About The Player Database

The Player Database is an open resource that gives greater access to competitive fighting game data. I've been watching competitive smash for a few years and often found myself asking about individual results and head to heads between top players, but I hadn't found a public database that tracks all of this information. As an aspiring programmer I figured this was a perfect first project. After working on it for a year, I am happy to finally release it! You can view the data through this website, or if you'd like you can download the sqlite database on GitHub.

Currently the database has all Smash 4 PGR level tournaments, and all non-ladder publicly listed Melee/Ultimate tournaments on start.gg with 4 or more entrants. I've built the backend to make things pretty generic, so I hope to expand to other games in the greater FGC. If you'd like to help getting a game's data up you can DM me on twitter, I would love the help.

Problems in The Dataset

Holes

All Smash 4 data is from PGR ranked tournaments, and even though Smash 4 is a recent game, there are a lot of holes in the dataset. The most extreme example of this is EVO 2015/2016, where the brackets are missing a great majority of sets. Similarly most Japanese tournaments use Challonge for top 96/128 but pen and paper for pools, which we can't process automatically. A lot of tournaments also don't report full data, meaning game counts and character data have been lost.

Split Accounts

The tournaments in the database are split among Challonge, start.gg, and a few self-hosted brackets. Because of this, merging all of the players' results has been quite difficult; for the sake of accuracy, not all merging can or should be done programmatically. For instance Tweek might appear in some Challonge brackets as TSM | Tweek, which needs to then be mapped into Tweek's main account.

Luckily start.gg has made great strides in standardizing esports data by having an account system that players can use to sign up for tournaments. In an ideal world, all data would already be tied to a single account per player, meaning this merge step would not be needed and we could have a perfect dataset. But even with start.gg some tournaments imported players into the tournament, generating new accounts, essentially duplicating the player in the database. If you notice two players that should be one, please let me know.