Welcome to the PGstats open beta!

We'd like to use cookies to help analyze and create a better experience for you.

You can find out more in our Privacy Policy and Terms of Service.

Win Rates and Pick Rates at the Online Qualifiers

By Jack "Jackie Peanuts" Moore | 06/01/21

The online qualifiers have concluded, a competition that included a whopping 36,925 games played across 11 regions. Never before have we had a single competition produce such a treasure trove of data, as every single one of these games had character data recorded thanks to the magic of smash.gg. Taking a similar approach to the one I used to analyze character trends over the past few patches, let's take a look at each character's Pick Rates and Win Rates at the online qualifiers:

click to expand. Pichu is hidden behind Peach; Toon Link is hidden behind Greninja and Little Mac
The horizontal axis of this chart represents each character's Pick Rate, how often they were chosen on a per-game basis. The vertical axis represents each character's Win Rate. The correlation between the two, as with our previous analysis, was effectively nonexistent, with an R² of 0.002.
With that total lack of correlation between Pick and Win Rates, it should come as no surprise that the Top 5s for each list are completely different. Duck Hunt barely beats out Steve for the best win rate in the qualifiers, but Steve was the only character in the Top 5 Win Rates to also have an above-average Pick Rate (1.8%). If we combined Simon and Richter into one character, however, Diddy Kong would slide into the fifth spot for top Win Rates at 59.6%, and his 1.5% Pick Rate was also solidly above average. Min Min (1.8% Pick Rate, 58.7% Win Rate) and Sonic (1.7% Pick Rate, 58.3% Win Rate) were the next most successful characters of those with above-average Pick Rates.
Pokémon Trainer and Snake, at 2.4% Pick Rates, join Pyra/Mythra, Wolf, Palutena R.O.B. and Ness as the seven characters to be picked at twice the average rate. Of these characters, five had positive win rates. Presented in descending order of Win Rate, they were Snake (56.7%), R.O.B. (54.4%), Palutena (53.2%), Ness (51.2%) and Pokémon Trainer (50.2%). Wolf (45.6%) and Pyra (40.7%) were the most used characters, but as we saw in our previous analysis, many popular characters experience a drag on their win rate as a direct result of that popularity—people either see others succeeding with the characters and try to make it work themselves or find the characters compelling enough to keep trying as they lose.
Four characters, I think, exemplify this, most notably Joker, whose 2.3% Pick Rate was just under twice average, and whose 39.4% win rate was second-lowest among characters with an average or better Pick Rate. Captain Falcon (1.9% Pick Rate, 42.2% Win Rate), Sephiroth (1.8% Pick Rate, 41.4% Win Rate) and Lucina (1.5% Pick Rate, 35.2% Win Rate) also continue to see heavy usage despite a lack of success.
At the top right of the graph, we see a cluster of characters who were very successful but see little usage. These characters include Duck Hunt, Dark Samus, Ice Climbers, Ryu, Lucario, Mii Swordfighter, Richter, Bayonetta, and Ken, all of whom have a Win Rate of 57% or better despite a sub-average Pick Rate. Some of these can be partially explained by one or two successful players— for Lucario, for Dark Samus, and for Ice Climbers, for the shotos, for example. But many of these characters are also either heavy zoners (Duck Hunt, Richter, Dark Samus) who can exploit the online delay or characters whose gameplan isn't part of a more common archetype (Bayonetta, Ice Climbers) and thus can exploit opponent matchup knowledge (or lack thereof).
Some characters whose lack of success was noted in the previous analysis like Fox and Chrom did improve in this analysis, which I suspect has to do with the prospect of offline events beyond the qualifiers for players like Light and Rivers, who have either previously avoided online events entirely or chosen to play in smaller online events. As some suggested after our previous post on Pick and Win Rates, a number of top players for characters like Fox and Chrom who suffer online simply choose not to play online, and the fact that these characters improved in the qualifiers, which for some top players was the only online event they entered, suggests to me there was something to that hypothesis. It will be fascinating to see how these characters perform as offline events return.
Finally, here is an alternate view with the X-Axis set the Pick Rate Rank rather than the raw Pick Rate, which should help with seeing through some of the more clustered parts of the graph the top of this post.
Did something else in this graph stand out to you? Tell us at @PGstats and follow for all the latest Smash analysis!