, but I decided that I would keep Heracross's role simple and play it safe. Megahorn would make Heracross far more dangerous, but it's hard to risk because of the miss chance.

170 SpA Latios Psychic vs. 160 HP / 0 SpD Heracross: 292-344 (85.6 - 100.8%) -- 6.3% chance to OHKO
252+ SpA Starmie Psychic vs. 160 HP / 0 SpD Heracross: 285-336 (83.5 - 98.5%) -- guaranteed 2HKO
252+ Atk Regirock Rock Slide vs. 160 HP / 0 Def Heracross: 143-169 (41.9 - 49.5%) -- guaranteed 3HKO

1641129465323.png


ENDIKU (Latios) @ Lum Berry
Ability: Levitate
EVs: 40 HP / 252 SpA / 216 Spe
Timid Nature (+Spe, -Atk)
- Calm Mind
- Psychic
- Dragon Claw
- Thunderbolt


Not much has changed; this doesn't need an overly complicated EV distribution anymore, since all SpA matters and the "free level 100 bulk" ensures it already lives through Ursaring Double-Edge and stuff like Glalie's STAB unboosted. Ice Beam was replaced by Dragon Claw to have a better matchup against most Psychics (e.g. Hypno/Gardevoir/Espeon), fellow Lati twins, and Houndoom. At +1 it does most things Ice Beam would do as well, except against some bulkier Ground-types. The Steelix matchup got a LOT worse, sadly, but most of the time you still power through it. "Cited from Actaeon's post above.

Set Explanation

Latios is a very safe third pokemon and can clutch out wins in some dire situations. I usually run ice beam, but I found dragon claw was a necessity for psychic types and Houndoom. Without dragon-claw, there's a very good chance you just lose to these, which are too dangerous.

Overly Comprehensive List of Difficult matchups​

Bird Keepers and hex trainers can be dangerous because of
  • Fast Crobat
  • Zapdos- T-wave shenanigans, sub or t-bolt crit
  • Articuno against Latios
  • Status hax from tri attack
Team is weak to QC physicals if Milotic isn't around.
Psychic types and Houndoom are tough matchups if Milotic isn't there. Riskier for sure.
Electric trainers with status conditions and Jolteon.
This team can't hit shedinja. ROFL. Milotic ties or wins depending on whether it actually uses an attacking move or not.

If Milotic wins, usually I'm in a good spot.

I've been grinding for 3+ weeks trying to reach 200 wins. I'm super gutted only reaching 148. I actually got to 148 on my second attempt. Lost to Bite Espeon critting Latios (which forced me to run dragon claw. The next best attempt reached 125, which I will attach a replay below. It's super bullsh+t. Lost AGAIN to that Aerial Ace Registeel.

Rage inducing replay. I just had to avoid a crit there to win. Lost Body by like 1 or 2%.
Wonderful post and you explained very well the thought process of the last two weeks. Just making this extra post to second/emphasize the enormous amount of brainstorming and trouble you went through, totally understand the disappointment. Milotic was a huge finding and 148 is nothing to underestimate in the street fights of Arena, congratulations!
 
Reporting an ongoing streak of 252 wins in Level 50 Battle Tower Doubles on cartridge.

270799717_608045417087309_3780408809363670994_n.jpg


381.gif
376.gif
146.gif


When starting a Generation III streak, the mechanic change that stood out to me as ripe for exploitation was the fact that spread moves that affect all other Pokemon on the field receive no damage reduction in this generation. After I struggled to get a solid streak on my explosion-focused team, I turned to Earthquake. The first easy choice was to slap a choice band on Swampert - the strongest EQ user available. From there, I initially tried playing with Zapdos opening alongside Swampert and Registeel and Latios in the rear. I eventually rotated Latios to the front of the roster, swapped Registeel for Metagross and played around with the fourth slot (I tried Weezing and Togetic here) before settling on Moltres.



Latios @ Lum Berry
Ability: Levitate
Level: 50
EVs: 4 HP / 252 SpA / 252 Spe
IVs: 31/x/31/31/31/31
Modest Nature
- Psychic
- Ice Beam
- Thunderbolt
- Protect

I don't think that the rationale behind picking up Latios really needs to be explained at this point. Many have said it before, and I agree - this is the strongest Pokemon in the entirity of the Emerald Frontier. There are probably only twothings worth mentioning about this set.
First is the choice to run Modest max speed instead of Timid with 220/216 EVs. When making this choice, I looked at the Pokemon from 162 to 172 speed and decided that each could be separately managed. Notable threats include: Starmie and Aerodactyl (Timid Latios doesn't OHKO these anyway), Dugtrio (see threat list discussion below), Raikou (Lum Berry blocks Thunder Wave, is OHKOed by Swampert), Sceptile (will always attack Swampert) and Gengar (see threat list discussion). I opted into max speed to speed-tie with most Gengar. I'm happy with that decision; winning the speed tie is very helpful but generally not mandatory. I don't have specific calculations for Modest vs Timid, but from memory it helps me grab some KOs on bulky waters with Thunderbolt + Swampert's Earthquake.
Second is the moveset - notably the absence of Dragon Claw. I initially has Dragon Claw over Ice Beam but noted it was only used against opposing Lati@s (admittedly, now a threat) and some Psychic types (which generally survived anyway). Ice Beam improves a number of key matchups, including Salamence (252+ SpA Latios Dragon Claw vs. 255 HP / 0 SpD Salamence-4: 183-216 (90.5 - 106.9%) -- 43.8% chance to OHKO), Claydol, Zapdos, and Grass types generally. Protect meanwhile has proven too useful. Neutral attacks, Thunder Waves as well as Dark and Ghost type attacks will typically go into Latios. Protect allows Latios to tank one of these hits, and then protect on the second turn while the AI magnets onto the low-HP target. It's also required alongside an Exploding Metagross.

swampert.png

Swampert @ Choice Band
Ability: Torrent
Level: 50
EVs: 164 HP / 252 Atk / 4 SpA / 88 Spe
IVs: 31/31/30/31/30/30
Adamant Nature
- Earthquake
- Hidden Power [Rock]
- Double-Edge
- Ice Beam

An easy choice as the strongest Earthquake user available, made stronger with the addition of a Choice Band. HP Rock is occasionally used to fend off opposing flying types (notably, Moltres, who can cause trouble with Double Team). Double-Edge is most often used as a strong single-target nuke in late game situations where Swampert is swapped out early, and then returns to the field either alongside Metagross, or after Latios and fainted and I don't wish to lock myself into EQ in case Moltres is also taken out. Ice Beam is used on rare occasions against Flygon or Salamence after one or more Intimidate attack drops, and can provide useful chip damage against Cradily. Moves that I have tried previously include Surf (utterly useless) and Brick Break (dropped in favour of the more versatile Double-Edge).

88 EVs puts Swampert at 91 speed, just ahead of the rather packed 90 speed tier. Otherwise he's got max attack, with the rest dumped in HP and 4 EVs in SpA to prevent any going to waste. Some notable threats between 80 (uninvested Swampert) and 90 speed include Scizor, Walrein, Hypno, Metagross, Dewgong and Ludicolo.

metagross.png

Metagross @ Cheri Berry
Ability: Clear Body
Level: 50
EVs: 204 HP / 252 Atk / 52 Spe
IVs: 31/31/31/odd/30/31
Adamant Nature
- Hidden Power [Steel]
- Earthquake
- Shadow Ball
- Explosion

This Metagross is a hangover from my old explosion team. It had the same moveset and EV spread, except with a skew toward physical defence. Moving all remaining EVs to HP provides the best overall bulk, and I haven't made attempts to optimise for specific KOs because Metagross is rarely at full health, as it is brought in on the switch in most instances. 52 Speed EVs reaches 97 Speed, allowing it to outspeed Nidoqueen and Claydol, as well as all the 90 speed threats mentioned above. The only change I would consider here is to potentially drop some Attack EVs to add into bulk, but I don't feel any real need to do so at the moment.

When I brought Metagross over to this team, I initially had Protect over Explosion, in fear of Swampert being locked into EQ. However, I found that the bigger threat was having Moltres KOed and having to bring Metagross in to an EQ in the same turn, in which case Protect was of no use. Explosion has gotten me out of a few tricky late game situations in the final 2v2, such as Moltres & Metagross vs two bulky waters.

moltres.png

Moltres @ Leftovers
Ability: Pressure
Level: 50
EVs: 252 HP / 42 Def / 40 SpA / 76 SpD / 100 Spe
IVs: 31/even/31/30/31/31/31
Modest Nature
- Flamethrower
- Hidden Power [Grass]
- Toxic
- Protect

Moltres is the last member of the team. In all honesty, it doesn't have much chance to shine here, and (surprisingly for a Fire/Flying type) is used primarily as a defensive pivot. It helpfully sports a 4x resistance to both Grass and Bug type moves, and then entices Rock type moves out of enemies which, with Protect plus a switch back into Swampert or Metagross, can often buy two free turns for allies to throw out damage. Moltres was chosen over Charizard for its superior bulk, which allows it to generally switch into Dark and Ghost type attacks levied at Latios for minimal damage (e.g. 255+ SpA Houndoom-2 Crunch vs. 252 HP / 0 SpD Moltres: 77-91 (39 - 46.1%) -- guaranteed 3HKO after Leftovers recovery; 255+ Atk Gengar-2 Shadow Ball vs. 252 HP / 42 Def Moltres: 51-61 (25.8 - 30.9%) -- 6.7% chance to 4HKO after Leftovers recovery).

