Combining of Formes: A Comprehensive Stats-Counting Policy

Antar

is a Battle Server Administratoris a Programmeris a Super Moderatoris a Community Contributor
Official Data Miner
#1
So even before the decision to count megas separately, there's been some ambiguity about which formes get combined.

For example:
  • Gastrodon-East and Gastrodon get counted together, because that's an appearance-only variation with no competitive significance
  • Rotom formes get counted separately because they have different types (as well as signature moves)
But what about Basculins, where Red-striped and Blue-striped can have different abilities? Or the Cosplay Pikachus, which have their own signature moves? Or Zen-mode Darmanitan, whose behavior is more similar to Mega evolution than dissimilar?

And then there's the matter of Hackmons tiers, where you can have Aegislash-Blades that can't revert to Aegislash-Shield and Megas that can themselves mega-evolve?

The policy so far had been "whatever" (I actually had to look it up--as it stands, in all the above cases, usage gets combined) because the weird cases are PU (or FU) in usage, or it's hackmons metagames, where we don't tier by usage, but I would really like to lay out an explicit and universal (meaning: applies to hackmons) policy for this moving forward, especially as I rewrite the usage stats scripts.

So here's my proposal. Let's start with it and move on from there:
  1. Pokemon that don't change formes are always combined into their base forme as long as the stats and type are the same (so Pikachus get combined, as do the Basculins, but not Rotoms or Gourgeists). These differences are on the same order of difference as event moves and hidden abilities, and we shouldn't be tiering those separately.
  2. Pokemon that change forme are counted uniquely based on the set of formes they can take on, even in the case of reversible forme changes. So Megas are counted separately, Darmanitan-Zen and Meloetta-Piroutette are counted separately, as is no-attacking-move Aegislash, and in Hackmons, Charizard holding Charizardite-Y is counted separately from Charizard-Mega-X holding Charizardite-Y. On the other hand, Metwo-Mega-X holding a Mewtwonite-Y and Mewtwo-Mega-Y holding a Mewtwonite-X are counted together (proposed name: Mewtwo-Mega-XY).
  3. The same rules apply to tiering for non-mega-formes as apply to mega-formes, that is, a base cannot be tiered higher than its forme-changing variant (to avoid situations such as forcing Darmanitan to run Zen Mode or Meloetta to run Relic Song in lower tiers)

To me, this is the simplest and most consistent policy, but it is a bit different from what we're doing now (note that about 10% of RU Meloettas run Relic Song--not enough to change its tier), but I probably wouldn't be rolling this change out until I'm ready to replace the usage stats scripts.
 

Antar

is a Battle Server Administratoris a Programmeris a Super Moderatoris a Community Contributor
Official Data Miner
#2
A related issue is naming conventions. And as much as I dislike having global policy dictated by an OM, I *would* like consistency with Hackmons, so consider the following sets of formes a Pokemon can take on, how should one refer to each?
  1. Mewtwo
  2. Mewtwo, Mewtwo-X
  3. Mewtwo, Mewtwo-Y
  4. Mewtwo-X
  5. Mewtwo-Y
  6. Mewtwo-X, Mewtwo-Y
(where 4-6 are Hackmons only). So the question is how to differentiate between 2 and 4 and 3 and 5. I momentarily considered referring to 2 and 3 as Mewtwo-OX and Mewtwo-OY, respectively, but in the interest in keeping with legacy names, I think the following should be acceptible:
  1. Mewtwo
  2. Mewtwo-X
  3. Mewtwo-Y
  4. Mewtwo-XX
  5. Mewtwo-YY
  6. Mewtwo-XY
 
#3
A related issue is naming conventions. And as much as I dislike having global policy dictated by an OM, I *would* like consistency with Hackmons, so consider the following sets of formes a Pokemon can take on, how should one refer to each?
  1. Mewtwo
  2. Mewtwo, Mewtwo-X
  3. Mewtwo, Mewtwo-Y
  4. Mewtwo-X
  5. Mewtwo-Y
  6. Mewtwo-X, Mewtwo-Y
(where 4-6 are Hackmons only). So the question is how to differentiate between 2 and 4 and 3 and 5. I momentarily considered referring to 2 and 3 as Mewtwo-OX and Mewtwo-OY, respectively, but in the interest in keeping with legacy names, I think the following should be acceptible:
  1. Mewtwo
  2. Mewtwo-X
  3. Mewtwo-Y
  4. Mewtwo-XX
  5. Mewtwo-YY
  6. Mewtwo-XY
Saying Mewtwo + Mewtwo-X = Mewtwo-X and Mewtwo + Mewtwo-Y = Mewtwo-Y is going to be very confusing. You are suggesting Mewtwo-X or Y alone by appearance, which may lead the reader to not assume what you want them to assume. Why not something like this:

