Performance Leaderboard

Bayesian Smoothed Win Rate

This leaderboard uses Bayesian smoothing to provide more reliable win rate estimates, especially for players with fewer games. The smoothed win rate balances a player's actual performance with the overall average, giving more weight to actual results as sample size increases.

Calculation Method

Smoothed Win Rate = (Player Wins + 13 × Global Average) ÷ (Player Games + 13)

The global average is currently 50.2% — the overall win rate across all competitive matches (naturally near 50% since every match has one winner and one loser).

W = 13 games — the median number of matches played by players in the selected period. This self-calibrates as the dataset grows: players with fewer games than the median are pulled toward the global average; players with more games are ranked primarily on their actual results.

Why This Approach?

Raw win percentages can be misleading for players with few games. A player who goes 3-0 would have 100% win rate, but this doesn't reflect true skill level. Bayesian smoothing gives these players a rating closer to the global average until they accumulate more games to prove their performance.