Pickleball players seeking to understand how their DUPR rating is calculated and updated after each game.
The Duper algorithm is a consistent system that processes player data to generate ratings for everyone on the same scale.
The algorithm balances transparency with proprietary 'secret sauce,' involving thousands of lines of code to handle edge cases.
Player ratings update instantaneously after each match, a popular feature, especially with tournament data.
The algorithm compares expected scores based on team ratings to actual match scores, adjusting ratings accordingly.
Unlike old systems, ratings now adjust based on score margins, not just wins/losses, for more precise changes.
Ratings are like stacked Legos; the order has minimal impact. The goal is accurate placement, not extreme decimal precision.
Reliability indicates rating confidence based on match recency and partner variety, but doesn't directly alter the rating.
Matches have weights based on upload source and format. These weights decay over time, reducing their impact.
A new match's weight, relative to a player's total historical weight, determines how much their rating changes.
Higher reliability means smaller rating changes per match; lower reliability allows for more significant adjustments.
Both partners move in the same direction (up or down) based on team performance, but the amount varies.
Improvement requires skill development and playing in Duper-reported events to provide accurate data for rating updates.
Lack of match history can lead to inaccurate ratings. The new algorithm aims for correct ratings, unlike older systems.