  1. Mewtwo-N
  2. Mewtwo-XN
  3. Mewtwo-YN
  4. Mewtwo-X
  5. Mewtwo-Y
  6. Mewtwo-XY
Sort of like what we do with Deoxys, where Deoxys-N is considered to be Deoxys' base forme. It is a lot less confusing to the reader than having to learn that Mewtwo-X does not refer to Mewtwo-X alone.
 
#4
Pokemon that don't change formes are always combined into their base forme as long as the stats and type are the same (so Pikachus get combined, as do the Basculins, but not Rotoms or Gourgeists). These differences are on the same order of difference as event moves and hidden abilities, and we shouldn't be tiering those separately.
This one could be tricky. To take your example of the Gourgeist forms, they differ in stats but retain the same moves and typing. The end result that's been observed competitively in PU is that there's little functional difference between the formes. G-XL and G-S see the most use due to the disparity in bulk/speed but both tend to serve the same purpose for these teams. In theory, you could have all four formes serving the exact same purpose on the exact same team with nothing else to distinguish them. Taken to the further extreme you could have a build where they are popular on but no one Gourgeist is necessary, such that each Gourgeist individually has 2% usage in a tier but 8% collectively. This situation might be the one where a universal approach won't work best.
 
#6
Applying some objective criteria that can be applied to each case is fine, but looking at the individual cases:

Any spread you can get with the middle Gourgeist formes can just be done better by XL and S (I did the math a while back), so Large has absolutely no competitive use in any tier where XL is allowed and Average has absolutely no competitive use in any tier where Small is allowed. However, S and XL have very different uses with XL functioning more as a wall and S functioning more as a utility attacker (it happens that XL is way better at its role and S sees almost no use in PU specifically, just because XL's role is way more useful). Because of this I think it definitely makes sense to tier them separately.

The Basculin forms are competitively identical because there is absolutely zero reason to not run Adaptability on Basculin (which is the ability shared by both forms), and either way I think it makes more sense to just treat it as a single Pokemon with four abilities.

Pikachu couldn't be less relevant of a Pokemon, but the normal form is much better than the cosplays regardless because the cosplays don't get Extremespeed which is the only thing that makes Pikachu remotely worth salvaging. I really don't care at all if these get tiered separately or not.

Also, this hasn't been mentioned yet, but I can't imagine we'd do anything but tier the Alola forms separately because it would make no sense to tier them together (referring to them as, for example, Exeggutor-Alola or Exeggutor-A).
 

Bughouse

Like ships in the night, you're passing me by
is a member of the Site Staffis a Forum Moderator Alumnusis a CAP Contributor Alumnusis a Tiering Contributor Alumnusis a Contributor Alumnus
#8
You also forgot the case of Meowstic, which you currently calculate usage for combined. The male and female have the same stats, typing, and regular abilities. But they are totally different Pokemon because they are defined by their hidden abilities. Female gets Competitive and so is used as an anti-defog powerhouse. Male gets Prankster and so is used as a screener. This is a much better test case than Basculin, who both run Adaptability anyway.

As to things like Darmanitan-Z and Meloetta-P and Cherrim-S or whatever other form changes like that can happen in-battle (aside from mega evolution), the answer is simple. If you can't start the battle with it, it's not something that should be tiered separately from its base Pokemon. If you can in Hackmons, then sure they could be tiered differently in a hypothetical Hackmons UU situation. But for standard play, they can't.
 

Vinc2612

IT'S VINCE DUMBASS
is a Tournament Director Alumnusis a Forum Moderator Alumnusis a Battle Server Moderator Alumnus
#9
I don't like taking hackmons in account because in my opinion, we need to separate two cases in the tiering of forms: Pokémon that changes form outside of the battle, and Pokémon that changes form during the battle. Hackmons make that separation impossible, and makes the whole thing more complicated.