The moveset serves two main purposes: to switch in on Grass types and hit back with Flamethrower, and to Toxic stall Double Team users. HP Grass meanwhile helps deal with any Rock/Ground types, as well as provide useful chip damage to bulky waters. Protect is crucial for buying time, and the AI targetting can get very predictable when you account for the defensive synergy between Moltres and it's allies.

40 SpA Modest Moltres gets the first bonus point, and gives a 87.5% chance to OHKO defensive Golem sets. 100 Speed outspeeds Breloom by one point, as well as being a few points ahead of Metagross. The rest of the EVs are divided to give all-round even bulk.

While I'm very happy with the use of Moltres in this slot, I am very much still considering changes to both the spread and moveset. I should run calculations to see if there is a more helpful SpA stat I should be aiming for. I've also considered dropping the Modest and SpA investment entirely in favour of more bulk, but would need to consider any KOs I might lose in doing so. The choice to go Modest was more in favour of general stat efficiecny, and not any specific purpose. Toxic is not all that useful (although occasionally very important), while HP grass is generally underwhelming, although I'm not in the mood for shooting for a new Moltres.

  • OHKO users
    112.gif
    - Quick claw OHKO moves (or, in the case of Dugtrio, just very fast) are difficult for this team. OHKO users that only have Fissure (e.g. Donphan and Dugtrio) can usually be dealt with by swapping Swampert for Moltres and targetting them with Latios. However, this luxury doesn't apply to anything with Sheer Cold or Horn Drill. It helps that the none of the water type OHKO users with Quick Claw run a water move other than Surf, which Moltres can generally bait out and shrug off. Quick Claw + Explosion users (e.g. Golem-3 and 4) are a similar threat that I lump in the same boat as OHKO users that can cause trouble, although can be worked around by always taking them out in a single turn to minimise the chance of them exploding.
  • Levitating Psychics
    (with Trace) - These are probably the biggest non-RNG related threat to the team. They are immune to Earthquake, can not be OHKOed (or in the case of Gardevoir, even 2HKOed) by Latios and can each carry a number of moves threatening to the team. Lati@s are generally dealt with by sacrificing Latios and picking up the KO with Ice Beam + either Double-Edge from Swampert or Shadow Ball from Metagross. Claydol sinks to two Ice Beams. Gardevoir is the biggest threat as it can also carry Magical Leaf. Swampert will always survive this (255+ SpA Gardevoir Magical Leaf vs. 164 HP / 0 SpD Swampert: 163-192 (83.1 - 97.9%) -- guaranteed 2HKO) but is vulnerable to a critical hit. This generally necessitates a switch to Metagross, who can pick up the KO with Shadow Ball after Latios does approx. 30% with Thunderbolt.
  • Double Team + Rest/Toxic Immunity
    and more... - These can't be Toxic stalled. Thankfully, most of themare vulnerable to an OHKO from at least one team member, so not too much luck is required. There are some others I could add to this list (e.g. Venusaur, Shiftry, Wailord) but they can be outsped and KOed turn one before they start spamming Double Team, so not a big threat.
  • Fast Status
    (excluding Thunder Wave) - Fast confusion, sleep, or attract directed at Swampert can hurt the team. Crobat goes down to a single Psychic, and Latios will never be confused because of Lum Berry. Espeon should be a problem in theory, but prefers to use Attract or Bite on Latios, leaving it open to an Earthquake from Swampert. If I had put more forethought into this issue, I would have made Swampert female to make this more predictable. Taurus goes down to 2 Psychics or Earthquakes or one of each. As it takes 3 turns to confuse both Latios and Swampert, I can always get off two attacks before being totally crippled, even if Latios loses the speed tie. Gengar is perhaps the most threatening of the bunch. 95%+ of the time it is fine - it will go down to a Psychic and even if I lose the speed tie and Swampert is confused or asleep. The problem is that Gengar also carries Shadow Ball which will KO Latios with a crit, so this is always on the table. This can put me in a tough position. Metagross will OHKO (252+ Atk Metagross Shadow Ball vs. 0 HP / 0 Def Gengar-2: 156-184 (115.5 - 136.2%) -- guaranteed OHKO) but is susceptible to burn or confusion, and is not a nice switch if Swampert is locked into Earthquake. Moltres can 2HKO with Flamethrower (40+ SpA Moltres Flamethrower vs. 0 HP / 0 SpD Gengar-2: 94-111 (69.6 - 82.2%) -- guaranteed 2HKO after Leftovers recovery) but that gives Gengar two chances to confuse it. I've been put in this situation twice, and managed to fight through confusion eventually but it can become very uncomfortable.
  • Ludicolo
    - This thing should be a huge threat, but for some reason hasn't caused issues. It has access to Double Team, can hit any member of the team for super effective damage, can survive any single hit (other than Explosion, which is too risky to use after a Double Team) and even survives Psychic + Earthquake. Thankfully, the only set to carry a Grass attack is Ludicolo-3, which will almost always go for Rain Dance or Double Team on turn one, an can not OHKO Swampert in any event (0 SpA Ludicolo-3 Giga Drain vs. 164 HP / 0 SpD Swampert: 142-168 (72.4 - 85.7%) -- guaranteed 2HKO). My current strategy is to generally ignore it, with the aim of picking up the KO on turn two or three, or landing a Toxic in a late game Double Team scenario. I fear this could bite me with some untimely status side effects or missed attacks, but has worked for me so far.

Potential improvements:
  • Rejigging the EV spreads on each of Moltres and Metagross. There is room here for optimisation and calculations;
  • Replacing Metagross with Thick Fat Snorlax (likely with the same moveset, but dropping HP Steel for Return or Double Edge); or
  • Swapping the Lum and Cheri Berry, as Metagross is more likely to switch in on ice moves that may freeze.
I'll continue to push forward with this team until a loss, but based on the vulnerability to certain threat (particularly QC+OHKO and Double Team as above) I think this team doesn't have the necessary tools to keep RNG strategies in check. It is inevitable that it will fall eventually, but I hope to see how far I can get it before that happens.
 
Last edited:
Reporting an ongoing streak of 252 wins in Level 50 Battle Tower Doubles on cartridge.

View attachment 396594

View attachment 396543
View attachment 396546View attachment 396548

When starting a Generation III streak, the mechanic change that stood out to me as ripe for exploitation was the fact that spread moves that affect all other Pokemon on the field receive no damage reduction in this generation. After I struggled to get a solid streak on my explosion-focused team, I turned to Earthquake. The first easy choice was to slap a choice band on Swampert - the strongest EQ user available. From there, I initially tried playing with Zapdos opening alongside Swampert and Registeel and Latios in the rear. I eventually rotated Latios to the front of the roster, swapped Registeel for Metagross and played around with the fourth slot (I tried Weezing and Togetic here) before settling on Moltres.



Latios @ Lum Berry
Ability: Levitate
Level: 50
EVs: 4 HP / 252 SpA / 252 Spe
IVs: 31/x/31/31/31/31
Modest Nature
- Psychic
- Ice Beam
- Thunderbolt
- Protect

I don't think that the rationale behind picking up Latios really needs to be explained at this point. Many have said it before, and I agree - this is the strongest Pokemon in the entirity of the Emerald Frontier. There are probably only twothings worth mentioning about this set.
First is the choice to run Modest max speed instead of Timid with 220/216 EVs. When making this choice, I looked at the Pokemon from 162 to 172 speed and decided that each could be separately managed. Notable threats include: Starmie and Aerodactyl (Timid Latios doesn't OHKO these anyway), Dugtrio (see threat list discussion below), Raikou (Lum Berry blocks Thunder Wave, is OHKOed by Swampert), Sceptile (will always attack Swampert) and Gengar (see threat list discussion). I opted into max speed to speed-tie with most Gengar. I'm happy with that decision; winning the speed tie is very helpful but generally not mandatory. I don't have specific calculations for Modest vs Timid, but from memory it helps me grab some KOs on bulky waters with Thunderbolt + Swampert's Earthquake.
Second is the moveset - notably the absence of Dragon Claw. I initially has Dragon Claw over Ice Beam but noted it was only used against opposing Lati@s (admittedly, now a threat) and some Psychic types (which generally survived anyway). Ice Beam improves a number of key matchups, including Salamence (252+ SpA Latios Dragon Claw vs. 255 HP / 0 SpD Salamence-4: 183-216 (90.5 - 106.9%) -- 43.8% chance to OHKO), Claydol, Zapdos, and Grass types generally. Protect meanwhile has proven too useful. Neutral attacks, Thunder Waves as well as Dark and Ghost type attacks will typically go into Latios. Protect allows Latios to tank one of these hits, and then protect on the second turn while the AI magnets onto the low-HP target. It's also required alongside an Exploding Metagross.

