By Andrew "PracticalTAS" Nestico | 03/29/21

For the first time since 2015, Super Smash Bros. Melee has a new tier list. Coordinated by Ryobeat and PracticalTAS, this tier list is the result of a survey completed by 63 of the 2019 MPGR top 100, plus Armada. Without further ado, PGstats presents the 13th Official Melee Tier List:

Read below for answers to frequently asked questions about this and other Melee tier lists.
Tiers don’t exist.
If you’re serious and not just invoking a meme from 2007, you should know that this argument has been heard many times before. Fundamentally, tiers don’t really affect the vast majority of a game’s playerbase, so they may not really affect you, but they absolutely do exist at the top level of play. At the very least, the fact that we asked a hundred top players for their tier list and none of them said “tiers don’t exist” should make it clear that there’s a consensus among said top players that tiers do, in fact, exist.
Who made this?
The tier list survey was created by Ryobeat; pestering top players to fill it out was done by Ryobeat and PracticalTAS; the subsequent math was done by PracticalTAS; the graphic was created by keiththehuman, and the content and release were done by the PGstats team led by Jackie Peanuts. The actual ballots were filled out by 64 top players, who either volunteered to do so or permitted us to input their existing public tier lists.
Why didn’t the Melee Back Room make this?
If you don’t know what that is, the Melee Back Room was a group of distinguished users on Smashboards, including top players and tournament organizers, who collectively made many big decisions for the community. They made the first 12 Melee tier lists. However, the whole site has seen steadily dwindling traffic over the past several years, and the Back Room has been inactive as well. So the reason why they didn’t do this is because they aren’t doing much of anything nowadays.
So why did you make this? And what makes you qualified?
We made this because the last tier list was made in 2015 and is in need of a few updates. And each person who submitted a ballot was an MPGR top 100 player in 2019 (besides Armada, who would have easily made the list if he hadn’t retired in 2018). As for PGstats, we make the aforementioned MPGR.
What’s the math behind the tier list?
The math is fairly simple, but with a few quirks.
  • Each player was told to rank all 26 characters from 1 to 26
  • Ties were allowed and ranks were normalized using average rank; for instance, a two-way tie for first was treated as giving both characters a rank of 1.5 and the third place character a rank of 3 as expected
  • Next, the top and bottom 5% of scores for each character (which turned out to be 3 ballots each) were dropped as outliers
  • The remaining ranks were averaged to give each character’s average score
  • Once each character had a score, K-Means Clustering was used to generate tiers (this allows us to test varying numbers of tiers and choose the best-looking result)
  • We ended up choosing 6 tiers, because it was the smallest number that captured most of the large gaps between characters
6 tiers? But there are also sub-tiers.
Yes, B tier and C tier are split into two sub-tiers each. This is a first for a Melee tier list and was done because we were close to choosing 8 tiers instead of 6; if we had chosen 8, S and A tier would be the same, and the B and C tiers would be split at the B+/B- and C+/C- splits (the only existing split that wouldn’t be in the same spot is the D/F split, which would move to before Ness). Basically it’s an additional piece of info noting that the top and bottom of B and C tier are slightly further away from each other than the top and bottom of the other tiers, but fundamentally the list has 6 tiers rather than 8.
Is this the same process as past tier lists?
We don’t know for sure, but all indications are that the outlier removal, normalization, and clustering steps were not used in the past. We think each of them improve the final product, so we felt justified in making that change.
How does the number of ballots stack up to past tier lists?
Keep in mind that the Melee Back Room has probably never had 100 members at once, but this tier list’s 64 ballots appears to be the largest sample size for a Melee tier list yet. The previous largest we saw was 39 ballots for the 2010 tier list.
How does the community tier list work?
The community tier list (coming Wednesday, March 31 at took in 1077 ballots from Twitter and Reddit users in the Smash Bros community. It uses the same methodology as above on a larger scale (dropping 5% as outliers means dropping 53 scores from the top and bottom of each character. It’s not an official list by any means, but it’s interesting to see how closely the community matches the opinions of the top players.
Why were players told to assume wobbling was legal for this list?
The survey was opened to responses back in early 2020, before the majority of events started to ban wobbling. To maintain consistency, this wasn’t changed for the last ballots that were submitted in early 2021. Who knows, in a few years we might see another tier list with wobbling banned.
The tier list in text form:
Tier | Rank | Character | Score | Rank Change | Prev Rank
S 1 Fox 1.68 0 1
S 2 Marth 2.36 +1 3
S 3 Jigglypuff 3.18 +2 5
S 4 Falco 3.56 -2 2
A 5 Sheik 4.66 -1 4
A 6 Captain Falcon 5.82 +2 8
A 7 Peach 6.84 -1 6
B+ 8 Ice Climbers 8.74 -1 7
B+ 9 Pikachu 9.62 0 9
B+ 10 Yoshi 9.69 +2 12
B+ 11 Samus 10.11 -1 10
B- 12 Luigi 12.23 +1 13
B- 13 Dr. Mario 12.61 -2 11
C+ 14 Ganondorf 14.83 0 14
C+ 15 Mario 15.53 0 15
C- 16 Donkey Kong 16.42 +1 17
C- 17 Young Link 17.31 -1 16
C- 18 Link 17.66 0 18
C- 19 Mr. Game & Watch 17.95 0 19
D 20 Mewtwo 20.22 +1 21
D 21 Roy 21.63 -1 20
D 22 Pichu 22.07 +2 24
D 23 Ness 22.78 0 23
D 24 Zelda 23.49 -2 22
F 25 Kirby 24.26 +1 26
F 26 Bowser 25.74 -1 25
PracticalTAS is one of the Smash community's most well-known statisticians, and he has worked on the community's official Melee rankings since 2015. He previously served as head editor for SSBMRank before taking over as project lead in 2018, when the project was rebranded to MPGR. He is also PGstats' statistician for Smash Ultimate and various FGC rankings. Outside of PGstats, he runs a Smash-centric Youtube channel, and you can most often find him tweeting about Melee minutiae at @PracticalTAS.
Stay tuned to (or @PGstats on Twitter) for more content surrounding the 13th Official Melee Tier List over the following weeks.
Tier List