1/ Pokémon that change form outside of the battle

If the Pokémon are exactly the same competitive-wise (read Gastrodon) OR one form totally outclass the second by the possibility of doing exactly the same thing + something else (Keldeo-R > Keldeo, we don't need the latter in BL), those should be tiered together. If it can be formulated differently so it works for Pikachu, that's even better (because in theory, standard Pikachu can use static / be male while cosplay can't).

If the Pokémon have different typing, different stats or different abilities, they should be tiered differently (Rotom, Meowstic, Basculin...).


2/ Pokémon that change form inside of the battle

If the Pokémon is a Mega, then the base form is tiered separately, but the mega cannot be in a lower tier than the base form.
If the Pokémon is not a Mega, then the Pokémon is tiered as a whole.
If the Pokémon is a flower hedgehog, then we can all be thankful for Shaymin-Sky to be ranked higher than the land form. But this could become an issue in the future if another Pokémon gets something similar.
 

Antar

is a Battle Server Administratoris a Programmeris a Super Moderatoris a Community Contributor
Official Data Miner
#10
You also forgot the case of Meowstic, which you currently calculate usage for combined. The male and female have the same stats, typing, and regular abilities. But they are totally different Pokemon because they are defined by their hidden abilities. Female gets Competitive and so is used as an anti-defog powerhouse. Male gets Prankster and so is used as a screener. This is a much better test case than Basculin, who both run Adaptability anyway.
Yes, under my proposal, Meowstics would get tiered together, as they are now. The ability difference to me is no more significant than a difference between Magic Guard and Unaware Clefable, and they get combined as well.

As to things like Darmanitan-Z and Meloetta-P and Cherrim-S or whatever other form changes like that can happen in-battle (aside from mega evolution), the answer is simple. If you can't start the battle with it, it's not something that should be tiered separately from its base Pokemon. If you can in Hackmons, then sure they could be tiered differently in a hypothetical Hackmons UU situation. But for standard play, they can't.
Note that all Cherrim (and Castform) outside of Hackmons have theoretical access to their weather formes since they can't control whether an opposing Pokemon uses a weather move.

Hackmons-consistent naming convention would be to call Forecast Castform "Castform" and non-Forecast-Castform "Castform-NoWeather."

Your point regarding Pokemon not starting in the forme, in my opinion, needlessly makes Megas an exception rather than the basis for the rule. I see no downside to tiering DZ, RS Melo or Utility Aegislash separately*, and the advantage is a drastic simplification in our tiering policy such that we no longer have to say "We tier Megas separately because they're special."

Vinc2612, I believe this largely addresses the points in your post.

*a consequence of this policy is that Aegislash with no attacking moves could be allowed outside of Ubers. I wonder where such a mon would fall...
 

Vinc2612

IT'S VINCE DUMBASS
is a Tournament Director Alumnusis a Forum Moderator Alumnusis a Battle Server Moderator Alumnus
#11
and the advantage is a drastic simplification in our tiering policy

*a consequence of this policy is that Aegislash with no attacking moves could be allowed outside of Ubers. I wonder where such a mon would fall...
That's where I disagree and the whole point of my post. I don't see that as a drastic simplification, it's actually more complicated.


Yes, under my proposal, Meowstics would get tiered together, as they are now. The ability difference to me is no more significant than a difference between Magic Guard and Unaware Clefable, and they get combined as well.
I didn't see it that way. It makes sense and I'm fine removing the ability part. I was just trying to make it as simple as possible, so I care more about the in-battle form changes for that matter.

In the cases of DZ or RS Meloetta, I agree it doesn't matter. The alternate forms would respectively be (UU) and (RU) because the starting form is. But having Aegislash "allowed without any offensive move" sounds like "Greninja without protean" or "Blaziken without Speed Boost". You're effectively banning half a Pokémon. And in the case of Shaymin-Sky, where you can't control the reversion to the Land form, it would be even more of a mess (well, not atm because the land is BL2, but what if their situation was inversed/will be inversed in the future?)
 

