Who is STATISTICALLY the Best Pokémon?

In today's episode, I set out to PROVE which pokemon is the best, using statistics!

Feel free to use it/improve upon it if you want, just let people know where you got it!

[Double-Negative]

Huge power doesn’t double the base stat. It doubles the stat. That makes it even more busted

[wildmurkrow]

Fun Fact about Aegislash it's stats were actually nerfed from 520 to 500 in Gen 8.

[jeziahfrancis]

Also, huge power doubles the actual stat of a Pokemon, rather than the base stat. They don't scale linearly. Mega Mawhile would actually have a base attack around 260, lol

[awkwardvulpix]

My only criticism. Megas are obviously better than normal pokemon. But I believe the item slot is extremely valuable. I would've included a Pokémon's ability to hold an item into the equation to some extent. While it almost exclusively focuses against the megas, for instance, Garchomp is used more than Mega Garchomp despite the Mega having a higher attack stat. While part of the reason may come down to the speed decrease, I believe the inability to run Scarf Mega, Rocky Helmet mega, Yache Mega, etc, drags down Mega Garchomp a lot.

[cice]

Subscribed, but would love a reblanced list in collaboration with a competitive player; their knowledge on abilities and moves would really help make the jank less jank. Additionally, having a separate list for each generation would be more accurate, as not all Pokémon are always available or even the same between games.

Maybe cross referencing an ability tier list and giving them numerical values based on their ranking? 10 is S 0 is F tier.

[jschmidty]

Even if Luxray has access to two powerful abilities, it doesn't seem fair to count points for this twice when a pokemon can only have one ability at a time

[plentyofpaper]

Lots of problems with this method (as I'm sure you're aware.) Perfecting the formula would be a truly daunting task, but I'd like to comment in with some direction that a more thorough analysis might take.
Assigning equal value to each stat is a real shortcoming. Especially on the defensive side.
HP can be used for both physical and special bulk, making it generally more useful than defense or special defense.
Further analysis though will reveal a more complicated issue. The damage formula is (mostly) just multiplication. So, a Pokemon's durability is basically it's HP multiplied by its defense or special defense respectively.
A Pokemon with base 120 HP and base 120 Defense is drastically bulkier than a Pokemon with 239 HP and 1 Defense, or a Pokemon with 1 HP and 239 Defense. This mistake is why you have Shuckle on your list. It's bulky, but not nearly as bulky as the sum of its defensive stats would suggest. The product is much more relevant.
So defensive stats should be weighted as a product of HP and Defense/Special Defense and evaluated from there... but there's a further complication.
At level 100, a Stat = (2*Bask stat + 5 + IVs + EVs/4)*Nature bonus. Except for HP, which has a 3 as a base stat coefficient instead of 2. (3*Bask stat + 5 + IVs + EVs/4)*Nature bonus.
The change in coefficients, coupled with the influence of EVs, IVs, and that arbitrary +5 complicates the scenario so you can't just multiply base stats.
You can create a complex formula to evaluate them, but that's more work than I want to dive into for now.
My lazy approximation (which should be a decent estimate) would be:
Physical bulk = 1.3*Base Defense * Base HP and similar for the special defense.
1.5 would be the more direct ratio of the raw stats, but EVs and IVs have a more fixed influence, which dilutes the multiplication effect, so I'm ballparking 1.3 as a crude estimate.
Additionally, balanced defenses are preferred so your tanky beast can stand up to anything. Blissey is basically invincible on the special defensive side, but poor on the physical side. Sacrificing some special bulk for some physical bulk would be preferred.
As such, the lower defensive bulk should be weighted more than the higher defensive bulk.
Both are still relevant though, so I'd probably weight the higher one at 70% the value of the lower one.
On the offensive side, there's only one attacking stat, plus the movepool to calculate damage. So much simpler.
But the attacker gets to choose which attacking stat is going to be relevant. So the higher attacking value should be rated more strongly than the lower one. A Pokemon with base 150 attack and base 80 special attack is really no better than a Pokemon with base 150 attack and base 0 special special attack. Although a Pokemon with base 130 and 110 is likely to favor physical, but still get some use out of its special attack.
Definitely an art form to determine how to evaluate the stats (especially since movepool makes a huge difference.) I'd take the higher base stat at full value. Then I'd 1/4 (lower attack stat/higher attack stat)^3. For balanced attack stats, that's a 25% boost in attacking potential. For something with twice as much in one attacking stat, that's only 1/16th boost. For something with a 10% lower secondary attack stat, that's an 18% boost. I think those are fair numbers (although honestly, the half value really should be zero.)
Speed is really difficult to weight properly. It's most useful for attackers, but has defensive use too. And in any particular engagement, it's either higher or lower than the opponents.
Also, since you need to KO opponents to win, offense should be valued at least a little higher than defense. You can win through passive damage stalling, but in general, the more offensive team will be able to set the pace of the battle. I'd value defensive strength as maybe 80% as value as offensive strength.

