by Caspar “Mobiusman”  van der Sman (Twitter: @moby_osman)

As a Melee spectator, 2016 so far has been a better year than I could have hoped for. There has been a large tournament featuring top talent somewhere in the world almost every weekend, plus two invitationals giving us more head-to-heads between top players than ever before. While a few months ago it looked possible that Armada had solved the metagame, in fact each major has brought out exciting new storylines, breakout performances, and upsets. More tournaments means more data, and now we have a pool of results bigger than at any other point in Melee history with which to evaluate individual performances and think about trends for the future. Inspired by the success of smashranking.eu, I think we’re now at a point where there is enough data to make a statistical ranking of players worthwhile. Of course, SSBMRank provides a more well-rounded perspective on the state of the game, and no numerical analysis will be able to replace it. But SSBMRank is a huge amount of work to collate, while stats have the advantage of being much faster to update with each new set of results.

Some of you may have already seen my posts on Reddit, but to give a brief overview of my approach: these scores are generated by the Glicko chess algorithm. As well as measuring a player’s skill, Glicko measures a ‘rating deviation’ that decreases with each match played but increases over time. Playing an opponent with a high deviation results in a smaller change in one’s own rating, as the opponent’s “true rating” is not yet determined. A deviation of 50 does not mean that a player’s skill can be considered anywhere in the range of �}50 from their score, but the higher deviations below do correspond less confident ratings. There is plenty more technical detail available on Glicko.net and Wikipedia. You might also find interesting this article from Microsoft explaining that compared to Glicko, TrueSkill is designed to evaluate multiplayer games and games that end in a draw. As neither of these apply to Melee singles, I believe Glicko is the better fit.

The data set I used includes every tournament with over 200 entrants in the last 12 months, plus Smash Summits 1&2 and Battle of the Five Gods. The early stages of the brackets are not included, as many tournaments don’t have publicly recorded pools results, but still this totals 4,471 matches. A few instances of top players sandbagging have been discounted – Wobbles at Evo, and Armada, Leffen and Ice at Heir II – as these throw out the ratings quite a bit (for example, Android beat both Leffen’s Yoshi and Ice’s Captain Falcon at Heir). Here are the results, and then I’ll talk a bit about what they tell us.

There are no real surprises at the top of the leaderboard, with the first few slots lining up as I think most people would order them. After EGLX, it looked like Leffen would not be able to maintain his placings from last summer, but at GOML he silenced his skeptics by defeating every other player in the top 5 in a row. In the era of the “big six” last year we might have expected to see the largest jump in points be between 6th and 7th place. Today, the biggest gaps are instead between Mango and Mew2King, and then between Westballz and Axe. I don’t think anybody could accuse M2K of not being a “god” in the modern era after beating Armada 3-0 at Smash Summit 2, but that match was a ray of hope in a year that’s otherwise been marked by inconsistency. M2K had dramatic 9th places losses to Nintendude and n0ne that could not be expected of any other top player. Plup and Westballz meanwhile have a clear lead on the rest of the field when it comes to establishing “demi-god” status, having each taken sets off Mango and Leffen in recent memory. Westballz took a tough loss to Laudandus at Pound, but it wasn’t enough to undo his very long list of wins over players like Shroomed, Lucky, Ice, and Axe.

Probably the most surprising aspect of this list is Colbol’s ranking, and specifically his placing above PPMD. There are a few reasons for this: firstly, 2016 has simply been a very good year for Colbol. Between 9th place at Genesis, 7th at Pound, and 1st at Fight Pitt 6 (not to mention 2nd at Frame One, which was not included in the data set), he has been a model of consistency, with no losses to players outside the top 20. Secondly, Colbol was the person to end two breakout performances from Ice Climbers players: Infinite Numbers at Pound (who had beaten Alex19, MacD and MikeHaze), and dizzkidboogie at Fight Pitt (who had taken down Abate, HugS, n0ne, and Duck). The upsets these players caused meant that eliminating them was worth more points than it might normally have been. The third factor is Battle of the Five Gods. The invitational tournaments we’ve had this year have been an invaluable source of match results between top players, but with Battle accounting for 50% of all of PPMD’s matches this year, they are of course going to skew things a little. While PP was subjected to two grueling round robins with the world’s best, Colbol has not been to any invitationals so far, giving him comparatively few encounters with other top players.

Ice has remained a consistent performer, with a long string of 9th place finishes at majors stretching all the way back to Evo. He managed to take one step further to get 7th at Genesis, including a win over Plup, but apart from this and one set against Westballz on his home turf, he has failed to cause any upsets. Still, having no losses at large tournaments to anyone not in the top 20 of SSBMRank 2015 (apart from that one pools match we won’t talk about) is enough to put him at the head of this very crowded part of the rankings. In stark contrast to Ice’s consistency we have Wobbles, who had possibly the single most impressive tournament run of the year so far at Battle of the Five Gods. Outside of this, he suffered upsets at the hands of Milkman and DJ Nintendo at Genesis, and had an unfortunate 0-2 record at Smash Summit 2. Wobbles has been absent from the last few majors, but will get a chance to defend his home turf soon at Low Tier City 4, where he has historically done very well.

While overall the last 12 months have still been very good for Druggedfox, GOML marked a low point, with his new main Fox losing to The Moon and Ryan Ford. Having taken sets off SFAT, Ice, S2J, Wizzrobe, and famously Mango with his Sheik last year, it will be interesting to see if Druggedfox simply needs more time to settle into his character and can replicate his past success. The other breakout stars of the last year placing highly on this list are Duck and Wizzrobe. Duck is best known for beating Leffen at both Dreamhack Winter and EGLX, but has also picked up recent wins over SFAT, Silent Wolf, and Westballz. The only blemishes on his 2016 record are losses to dizzkidboogie at Fight Pitt and a surprising upset at the hands of Bizzarro Flame at Pound. Dreamhack Austin finally brought out the performance from Wizzrobe that I think many people had been waiting for, with wins over MacD, Nintendude, and SFAT. At GOML he remained consistent, solidly beating Professor Pro 3-0 and losing only to Armada and Mango. If he can maintain this momentum going into this summer’s majors he may be able to establish himself beyond doubt as the new best Falcon player in the world.

The former bearer of that title is notably absent from this list. The return of Hax has been probably the most popular storyline of 2016 so far, but there is simply not enough data for the algorithm to assign him an accurate ranking (it gave him an deviation of 103). At Pound, Hax demonstrated he still has the potential to be top 10 in the world by taking down SFAT, Nintendude and Mango’s Marth. Like everybody else, I’m hoping we don’t have to wait too long to see him do it again.

This approach to analyzing results is only as good as the quality of the data available. With this in mind I am very excited looking ahead to the tournaments to come. There are three enormous events at least coming up over the next two months: Smash’n’Splash, CEO and Evo. We are really spoiled by Smash’s upward trend – more matches to watch, more storylines to write about, and more data to analyse. I don’t think that stats will ever be a true replacement for the more traditional panel-based ranking, but I do think we have better tools now to construct them than ever before.