Antar

is a Battle Server Administratoris a Programmeris a Super Moderatoris a Community Contributor
Official Data Miner
#12
Vinc2612, first off, Shaymin-Sky is Shaymin-Sky. Or, rather:

{shaymin, shayminsky} = Shaymin-Sky.

There is no Shaymin-Sky that cannot revert to Shaymin under any circumstances. The Pokemon that can only take on the form "shayminsky" does not exist.

Re: Aegislash -- I'm sorry, I wasn't being careful bringing up kind of a weird hypothetical. What I meant is that OU could CHOOSE to unban Aegislash-with-no-attacking-moves, but if they want to keep all Aegislash banned, that's fine and not at all inconsistent.
 

Vinc2612

IT'S VINCE DUMBASS
is a Tournament Director Alumnusis a Forum Moderator Alumnusis a Battle Server Moderator Alumnus
#13
Shaymin-Sky reverts to Shaymin-Land when frozen. It can change its form during the battle, even if it's a really specific case in the tiering
 

Antar

is a Battle Server Administratoris a Programmeris a Super Moderatoris a Community Contributor
Official Data Miner
#14
Shaymin-Sky reverts to Shaymin-Land when frozen. It can change its form during the battle, even if it's a really specific case in the tiering
Yes. That's what I'm saying. Just as all Forecast Castforms have the potential to take on the form Castform-Rainy because an opponent could use Rain Dance, so all Shaymin-Skys have the potential to revert to Land forme.
 

Vinc2612

IT'S VINCE DUMBASS
is a Tournament Director Alumnusis a Forum Moderator Alumnusis a Battle Server Moderator Alumnus
#15
Oh nvm, I got lost in the double negation. I also realised that I didn't understand your whole point.

If I (finally) understand correctly, you don't want to separate stuff as Cherrim and Cherrim-S outside of hackmons because all Cherrim are also potential Cherrim-S. But you do want to separate Darmanitan and Darmanitan-Zen because regular Darmanitan is not forced to go the Zen Mode route, by picking Sheer Force. And in this example, Cherrim would be PU, Darmanitan UU, and Darmanitan-Zen (UU).
Just repeating stuff just to be sure I understood correctly.

Ok then I can finally understand where does it come from, and I don't have any objection remaining. Sorry for the whole misunderstanding.
 
#16
Just for clarification purposes is there a particular reason Hackmons is receiving so much consideration when it comes to naming convention? Things like the Mewtwo example have no relevance outside of that currently non-official tier.
 

Antar

is a Battle Server Administratoris a Programmeris a Super Moderatoris a Community Contributor
Official Data Miner
#17
Texas Cloverleaf, it's so usage stats are consistent across all tiers supported by PS, and Hackmons is the home of most of the edge-cases. Basically, if we get a system that works for Hackmons, it'll work for pretty much anything (and I can be decently assured that it'll be robust for future generations). At some point I'll need to tackle some of the stranger mod tiers, but that's out-of-scope for now.
 

Karxrida

Travels at the speed of DARKNESS
is a Community Contributor
#18
I don't like taking hackmons in account because in my opinion, we need to separate two cases in the tiering of forms: Pokémon that changes form outside of the battle, and Pokémon that changes form during the battle. Hackmons make that separation impossible, and makes the whole thing more complicated.


1/ Pokémon that change form outside of the battle

If the Pokémon are exactly the same competitive-wise (read Gastrodon) OR one form totally outclass the second by the possibility of doing exactly the same thing + something else (Keldeo-R > Keldeo, we don't need the latter in BL), those should be tiered together. If it can be formulated differently so it works for Pikachu, that's even better (because in theory, standard Pikachu can use static / be male while cosplay can't).

If the Pokémon have different typing, different stats or different abilities, they should be tiered differently (Rotom, Meowstic, Basculin...).


2/ Pokémon that change form inside of the battle

If the Pokémon is a Mega, then the base form is tiered separately, but the mega cannot be in a lower tier than the base form.
If the Pokémon is not a Mega, then the Pokémon is tiered as a whole.
If the Pokémon is a flower hedgehog, then we can all be thankful for Shaymin-Sky to be ranked higher than the land form. But this could become an issue in the future if another Pokémon gets something similar.
Normal Keldeo can legally have Secret Sword by trading it from BW2 to BW (which doesn't have Resolute Form in the data but does have SS) and then sending it through Bank. Resolute is completely aesthetic for all intents and purposes.
 