View attachment 396545
Swampert @ Choice Band
Ability: Torrent
Level: 50
EVs: 164 HP / 252 Atk / 4 SpA / 88 Spe
IVs: 31/31/30/31/30/30
Adamant Nature
- Earthquake
- Hidden Power [Rock]
- Double-Edge
- Ice Beam

An easy choice as the strongest Earthquake user available, made stronger with the addition of a Choice Band. HP Rock is occasionally used to fend off opposing flying types (notably, Moltres, who can cause trouble with Double Team). Double-Edge is most often used as a strong single-target nuke in late game situations where Swampert is swapped out early, and then returns to the field either alongside Metagross, or after Latios and fainted and I don't wish to lock myself into EQ in case Moltres is also taken out. Ice Beam is used on rare occasions against Flygon or Salamence after one or more Intimidate attack drops, and can provide useful chip damage against Cradily. Moves that I have tried previously include Surf (utterly useless) and Brick Break (dropped in favour of the more versatile Double-Edge).

88 EVs puts Swampert at 91 speed, just ahead of the rather packed 90 speed tier. Otherwise he's got max attack, with the rest dumped in HP and 4 EVs in SpA to prevent any going to waste. Some notable threats between 80 (uninvested Swampert) and 90 speed include Scizor, Walrein, Hypno, Metagross, Dewgong and Ludicolo.

View attachment 396547
Metagross @ Cheri Berry
Ability: Clear Body
Level: 50
EVs: 204 HP / 252 Atk / 52 Spe
IVs: 31/31/31/odd/30/31
Adamant Nature
- Hidden Power [Steel]
- Earthquake
- Shadow Ball
- Explosion

This Metagross is a hangover from my old explosion team. It had the same moveset and EV spread, except with a skew toward physical defence. Moving all remaining EVs to HP provides the best overall bulk, and I haven't made attempts to optimise for specific KOs because Metagross is rarely at full health, as it is brought in on the switch in most instances. 52 Speed EVs reaches 97 Speed, allowing it to outspeed Nidoqueen and Claydol, as well as all the 90 speed threats mentioned above. The only change I would consider here is to potentially drop some Attack EVs to add into bulk, but I don't feel any real need to do so at the moment.

When I brought Metagross over to this team, I initially had Protect over Explosion, in fear of Swampert being locked into EQ. However, I found that the bigger threat was having Moltres KOed and having to bring Metagross in to an EQ in the same turn, in which case Protect was of no use. Explosion has gotten me out of a few tricky late game situations in the final 2v2, such as Moltres & Metagross vs two bulky waters.

View attachment 396549
Moltres @ Leftovers
Ability: Pressure
Level: 50
EVs: 252 HP / 42 Def / 40 SpA / 76 SpD / 100 Spe
IVs: 31/even/31/30/31/31/31
Modest Nature
- Flamethrower
- Hidden Power [Grass]
- Toxic
- Protect

Moltres is the last member of the team. In all honesty, it doesn't have much chance to shine here, and (surprisingly for a Fire/Flying type) is used primarily as a defensive pivot. It helpfully sports a 4x resistance to both Grass and Bug type moves, and then entices Rock type moves out of enemies which, with Protect plus a switch back into Swampert or Metagross, can often buy two free turns for allies to throw out damage. Moltres was chosen over Charizard for its superior bulk, which allows it to generally switch into Dark and Ghost type attacks levied at Latios for minimal damage (e.g. 255+ SpA Houndoom-2 Crunch vs. 252 HP / 0 SpD Moltres: 77-91 (39 - 46.1%) -- guaranteed 3HKO after Leftovers recovery; 255+ Atk Gengar-2 Shadow Ball vs. 252 HP / 42 Def Moltres: 51-61 (25.8 - 30.9%) -- 6.7% chance to 4HKO after Leftovers recovery).

The moveset serves two main purposes: to switch in on Grass types and hit back with Flamethrower, and to Toxic stall Double Team users. HP Grass meanwhile helps deal with any Rock/Ground types, as well as provide useful chip damage to bulky waters. Protect is crucial for buying time, and the AI targetting can get very predictable when you account for the defensive synergy between Moltres and it's allies.

40 SpA Modest Moltres gets the first bonus point, and gives a 87.5% chance to OHKO defensive Golem sets. 100 Speed outspeeds Breloom by one point, as well as being a few points ahead of Metagross. The rest of the EVs are divided to give all-round even bulk.

While I'm very happy with the use of Moltres in this slot, I am very much still considering changes to both the spread and moveset. I should run calculations to see if there is a more helpful SpA stat I should be aiming for. I've also considered dropping the Modest and SpA investment entirely in favour of more bulk, but would need to consider any KOs I might lose in doing so. The choice to go Modest was more in favour of general stat efficiecny, and not any specific purpose. Toxic is not all that useful (although occasionally very important), while HP grass is generally underwhelming, although I'm not in the mood for shooting for a new Moltres.

  • OHKO users View attachment 396572
    - Quick claw OHKO moves (or, in the case of Dugtrio, just very fast) are difficult for this team. OHKO users that only have Fissure (e.g. Donphan and Dugtrio) can usually be dealt with by swapping Swampert for Moltres and targetting them with Latios. However, this luxury doesn't apply to anything with Sheer Cold or Horn Drill. It helps that the none of the water type OHKO users with Quick Claw run a water move other than Surf, which Moltres can generally bait out and shrug off. Quick Claw + Explosion users (e.g. Golem-3 and 4) are a similar threat that I lump in the same boat as OHKO users that can cause trouble, although can be worked around by always taking them out in a single turn to minimise the chance of them exploding.
  • Levitating Psychics
    (with Trace) - These are probably the biggest non-RNG related threat to the team. They are immune to Earthquake, can not be OHKOed (or in the case of Gardevoir, even 2HKOed) by Latios and can each carry a number of moves threatening to the team. Lati@s are generally dealt with by sacrificing Latios and picking up the KO with Ice Beam + either Double-Edge from Swampert or Shadow Ball from Metagross. Claydol sinks to two Ice Beams. Gardevoir is the biggest threat as it can also carry Magical Leaf. Swampert will always survive this (255+ SpA Gardevoir Magical Leaf vs. 164 HP / 0 SpD Swampert: 163-192 (83.1 - 97.9%) -- guaranteed 2HKO) but is vulnerable to a critical hit. This generally necessitates a switch to Metagross, who can pick up the KO with Shadow Ball after Latios does approx. 30% with Thunderbolt.
  • Double Team + Rest/Toxic Immunity
    and more... - These can't be Toxic stalled. Thankfully, most of themare vulnerable to an OHKO from at least one team member, so not too much luck is required. There are some others I could add to this list (e.g. Venusaur, Shiftry, Wailord) but they can be outsped and KOed turn one before they start spamming Double Team, so not a big threat.
  • Fast Status
    (excluding Thunder Wave) - Fast confusion, sleep, or attract directed at Swampert can hurt the team. Crobat goes down to a single Psychic, and Latios will never be confused because of Lum Berry. Espeon should be a problem in theory, but prefers to use Attract or Bite on Latios, leaving it open to an Earthquake from Swampert. If I had put more forethought into this issue, I would have made Swampert female to make this more predictable. Taurus goes down to 2 Psychics or Earthquakes or one of each. As it takes 3 turns to confuse both Latios and Swampert, I can always get off two attacks before being totally crippled, even if Latios loses the speed tie. Gengar is perhaps the most threatening of the bunch. 95%+ of the time it is fine - it will go down to a Psychic and even if I lose the speed tie and Swampert is confused or asleep. The problem is that Gengar also carries Shadow Ball which will KO Latios with a crit, so this is always on the table. This can put me in a tough position. Metagross will OHKO (252+ Atk Metagross Shadow Ball vs. 0 HP / 0 Def Gengar-2: 156-184 (115.5 - 136.2%) -- guaranteed OHKO) but is susceptible to burn or confusion, and is not a nice switch if Swampert is locked into Earthquake. Moltres can 2HKO with Flamethrower (40+ SpA Moltres Flamethrower vs. 0 HP / 0 SpD Gengar-2: 94-111 (69.6 - 82.2%) -- guaranteed 2HKO after Leftovers recovery) but that gives Gengar two chances to confuse it. I've been put in this situation twice, and managed to fight through confusion eventually but it can become very uncomfortable.
  • Ludicolo
    - This thing should be a huge threat, but for some reason hasn't caused issues. It has access to Double Team, can hit any member of the team for super effective damage, can survive any single hit (other than Explosion, which is too risky to use after a Double Team) and even survives Psychic + Earthquake. Thankfully, the only set to carry a Grass attack is Ludicolo-3, which will almost always go for Rain Dance or Double Team on turn one, an can not OHKO Swampert in any event (0 SpA Ludicolo-3 Giga Drain vs. 164 HP / 0 SpD Swampert: 142-168 (72.4 - 85.7%) -- guaranteed 2HKO). My current strategy is to generally ignore it, with the aim of picking up the KO on turn two or three, or landing a Toxic in a late game Double Team scenario. I fear this could bite me with some untimely status side effects or missed attacks, but has worked for me so far.