[dabluestone]

Definitely room for improvement. The way you did abilities needs work, generalizing all of the species with a very small list of abilities definitely skewed the final result. Some abilities that define certain species get left out, and a lot of the time how the ability synergies with the rest of the kit is what makes it a good ability

[matthewmeyering]

For anyone wondering about shedninja’s score with wonderguard accounted for in its resistances (allows it to be immune to all non-super effective attacks), it’s score ends up being 5.218, just .03 off of tying with mega-lucario for 10th place and being 13th overall :D

[TheChiptide]

Hey everyone, SUPER appreciate all the love and support I’ve been getting, glad you all love the video!

Also, let me save you some typing time, I know the way I calculated the ability score was really bad, that was the joke. This video was less about actually finding a concrete answer (that would have taken forever), and more about having fun, learning about how simple statistics can be, and also maybe a lesson in how people can use data to mislead people. But I appreciate the comments, even the ones correcting me, it wouldn’t be science without a peer-review process!

[Kryptictony]

The effort that went into this video...I'm so glad this showed up on my recommended. Quality YouTuber, you earnt yourself a subscriber, keep up the great work man! Can't wait for your future vids!

[Aquilenne]

Something to account for with Shedinja is that its base HP appears higher than it really is.

Technically a theoretical pokemon with base 1 hp via normal scaling would have 112hp uninvested, or 143 with IVs, while Shedinja has 1, and can't benefit from IVs. With each point of base HP adding 2 total hp, Shedinja has an effective base hp of -70 rather than the 1 it gets recorded as.

[Biro_silly]

the top ten are NOT competitive powerhouses, but the math was Chef's kiss(also galar has a LOT of Op MONS, they could probably be half of the top ten) Great video)

[UGMD]

I loved the unique angle you took in this video! While it isn’t accurate at all for singles or doubles, I thought the workarounds you took for laziness and from lack of competitive knowledge were pretty fun

[B.y]

We all know that Landorus-Therian is the secret number 0 spot

[Camphorous]

The key to pokemon is looking at network synergies, not just traits in isolation.

For example, the stats are only important in context of the move pool. If a pokemon has a huge attack stat but the high damage moves are only special, that's no good.

Likewise, moves that cover a pokemon's type weaknesses are more valuable than moves of a random type.

I also don't think abilities should be rated as a single category and should instead be split up into function types and added to the move pool or type analysis. For example, some abilities grant an immunity and some serve as a stat buff or a status condition.

I suppose I'd be looking at each stat as a trait, with the pokemon's ability to raise or make use of it factored in as a modifier, plus resistances as a second trait and field control (status conditions, switching, flinching, hazards) as a third trait.

[patfitch]

I think it would be cool to see a reworked version of this list which takes into account the synergies between different stats. A fast Pokemon with high attack is generally WAY better than a fast Pokemon with high defense. A Pokemon with high defenses won't get much use out of them if they don't have the hp to back it up.

[bastianandieschack]

I would think if you put gen 8 into account Zacian Crowned would... Take the crown, Fairy Steel, Copius amount of attack and speed, ability that further increases its attack

[IDerIBarde]

Ok... Love the idea of all of this. Few Thinking's

1. Always scale linearly: 1/x is weight not linearly. -x + Offset is.
2. SpA is not useful is A is high.
3. Ability rankings for all abilities exist on YouTube. Rank them with a constant offset. If you normalize you can easily use a few videos for this.
4. Include usage. E.g. in VGC wins or successful nuzzlocks.
5. Exclude Megas.

[Ryukurai]

One thing I feel like you kind of didn't mention or cover is how having both a high attack and special attack essentially suffer from diminishing returns - since the vast majority of pokemon only care about one or the other (and if a pokemon has high in both, it's likely they're being handicapped since you very rarely need both - and having a higher defensive stat or speed stat would matter much more).