Last edited:

Antar

is a Battle Server Administratoris a Programmeris a Super Moderatoris a Community Contributor
Official Data Miner
#19
User Pyritie PMmed me to point out that the Meowstics have different move pools in addition to different abilities. Doesn't change my opinion--it's more dramatic a difference than the Basculins, but it doesn't quite reach the magnitude of typing or stat change and NOT combining opens up the policy to event mons with exclusive moves and otherwise unobtainable HAs.
 

Quite Quiet

sleep life away
is a Tournament Directoris a member of the Site Staffis a Super Moderatoris a Community Contributoris a Tiering Contributoris a Smogon Media Contributoris a Battle Server Moderator
Articles Co-Leader
#20
I'm asking for clarification:
Pokemon that don't change formes are always combined into their base forme as long as the stats and type are the same
Does this mean you want to tier every Arceus typing and Giratina-O separately from their base variants as well? Neither can change form in battle, but they also has differences in stats (Giratina-O) or typing (Arceus) compared to their base forms.
 

Antar

is a Battle Server Administratoris a Programmeris a Super Moderatoris a Community Contributor
Official Data Miner
#22
Quite Quiet, correct. There'd be no change to our policy regarding Giratina and Arceus (that is, they still get tiered separately). Flip side is that Genesects still get tiered together because the only difference is change in one attack's type.
 

Antar

is a Battle Server Administratoris a Programmeris a Super Moderatoris a Community Contributor
Official Data Miner
#23
For those uncomfortable with a hypothetical "Aegislash-Shield" being allowed in lower tiers, Darvin just pointed out the following to me in a PM:

Also, on a tangentially related point: even if it doesn't have attacking moves an Aegislash can still struggle and enter blade forme so at present all legal Aegislash can potentially change forme.
So, by the same logic as there is no ['Shaymin-Sky'], so there is no ['Aegislash-Shield'] outside of Hackmons!

On that note, what I'd probably do is apply the following convention (note that the latter two formes can only exist in Hackmons):
  • Aegislash-Shield, Aegislash-Blade -> Aegislash
  • Aegislash-Shield -> Aegislash-Shield
  • Aegislash-Blade -> Aegisladh-Blade
 

MattL

I have discovered a truly remarkable CT which this box is t-
is a Smogon Media Contributor Alumnusis a Battle Server Moderator Alumnus
#24
Yes, under my proposal, Meowstics would get tiered together, as they are now. The ability difference to me is no more significant than a difference between Magic Guard and Unaware Clefable, and they get combined as well.
I want to continue with this point because it's possible that future generations could introduce situations for which your 3-point proposal isn't ideal. It's misguided to try to create a policy that works for all hypothetical cases, but I'm still going to present one because it's in the realm of possibility while being more extreme than the quoted example above.

Suppose that a Pokemon is introduced with two formes, an offensive forme with Huge Power, and a defensive forme with Fur Coat. Both have the same typing and stats, and both formes are competitively viable (this seems really unlikely, but a Pokemon so powerful it was banned from Ubers also seemed really unlikely, didn't it). Under your policy, these two formes would be tiered together, but the difference between them in battle is far greater than the Gourgeist formes, for example.

If situations like this arise, where going against your policy is the "correct" decision, what would your course of action be? You could:
  1. Enforce the policy for consistency's sake, despite it being suboptimal in this situation. I don't think this is the best solution because making the correct decision should trump following precedent.
  2. Redefine the policy each time a situation like this appears to be suitable to each existing situation. I don't like this one either because eventually we could come to a point where there doesn't exist a good policy that works for every situation, and we should probably try not to have a policy that is constantly subject to change whenever GF throws something at us.
  3. Treat each case ad-hoc. Despite this being much more work, it is feasible due to the small number of cases there are. The problems with this approach (besides the increased effort to implement) are deciding who gets to vote and the loss of objectivity and structure to the policy, and these could easily become major problems.
  4. Have a similar discussion to this one whenever we feel like a change needs to be made. This might be the best approach because it's probably impossible to create a policy that won't need to be adjusted.