Potential improvements:
  • Rejigging the EV spreads on each of Moltres and Metagross. There is room here for optimisation and calculations;
  • Replacing Metagross with Thick Fat Snorlax (likely with the same moveset, but dropping HP Steel for Return or Double Edge); or
  • Swapping the Lum and Cheri Berry, as Metagross is more likely to switch in on ice moves that may freeze.
I'll continue to push forward with this team until a loss, but based on the vulnerability to certain threat (particularly QC+OHKO and Double Team as above) I think this team doesn't have the necessary tools to keep RNG strategies in check. It is inevitable that it will fall eventually, but I hope to see how far I can get it before that happens.
Congrats on the streak and claiming the #1 spot on Tower Doubles, for now... Alternatively, I also have some suggestions on your team so you can improve on your streak! I see you listed Ice Beam on Swampert with the sole purpose of hitting Cradily, but if that's what this move is worth for, why not using Brick Break instead? Brick Break hits Cradily much more harder while having the advantage of hitting Umbreon and Blissey without resorting to Earthquake in case that Latios or Moltres are not around. You hardly will need Ice coverage on Swampert anyways since Latios runs the move and will easily out-damage whatever Swampert can do with Ice Beam. Additionally, you should still keep Metagross on your team IMO. It's your only switch-in to Gardevoir 2 and Espeon which IMO Snorlax worsens the match-up because otherwise, you're taking neutral damage from Psychic where even a crit or a Sp. Def drop will make you lose momentum.

As far as optimization goes, you should consider running 164 Speed EVs on Metagross which lets you outspeed a dangerous threat against your team: Mr. Mime 4. I'm sure you fear this match-up since Mr. Mime 1 can run Magical Leaf but Set 4 has super-effective coverage on Latios and Moltres while complicating the match up if Bright Powder activates a few times. The Speed EVs allows you to outspeed and smack it with Shadow Ball before it hits you with Fire Punch. It will look something like this: 94 HP / 252 Atk / 164 Speed. It still retains a decent amount of bulk while hitting as hard as possible and being fast enough to outspeed targets like Gyarados and Milotic.
 
I am still in the process of reaching 200+ wins in Arena. I feel very close to the 200 holy grail and I'm still optimizing some teams.
I came really close with 186 wins with Choice Band Metagross, Milotic and Latios. Lost to Starmie, Raikou and Metagross. Metagross Exploded on Starmie, Raikou para haxed Milotic on mirror coat and Latios couldn't beat Metagross.

This team on average hit 100+ wins very consistently for me, the sign of a really good team. I chose to play this team at level 50, as Metagross is consistently less of a threat compared to Tyranitar and damage ranges are better for Metagross and Latios.

Level 50 252+ Atk Choice Band Metagross Earthquake vs. Level 50 0 HP / 0 Def Metagross: 156-184 (100.6 - 118.7%) -- guaranteed OHKO
Level 100 252+ Atk Choice Band Metagross Earthquake vs. Level 100 0 HP / 0 Def Metagross: 295-348 (98 - 115.6%) -- 87.5% chance to OHKO
1642275364311.png


Metagross @ Choice Band
Level: 50
Adamant Nature
Ability: Clear Body
EVs: 252 Atk / 12 Def / 244 Spe

- Meteor Mash
- Explosion
- Earthquake
- Aerial Ace

Set features
  • Faster than the 120 speed tier
  • Maximum Attack
  • 12 defense to live earthquakes from Golem and Rhydon with greater probability.
  • Rhydon’s EQ = 75% to KO, Golem’s EQ = 12.5%
Actaeon had the idea to try out Choice Band Metagross as the lead. We’ve tried choice band in other ways, but it’s perfect in the lead slot. With solid natural bulk, good resistances, clear body ability to block intimidate and a decent speed stat, Metagross can wreck some havoc off the rip. The power boost from Choice Band is appreciated, allowing you to ko pokemon like Lanturn in one hit and even 3 hit ko bulky waters. The idea here is for Metagross to snowball out of control or trade in bad matchups. Fire types and Ground types that can ko Metagross in one hit are easily checked by Milotic.

I actually didn’t get into too many troublesome spots being locked into a bad move. The worst instance was facing Raikou/Articuno. I locked myself into EQ and Articuno proceeded to sub, double team and reverse sweep my team…. ROFL.

In general, you do want to be locked into Meteor Mash (HP steel isn’t worth it), as it has no immunities. Against most pokemon, you can afford to take that risk, as they cannot ko metagross in one hit anyway without a critical hit or ohko move.

Side note: it's generally really safe to lock yourself into Aerial Ace against parasol trainers and bug catchers. There's usually a lot of grass and bug pokemon in the back.

As a side note, I had shadowball instead of Explosion, but Explosion is absolutely mandatory. I’ve been in too many situations where I needed a troublesome pokemon gone.

The list of explosion targets include
  • Snorlax
  • Lapras
  • Zapdos
  • Suicune
  • Starmie (no shadowball). Might as well get this.
  • Gyarados. Can’t beat it anyway. Don’t want it to set up DD.
1642275308729.png

Milotic @ Leftovers
Level: 50
Calm Nature (+ SpD, -Atk)
Ability: Marvel Scale
EVs: 252 HP / 216 Def / 36 SpD / 4 Spe

- Dive
- Recover
- Icy-wind
- Mirror Coat

Features of the set (adapted to level 50)
  • Avoids the thunderbolt 2ko from Modest Latios
    • 252+ SpA Latios Thunderbolt vs. 252 HP / 36+ SpD Milotic: 88-104 (43.5 - 51.4%)
  • Rest in physical defense
  • Outspeeds base 100s after a speed drop
  • Outspeeds Magmar 1 and opposing Milotic
This Metagross team also showed me how effective Milotic is in the 2nd slot compared to the 1st slot. If Metagross gets revenge koed, Milotic gets incredibly valuable information that eases prediction. Gengar uses fire punch against Metagross? Obvious MC.

After playing Milotic close to thousands of battles by now, I’ve actually learned a lot about the optimal plays against troublesome opponents.
  • Vaporeon – Aim for Mind and Body if it acid armors. Aim for skill and body if Vaporeon uses Surf against Milotic. Borderline unwinnable against Body Slam Vaporeon (unless it paralyzes you- in which case Vaporeon will switch to surf- which you can MC to win).
  • Regice- Dive 3x. Take the guaranteed tie by spamming dive 3x. You automatically win if Regice doesn't have thunderbolt/thunder. If it uses Amnesia or rain dance on turn 2, you win this too.
  • QC ursaring/ QC granbull (if it's Double edge)- Dive, Recover, Recover. If you dodge a crit, you automatically win Skill and Body due to double edge recoil.
  • Opposing Milotic. Take the W against non-leftovers variants with Dive+ recover. Take the tie against leftovers variants.
  • Dive first against Lanturn. There’s a very good chance Lanturn will use Confuse ray, Rain dance or thunderwave. You also live 2 thunderbolts and Latios can revenge kill in the worst situation.
  • 252+ SpA Lanturn Thunderbolt vs. 252 HP / 36+ SpD Milotic: 93-110 (46 - 54.4%) -- 6.6% chance to 2HKO after Leftovers recovery
I also learned that Milotic is also an amazing partner for Latios. It completely handles physical pokemon like Regirock and Metagross (guaranteed tie at worst), beats Houndoom, Arcanine, beats/ties with Regice. It also beats and/or ties against double team spammers like Umbreon, Blissey. It doesn’t help set up Latios in the way Snorlax does, but these troublesome pokemon won’t even make it past Milotic.

Milotic also heavily prefers level 50 even with less overall bulk. Opposing Metagross is far less dangerous than Tyranitar (who has many haxy sets and boosted rockslides to deal with). Tyranitar also summons sand and is a brick wall for Latios to deal with, which is a problem in a 1-1 match.

1642275353731.png

ENKIDU (Latios) @ Lum Berry
Ability: Levitate
EVs: 28 HP / 12 Def / 244 SpA / 4 SpD / 220 Spe
Timid Nature (+Spe, -Atk)
- Calm Mind
- Psychic
- Dragon Claw
- Thunderbolt


Set Explanation
I’m not really going to into too much explanation with Latios. It checks grass types, electric types and it’s clearly the best special attacker in the game. Dragon claw is necessary against psychics and houndoom.

Troublesome opponents
Jolly Metagross. Lives EQ from CB Meta and explodes on Milotic.
- 252+ Atk Choice Band Metagross Earthquake vs. 170 HP / 0 Def Metagross: 156-184 (88.6 - 104.5%) -- 31.3% chance to OHKO
Heracross. Metagross can get koed from an EQ critical hit. Aerial Ace target.
Snorlax
Water Absorb QC Lapras.
Haxrein
Crit/miss against psychic types/Gengar. Ex Ice punch Gardevoir.
Crit/Para Hax from electrics (Raikou, Jolteon 4, Zapdos).
Legendary trainers seem to carry some combination of Regice/Zapdos/ Registeel/Articuno/ Raikou and Meta. Always dangerous. Need a good order as well.
Slightly shaky matchup against QC normal types like Granbull and Ursaring.
Status hax (confusion and para)

https://pokepast.es/1a0942889618131b
 

Attachments

Last edited:
Update: I have finally achieved 200+ wins! My streak has ended @207 with Gengar, Milotic and Latios. Milotic tied against Meganium who set up lightscreen and Latios couldn't beat shadow ball Exploud of all things with lightscreen up. I'm not even sad about it. I've lost close to 200-300 times in the Arena and I finally got the luck I was looking for. Too many heart breaking losses to Haxrein, Lapras and other hax to count. I'm only going to explain the different Gengar set, as I've already explained the Milotic set with a wealth of detail.

1642375534704.png

Gengar (F) @ Cheri Berry
Ability: Levitate
EVs: 84 Def / 204 SpA / 220 Spe
Timid Nature
- Thunderbolt
- Ice Punch
- Destiny Bond
- Substitute

You might notice that Gengar's evs are different from the 36 Def/ 252 SpA/ 220 Speed that I was always using. My reasoning for increasing physical defense is quite simple: Snorlax. During streaks, I kept getting screwed over by QC Snorlax (lost close to 5 times in a span of 2 days to it). I reduced Gengar's special attack to a reasonable number that risked losing some damage rolls. I settled on 204 SpA, which guarantees the ko against 252 HP Salamence.

252+ Atk Snorlax Shadow Ball vs. 0 HP / 36 Def Gengar: 127-150 (94 - 111.1%) -- 68.8% chance to OHKO. Far too risky.
252+ Atk Snorlax Shadow Ball vs. 0 HP / 84 Def Gengar: 119-140 (88.1 - 103.7%) -- 25% chance to OHKO
204 SpA Gengar Ice Punch vs. 252 HP / 0 SpD Salamence: 204-240 (100.9 - 118.8%) -- guaranteed OHKO

I have also experimented with many items for Gengar (petaya, lum, persim, magnet and cheri). I will go into my explanation why other items are not nearly as ideal for Gengar as cheri berry is.

-Lum Berry:
  • Too valuable of an item to use on Gengar. Gengar is already immune to poison, you will be using substitute against sleep powder grass types and jynx first and you already lose to prime confuse ray targets like Crobat without Destiny Bond. If no one else is using Lum, then Gengar may have it.
- Petaya Berry
  • Gengar should ko targets outright with a SE move or trade against dangerous targets. It's usually better to take the safe route, rather than try to take a greedy damage roll
- Magnet
  • Increases some damage rolls (like making 0 hp starmie a guaranteed ko) and increasing some 2koes against walrein and Lapras. However, against starmie, you shouldn't be risking the T-bolt against Psychic Starmie (could be the 255 hp one or could be Starmie 7, which carries bright powder). Against Walrein 4 and Lapras 2, if you see surf/ice beam, I'm not risking the 2ko, I take the trade after using substitute. No questions about it.
- Persim Berry
  • Very niche.
  • Given Gengar's high speed, I use substitute fairly liberally to scout and you tend to trade or don't win those matchups anyways.
-Cheri Berry.
  • This item is invaluable because Gengar loves it's speed.
  • Incidental paralysis from Zap Cannon Forretress, Thunder Nidoking and dragon breath Steelix is also a thing among others.
  • It eases prediction against known thunder wave spammers like Latios/Latias. You can safely destiny bond to catch every Latios/Latias. If either of the Lati twins reveals T-wave, you can substitute on the next turn and ice punch on turn 3. Otherwise, destiny bond is successful = trade.
  • Same applies for Zapdos, Raikou and misdreavus. I've noticed that against Zapdos though, you want to attack first, as it can double team or lightscreen. Zapdos also seems to switch to thunderbolt on turn 2 if thunder-wave fails.
  • Registeel. You HAVE TO ATTACK Registeel variants, as they are too bulky and you will lose if you don't tie MIND at least. I've painfully experienced losses to Registeel 6 because I didn't attack 3x in a row. However, Registeel 4 has thunder-wave. Having Cheri Berry here eases this prediction, on the off chance you are facing Registeel 4.
  • Regice. Destiny Bond first. If it reveals t-wave, you can safely set up a substitute on turn 2. Same principle applies.
I've also made it no secret how much I appreciate Gengar. It's truly a S tier in the Arena and it also does help Milotic quite a bit. Gengar loves setting up a substitute against status spammers, provides additional ohko move protection, checks normal types and fighting types and trades against troublesome matchups for Milotic. These are all known things that Milotic can struggle with and Gengar patches up these issues quite nicely.

The only thing it doesn't really solve is double team spammers (no Aerial Ace, weakness to DT faint attack umbreon) and Snorlax.

1642375598069.png

Milotic (F)@ Leftovers
Level: 50
Calm Nature (+ SpD, -Atk)
Ability: Marvel Scale
EVs: 252 HP / 216 Def / 36 SpD / 4 Spe

- Dive
- Recover
- Icy-wind
- Mirror Coat

Enkidu (Latios) (M) @ Lum Berry
Level : 50
Ability: Levitate
EVs: 28 HP / 12 Def / 244 SpA / 4 SpD / 220 Spe
Timid Nature (+Spe, -Atk)
- Calm Mind
- Dragon Claw
- Thunderbolt
- Psychic

https://pokepast.es/5cde506c0d83a2c2

Final Thoughts.
I want to thank Actaeon first. You were a big help optimizing sets, experimenting with me and for watching my streams. I am so glad we reached 200+ together. We've spent an absurd amount of time in the Arena (especially in the past month and a half) and we are finally rewarded for our troubles. We finally can move on to niche things and experiment for fun.
I want to thank Kommo-o, dgice, Valentino, Maizup Runeblade14 and anyone else I've missed. I appreciate your support and I'm glad people got to see this team put in the work.
 

Attachments

Last edited:

QuentinQuonce

formerly green_typhlosion
Update: I have finally achieved 200+ wins! My streak has ended @207 with Gengar, Milotic and Latios. Milotic tied against Meganium who set up lightscreen and Latios couldn't beat shadow ball Exploud of all things with lightscreen up. I'm not even sad about it. I've lost close to 200-300 times in the Arena and I finally got the luck I was looking for. Too many heart breaking losses to Haxrein, Lapras and other hax to count. I'm only going to explain the different Gengar set, as I've already explained the Milotic set with a wealth of detail.

View attachment 399037
Gengar (F) @ Cheri Berry
Ability: Levitate
EVs: 84 Def / 204 SpA / 220 Spe
Timid Nature
- Thunderbolt
- Ice Punch
- Destiny Bond
- Substitute

You might notice that Gengar's evs are different from the 36 Def/ 252 SpA/ 220 Speed that I was always using. My reasoning for increasing physical defense is quite simple: Snorlax. During streaks, I kept getting screwed over by QC Snorlax (lost close to 5 times in a span of 2 days to it). I reduced Gengar's special attack to a reasonable number that risked losing some damage rolls. I settled on 204 SpA, which guarantees the ko against 252 HP Salamence.

252+ Atk Snorlax Shadow Ball vs. 0 HP / 36 Def Gengar: 127-150 (94 - 111.1%) -- 68.8% chance to OHKO. Far too risky.
252+ Atk Snorlax Shadow Ball vs. 0 HP / 84 Def Gengar: 119-140 (88.1 - 103.7%) -- 25% chance to OHKO
204 SpA Gengar Ice Punch vs. 252 HP / 0 SpD Salamence: 204-240 (100.9 - 118.8%) -- guaranteed OHKO

I have also experimented with many items for Gengar (petaya, lum, persim, magnet and cheri). I will go into my explanation why other items are not nearly as ideal for Gengar as cheri berry is.

-Lum Berry:
  • Too valuable of an item to use on Gengar. Gengar is already immune to poison, you will be using substitute against sleep powder grass types and jynx first and you already lose to prime confuse ray targets like Crobat without Destiny Bond. If no one else is using Lum, then Gengar may have it.
- Petaya Berry
  • Gengar should ko targets outright with a SE move or trade against dangerous targets. It's usually better to take the safe route, rather than try to take a greedy damage roll
- Magnet
  • Increases some damage rolls (like making 0 hp starmie a guaranteed ko) and increasing some 2koes against walrein and Lapras. However, against starmie, you shouldn't be risking the T-bolt against Psychic Starmie (could be the 255 hp one or could be Starmie 7, which carries bright powder). Against Walrein 4 and Lapras 2, if you see surf/ice beam, I'm not risking the 2ko, I take the trade after using substitute. No questions about it.
- Persim Berry
  • Very niche.
  • Given Gengar's high speed, I use substitute fairly liberally to scout and you tend to trade or don't win those matchups anyways.
-Cheri Berry.
  • This item is invaluable because Gengar loves it's speed.
  • Incidental paralysis from Zap Cannon Forretress, Thunder Nidoking and dragon breath Steelix is also a thing among others.
  • It eases prediction against known thunder wave spammers like Latios/Latias. You can safely destiny bond to catch every Latios/Latias. If either of the Lati twins reveals T-wave, you can substitute on the next turn and ice punch on turn 3. Otherwise, destiny bond is successful = trade.
  • Same applies for Zapdos, Raikou and misdreavus. I've noticed that against Zapdos though, you want to attack first, as it can double team or lightscreen. Zapdos also seems to switch to thunderbolt on turn 2 if thunder-wave fails.
  • Registeel. You HAVE TO ATTACK Registeel variants, as they are too bulky and you will lose if you don't tie MIND at least. I've painfully experienced losses to Registeel 6 because I didn't attack 3x in a row. However, Registeel 4 has thunder-wave. Having Cheri Berry here eases this prediction, on the off chance you are facing Registeel 4.
  • Regice. Destiny Bond first. If it reveals t-wave, you can safely set up a substitute on turn 2. Same principle applies.
I've also made it no secret how much I appreciate Gengar. It's truly a S tier in the Arena and it also does help Milotic quite a bit. Gengar loves setting up a substitute against status spammers, provides additional ohko move protection, checks normal types and fighting types and trades against troublesome matchups for Milotic. These are all known things that Milotic can struggle with and Gengar patches up these issues quite nicely.

The only thing it doesn't really solve is double team spammers (no Aerial Ace, weakness to DT faint attack umbreon) and Snorlax.

View attachment 399038
Milotic (F)@ Leftovers
Level: 50
Calm Nature (+ SpD, -Atk)
Ability: Marvel Scale
EVs: 252 HP / 216 Def / 36 SpD / 4 Spe

- Dive
- Recover
- Icy-wind
- Mirror Coat

Enkidu (Latios) (M) @ Lum Berry
Level : 50
Ability: Levitate
EVs: 28 HP / 12 Def / 244 SpA / 4 SpD / 220 Spe
Timid Nature (+Spe, -Atk)
- Calm Mind
- Dragon Claw
- Thunderbolt
- Psychic

https://pokepast.es/5cde506c0d83a2c2

Final Thoughts.
I want to thank Actaeon first. You were a big help optimizing sets, experimenting with me and for watching my streams. I am so glad we reached 200+ together. We've spent an absurd amount of time in the Arena (especially in the past month and a half) and we are finally rewarded for our troubles. We finally can move on to niche things and experiment for fun.
I want to thank Kommo-o, dgice, Valentino, Maizup Runeblade14 and anyone else I've missed. I appreciate your support and I'm glad people got to see this team put in the work.
I've really enjoyed these posts. It almost - almost - makes me want to try Arena again and that's the highest praise I can think of given how much I dislike that facility. Congrats on your outstanding record!
 
Hi all, just wanted to share my team.

Lead
Salamance (Tobey M.) (M) @ Lum Berry
Level: 50
Ability: Intimidate
IVs: 30/31/21/x/20/31
Evs: 252 Atk / 6 Def / 32 SpD / 220 Spd
- Dragon Dance
- Aerial Ace
- Earthquake
- Rock Slide

Ludicolo (Happy) (F) @ Leftovers
Level: 50
Ability: Rain Dish
IVs: 31/x31/10/27/28
EVs: 164 HP / 94 Def / 252 SpD
- Leech Seed
- Ice Beam
- Substitute
- Double Team

Snorlax (M) @ Chesto Berry
Level: 50
Ability: Thick Fat
IVs: 0/31/31/x/31/14
EVs: 252 HP / 252 Def / 6 SpD
- Curse
- Return
- Brick Break
- Rest

So on my first run with this team I got to a streak of 69 )': I was watching Match of the Day and did not pay close attention to the Salon Maiden battle. Anyway, I wanted to build a team around one of my favourite moves, Leech Seed. I hadn't seen any streaks with a Pokemon using Leech Seed when I started this idea (since then I have seen howtogetpikachuonabus' team, and how he used a Meganium), and so I looked at all possible Leech Seed Pokemon, and decided to use Ludicolo. Whilst Ludicolo may not be as bulky as Venasaur or Meganaium, I realised that if I paired it with a Salamance whose Intimidate can give it some sort of pseudo-increased bulk. Furthermore Ludicolo's typing is a blessing as it has no special weaknesses. With Max Special defence it can often happily set up against STAB Ice Beams, Flamethrowers and Physics, once it has one or two Double Teams up and Leech Seed, it can quite happily stall.

I considered the 3rd spot in this team for quite sometime, I needed something that resisted ice, whilst also helping against bulky water mons and rock types. Initially I considered Magneton after looking at iSoNkei's Palace team and seeing he used a Salamance, Milotic, Magneton combination. I thought for a while about having some sort of utility magneton set with Rain Dance, Thunder, Thunder Wave, Reflect or something to that effect. I also considered trying a Lantern, it's typing seemed to fit well, it would help against the rock and flying types that the other two struggled with.

After some consideration I thought that Magneton and Lantern wouldn't be good enough, they were too frail and wouldn't be able to hit hard enough. It was at this point I found howtogetpikachuonabus' team, and realised that Snorlax could fit in nicely here. Thick Fat giving it a resistance to Ice (as such) and it's natural bulk being the perfect mon to set up with. This works even better when the opponent is Intimidated AND it has a Leech Seed recovery helping it.

I copied the max defence and HP investments used as it helps tank rock type moves better on switch in.

Anyway I just wanted to share this team here, I know 69 is a pathetic excuse for a streak but I easily could have gotten further had I not been distracted, I'll continue trying with this team when breaking from exams and hopefully reach a nice streak.
PS. This is on cartridge and all Pokemon were bred, idk how to RNG manipulate.
 
Update: I have finally achieved 200+ wins! My streak has ended @207 with Gengar, Milotic and Latios. Milotic tied against Meganium who set up lightscreen and Latios couldn't beat shadow ball Exploud of all things with lightscreen up. I'm not even sad about it. I've lost close to 200-300 times in the Arena and I finally got the luck I was looking for. Too many heart breaking losses to Haxrein, Lapras and other hax to count. I'm only going to explain the different Gengar set, as I've already explained the Milotic set with a wealth of detail.

View attachment 399037
Gengar (F) @ Cheri Berry
Ability: Levitate
EVs: 84 Def / 204 SpA / 220 Spe
Timid Nature
- Thunderbolt
- Ice Punch
- Destiny Bond
- Substitute

You might notice that Gengar's evs are different from the 36 Def/ 252 SpA/ 220 Speed that I was always using. My reasoning for increasing physical defense is quite simple: Snorlax. During streaks, I kept getting screwed over by QC Snorlax (lost close to 5 times in a span of 2 days to it). I reduced Gengar's special attack to a reasonable number that risked losing some damage rolls. I settled on 204 SpA, which guarantees the ko against 252 HP Salamence.

252+ Atk Snorlax Shadow Ball vs. 0 HP / 36 Def Gengar: 127-150 (94 - 111.1%) -- 68.8% chance to OHKO. Far too risky.
252+ Atk Snorlax Shadow Ball vs. 0 HP / 84 Def Gengar: 119-140 (88.1 - 103.7%) -- 25% chance to OHKO
204 SpA Gengar Ice Punch vs. 252 HP / 0 SpD Salamence: 204-240 (100.9 - 118.8%) -- guaranteed OHKO

I have also experimented with many items for Gengar (petaya, lum, persim, magnet and cheri). I will go into my explanation why other items are not nearly as ideal for Gengar as cheri berry is.

-Lum Berry:
  • Too valuable of an item to use on Gengar. Gengar is already immune to poison, you will be using substitute against sleep powder grass types and jynx first and you already lose to prime confuse ray targets like Crobat without Destiny Bond. If no one else is using Lum, then Gengar may have it.
- Petaya Berry
  • Gengar should ko targets outright with a SE move or trade against dangerous targets. It's usually better to take the safe route, rather than try to take a greedy damage roll
- Magnet
  • Increases some damage rolls (like making 0 hp starmie a guaranteed ko) and increasing some 2koes against walrein and Lapras. However, against starmie, you shouldn't be risking the T-bolt against Psychic Starmie (could be the 255 hp one or could be Starmie 7, which carries bright powder). Against Walrein 4 and Lapras 2, if you see surf/ice beam, I'm not risking the 2ko, I take the trade after using substitute. No questions about it.
- Persim Berry
  • Very niche.
  • Given Gengar's high speed, I use substitute fairly liberally to scout and you tend to trade or don't win those matchups anyways.
-Cheri Berry.
  • This item is invaluable because Gengar loves it's speed.
  • Incidental paralysis from Zap Cannon Forretress, Thunder Nidoking and dragon breath Steelix is also a thing among others.
  • It eases prediction against known thunder wave spammers like Latios/Latias. You can safely destiny bond to catch every Latios/Latias. If either of the Lati twins reveals T-wave, you can substitute on the next turn and ice punch on turn 3. Otherwise, destiny bond is successful = trade.
  • Same applies for Zapdos, Raikou and misdreavus. I've noticed that against Zapdos though, you want to attack first, as it can double team or lightscreen. Zapdos also seems to switch to thunderbolt on turn 2 if thunder-wave fails.
  • Registeel. You HAVE TO ATTACK Registeel variants, as they are too bulky and you will lose if you don't tie MIND at least. I've painfully experienced losses to Registeel 6 because I didn't attack 3x in a row. However, Registeel 4 has thunder-wave. Having Cheri Berry here eases this prediction, on the off chance you are facing Registeel 4.
  • Regice. Destiny Bond first. If it reveals t-wave, you can safely set up a substitute on turn 2. Same principle applies.
I've also made it no secret how much I appreciate Gengar. It's truly a S tier in the Arena and it also does help Milotic quite a bit. Gengar loves setting up a substitute against status spammers, provides additional ohko move protection, checks normal types and fighting types and trades against troublesome matchups for Milotic. These are all known things that Milotic can struggle with and Gengar patches up these issues quite nicely.

The only thing it doesn't really solve is double team spammers (no Aerial Ace, weakness to DT faint attack umbreon) and Snorlax.

View attachment 399038
Milotic (F)@ Leftovers
Level: 50
Calm Nature (+ SpD, -Atk)
Ability: Marvel Scale
EVs: 252 HP / 216 Def / 36 SpD / 4 Spe

- Dive
- Recover
- Icy-wind
- Mirror Coat

Enkidu (Latios) (M) @ Lum Berry
Level : 50
Ability: Levitate
EVs: 28 HP / 12 Def / 244 SpA / 4 SpD / 220 Spe
Timid Nature (+Spe, -Atk)
- Calm Mind
- Dragon Claw
- Thunderbolt
- Psychic

https://pokepast.es/5cde506c0d83a2c2

Final Thoughts.
I want to thank Actaeon first. You were a big help optimizing sets, experimenting with me and for watching my streams. I am so glad we reached 200+ together. We've spent an absurd amount of time in the Arena (especially in the past month and a half) and we are finally rewarded for our troubles. We finally can move on to niche things and experiment for fun.
I want to thank Kommo-o, dgice, Valentino, Maizup Runeblade14 and anyone else I've missed. I appreciate your support and I'm glad people got to see this team put in the work.
Congratulations my dude! Words cannot describe how much I'm in awe for such achievement. We saw on the Discord when you were streaming how much time and effort to make this come true. Congratulations and may this record last for a very long time!!
 
PhotoGrid_1642563456301.jpg

Greetings Frontierlings! I am here today because my "Lingering Questions" about the Battle Factory have been answered. Some light has been shed upon some of the coding in the Factory, which I feel is worth understanding for those who are attempting a long streak. Credit goes to Mow and Actaeon for undertaking the task of making the Factory Assistant, being able to read the code, and figuring out the relevant typo in the equation. Due to said typo in Emerald's code, the starting draft Pokemon in Round 9 all have random IVs between 0 and 31. This also means the elevations in Round 8, which are pulled from Round 9, will have random IVs. Consequently, if you are swapping often and have a large swap count, you will be punished for your swaps in Round 8 (with random IV elevations instead of 31 IVs), but you will be rewarded for them in Round 9 (with 31 IV elevations instead of random IVs). On the other hand, if you have swapped fewer than 15 times by Round 8, you will have all 31 IV draft picks, and Round 9 will consist entirely of random IVs. Random IVs have a 3/31, or 9.68%, possibility of being lower than 3, which would be worse than a swapped Pokemon's IVs. The most common way random IVs would make a game-changing difference is if you had a sub-31 speed IV and subsequently couldn't outspeed an opponent you would've outsped with 31 IVs. Always check your IVs in Rounds 8 and 9, and weigh your team selection using calcs of the relevant potential matchups your team might encounter.

Edit: Also, Round 9 is a Noland Round where you have to overcome his 31IV pokemon. Having random IV'ed starting picks instead of the usual 31IV, is why I would dub Round 9 the hardest round in the Factory.
 
Last edited:
Posting some results of a very simple program I wrote about EV'ing for overall mixed bulk here. We know the drill -- you're looking for EV's to be as bulky as possible on both the physical side and the special side of the spectrum for a mixed wall like Umbreon, Snorlax or Registeel.

Let's assume you don't want to specifically EV to avoid being OHKOed or 2HKOed by certain threats and that you're just interested in having theoretically the best overall bulk, that is to minimize general damage, without preferring to be bulky on either side.

Before going into details, here's 'rules of thumb' we use when EV'ing for mixed bulk that we already knew / should learn, such as:
  • Consider investing in HP for mixed bulk for most Pokemon, because it "counts" towards both sides;
  • The rule above doesn't apply anymore once your Defenses together are still less than your HP stat (think of Wailord / Wobbuffet for example);
  • Mixed bulk overall gets approximately maximized if you strive to equalize your Defenses.
  • Look for jump points after applying the rules above.
These 'jump points' are the interesting part and there are two kinds of jumps: some EVs will make your stat jump from 130 to 132 for example on some Pokemon, and the other kind of jump I'm talking about is the damage output jumps that occur at some combinations of (Special) Attack, Power and of course Defense stats. These jumps are caused by rounding in the stat and damage formulas, and are hard to predict by hand.

So what did I do? As usual, I brute-forced the whole thing and wrote a program that does the following:
  1. Takes as input the base HP, base Def and base SpD of the Pokemon you want to 'outbulk';
  2. Assumes level 50 and 31 IVs in all stats (can be changed if you're particularly interested in running a Hidden Power, or can't obtain a perfectly bred Pokemon for example)
  3. Creates a big for-loop that does steps 4 and 5 for all combinations of EV that add up to 510 and with a maximum of 255 in every stat (HP, Def, SpD). Also considers either a +Def or a +SpD nature to see which is best.
  4. It performs damage calculations with offensive stats ranging from 100 to 200 and with powers equal to 80, 90, 95, 100 and 120, encompassing the most important damage calculations you will face in practice (e.g. Rhydon's EQ, Metagross' MM, Latios Psychic/IB/Tbolt, Raikou Tbolt, Jolteon Tbolt, etc. etc.). It repeats these calculations with and without STAB on both physical and special sides.
  5. Adds up all the damages calculated in step 4, and divides this by the corresponding HP stat it calculated, giving a relative damage output for the specific spread from step 3. This number usually ranges from 500 to 1500 for most fully evolved Pokemon, i.e. they faint between 500 and 1500 times.
  6. Repeats until the spread is found with least relative damage output and returns this spread, together with the actual stats.
After step 6, you know the best Nature to boost for overall bulk and the EVs that come with it. The program also evaluates the expression '1000/relDmg', where 'relDmg' is the least relative damage it found (and thus corresponds to the most efficient EV spread). This expression is a measure of what I like to call (mixed) bulkiness or staying power, and can be used to compare how bulky different species are.

A staying power of more than 2 indicates you have an extremely bulky Pokemon. Here's a list, the order is HP/Def/SpD all the time and a '+' indicates you have a boosting Nature for that stat. Note: for some Pokemon like Registeel and Shuckle it doesn't matter which stat you boost of course. I have assumed +Def for these Pokemon, but you can easily change the spread around. These Pokemon, with equal Def and SpD basestats, are colored blue.

SpeciesSpreadStatsStaying power
Lugia252/172+/84213/189/1852.508
Registeel252/92+/164187/200/1912.302
Shuckle252/84+/172127/287/2722.253
Regice252/252+/4187/167/2212.240
Regirock252/4/252+187/221/1672.240
Ho-Oh252/252+/4213/156/1752.208
Snorlax236/252+/20265/128/1332.165
Blissey252/252+/4362/68/1512.118
Umbreon252/252/4+ or 252/164+/92202/162/166 or 202/166/1622.081
Suicune252/116+/140207/165/1532.063
Kyogre252/252+/4207/156/1612.060
Groudon252/4/252+207/161/1562.060
Deoxys-D252/84+/172157/210/2022.039
Slaking212/52/244+252/127/1272.004
Articuno252/172+/84 or 252/220/36+197/156/156 or 197/148/1651.929
Lapras252/164+/92 or 252/212/44+237/133/127 or 237/127/1331.928
Wobbuffet4/252+/252266/121/1101.916
Mew/Celebi/Jirachi252/92+/164207/145/1411.856
Chansey252/252+/4357/62/1261.850
Milotic252/252+/4202/144/1461.837
Walrein252/92+/164217/134/1311.801
Metagross252/12/244+187/152/1551.801
Relicanth252/4/252+207/151/1281.797
Mewtwo252/92+/164213/134/1311.767
Rayquaza252/92+/164212/134/1311.759
Cradily252/116+/140 or 252/220/36+193/145/1451.755
Vaporeon244/236+/28236/121/1191.772
Muk252/204+/52 or 252/252/4+212/133/127 or 212/127/1331.725
Swampert252/92+/164207/134/1311.718
Entei252/44+/212 or 252/132/124+222/122/1221.695
Meganium252/92+/164187/145/1411.677
Gyarados252/172+/84 or 252/252/4+202/133/131 or 202/131/1331.670
Miltank252/12/244+202/127/1331.644
Claydol252/132+/124 or 252/244/12+167/156/1561.635
Donphan252/4/252+197/141/1231.621
Kangaskhan252/84+/172212/122/1221.618
Hariyama20/236+/252222/121/1121.618
Dusclops252/84+/172147/177/1721.613
Rhydon252/4/252+212/141/1061.605
Zapdos252/124+/132 or 252/172/84+197/133/127 or 197/127/1331.603
Moltres252/132/124+ or 252/84+/172197/127/133 or 197/133/1271.603
Raikou252/252/4+ or 252/204+/52197/127/133 or 197/133/1271.603
Forretress252/4/252+182/161/1231.591
Lanturn124/252+/132216/121/1131.579
Skarmory252/4/252+172/161/1341.578
Wailord4/252+/252246/106/971.556
Whiscash220/132+/156 or 220/140/148+213/121/111 or 213/111/1211.542
Salamence252/84+/172202/122/1221.542

Please let me know if you want to see others!
 
Last edited:

QuentinQuonce

formerly green_typhlosion
Posting some results of a very simple program I wrote about EV'ing for overall mixed bulk here. We know the drill -- you're looking for EV's to be as bulky as possible on both the physical side and the special side of the spectrum for a mixed wall like Umbreon, Snorlax or Registeel.

Let's assume you don't want to specifically EV to avoid being OHKOed or 2HKOed by certain threats and that you're just interested in having theoretically the best overall bulk, that is to minimize general damage, without preferring to be bulky on either side.

Before going into details, here's 'rules of thumb' we use when EV'ing for mixed bulk that we already knew / should learn, such as:
  • Consider investing in HP for mixed bulk for most Pokemon, because it "counts" towards both sides;
  • The rule above doesn't apply anymore once your Defenses together are still less than your HP stat (think of Wailord / Wobbuffet for example);
  • Mixed bulk overall gets approximately maximized if you strive to equalize your Defenses.
  • Look for jump points after applying the rules above.
These 'jump points' are the interesting part and there are two kinds of jumps: some EVs will make your stat jump from 130 to 132 for example on some Pokemon, and the other kind of jump I'm talking about is the damage output jumps that occur at some combinations of (Special) Attack, Power and of course Defense stats. These jumps are caused by rounding in the stat and damage formulas, and are hard to predict by hand.

So what did I do? As usual, I brute-forced the whole thing and wrote a program that does the following:
  1. Takes as input the base HP, base Def and base SpD of the Pokemon you want to 'outbulk';
  2. Assumes level 50 and 31 IVs in all stats (can be changed if you're particularly interested in running a Hidden Power, or can't obtain a perfectly bred Pokemon for example)
  3. Creates a big for-loop that does steps 4 and 5 for all combinations of EV that add up to 510 and with a maximum of 255 in every stat (HP, Def, SpD). Also considers either a +Def or a +SpD nature to see which is best.
  4. It performs damage calculations with offensive stats ranging from 100 to 200 and with powers equal to 80, 90, 95, 100 and 120, encompassing the most important damage calculations you will face in practice (e.g. Rhydon's EQ, Metagross' MM, Latios Psychic/IB/Tbolt, Raikou Tbolt, Jolteon Tbolt, etc. etc.). It repeats these calculations with and without STAB on both physical and special sides.
  5. Adds up all the damages calculated in step 4, and divides this by the corresponding HP stat it calculated, giving a relative damage output for the specific spread from step 3. This number usually ranges from 500 to 1500 for most fully evolved Pokemon, i.e. they faint between 500 and 1500 times.
  6. Repeats until the spread is found with least relative damage output and returns this spread, together with the actual stats.
After step 6, you know the best Nature to boost for overall bulk and the EVs that come with it. The program also evaluates the expression '1000/relDmg', where 'relDmg' is the least relative damage it found (and thus corresponds to the most efficient EV spread). This expression is a measure of what I like to call (mixed) bulkiness or staying power, and can be used to compare how bulky different species are.

A staying power of more than 2 indicates you have an extremely bulky Pokemon. Here's a list, the order is HP/Def/SpD all the time and a '+' indicates you have a boosting Nature for that stat. Note: for some Pokemon like Registeel and Shuckle it doesn't matter which stat you boost of course. I have assumed +Def for these Pokemon, but you can easily change the spread around. These Pokemon, with equal Def and SpD basestats, are colored blue.

SpeciesSpreadStatsStaying power
Lugia252/172+/84213/189/1852.508
Registeel252/92+/164187/200/1912.302
Shuckle252/84+/172127/287/2722.253
Regice252/252+/4187/167/2212.240
Regirock252/4/252+187/221/1672.240
Ho-Oh252/252+/4213/156/1752.208
Snorlax236/252+/20265/128/1332.165
Blissey252/252+/4362/68/1512.118
Umbreon252/252/4+ or 252/164+/92202/162/166 or 202/166/1622.081
Suicune252/116+/140207/165/1532.063
Kyogre252/252+/4207/156/1612.060
Groudon252/4/252+207/161/1562.060
Deoxys-D252/84+/172157/210/2022.039
Slaking212/52/244+252/127/1272.004
Lapras252/164+/92 or 252/212/44+237/133/127 or 237/127/1331.928
Wobbuffet4/252+/252266/121/1101.916
Mew/Celebi/Jirachi252/92+/164207/145/1411.856
Milotic252/252+/4202/144/1461.837
Walrein252/92+/164217/134/1311.801
Metagross252/12/244+187/152/1551.801
Mewtwo252/92+/164213/134/1311.767
Rayquaza252/92+/164212/134/1311.759
Vaporeon244/236+/28236/121/1191.772
Swampert252/92+/164207/134/1311.718
Meganium252/92+/164187/145/1411.677
Dusclops252/84+/172147/177/1721.613
Forretress252/4/252+182/161/1231.591
Skarmory252/4/252+172/161/1341.578

Please let me know if you want to see others!
This is incredibly useful, great work!

I'm never 100% sure I'm getting it right when it comes to defensive EVs - my rationale has always been "even-numbered defences + HP that's divisible by four, plus one" so it's good to see more optimised numbers. Or if I'm running, say, CM Suicune I simply invest fully in physical defence because it matters less when you can boost the other one, but I've always thought there were probably better ways to go about it. Interesting to see which Pokemon rank highest, Mewtwo was a surprise.
 
This is incredibly useful, great work!

I'm never 100% sure I'm getting it right when it comes to defensive EVs - my rationale has always been "even-numbered defences + HP that's divisible by four, plus one" so it's good to see more optimised numbers. Or if I'm running, say, CM Suicune I simply invest fully in physical defence because it matters less when you can boost the other one, but I've always thought there were probably better ways to go about it. Interesting to see which Pokemon rank highest, Mewtwo was a surprise.
Thanks, don't use these spreads (without thought) if you're using CM for example -- in fact I don't think this is very useful at all, but it was an interesting project. For "real" purposes, always check things by hand and suit EV spreads to do exacty what you want them to do.
 

Users Who Are Viewing This Thread (Users: 2, Guests: 5)

Top