How we rate players, predict matches, and why our numbers are different Powered by Sheet Rating™
Sheet Rating™ is a rating system originally developed for chess by physicist Arpad Elo. The concept is simple: every player starts at a baseline rating. Win a match and your rating goes up. Lose and it goes down. The amount it moves depends on who you played.
Beat a player rated much higher than you? Big jump. Lose to someone far below you? Significant drop. Beat someone at your level? Small adjustment. The system is self-correcting — over enough matches, every player's rating converges on their true skill level.
We've adapted this framework for professional pickleball, with significant modifications to account for the unique characteristics of the sport.
We don't use chess defaults. Pickleball is a different sport with different dynamics — more upsets, faster skill development, and smaller rating gaps between top players. Our system is tuned accordingly.
Our model uses proprietary parameters tuned specifically for pickleball, calibrated across 7,500+ PPA matches. The parameters are optimized so that rating differences translate to accurate predicted win probabilities, reflecting pickleball's tighter competitive field. Our 87.4% prediction accuracy speaks for itself.
We also run a 5-pass convergence process across the full match history. Rather than processing matches once chronologically, we iterate multiple times so early-career matches are retroactively weighted against more accurate baseline ratings.
Result: Our model correctly predicts match outcomes 87.4% of the time — compared to 69% from the next-closest public model.
A dominant singles player isn't necessarily a dominant doubles player. The skills are different — court coverage vs. net play, raw athleticism vs. communication and positioning.
That's why we maintain completely separate ratings for each format:
Each format has its own Sheet Rating™ ladder, its own K-factor progression, and its own convergence pass. A player's Singles rating has zero influence on their Doubles rating, and vice versa.
In doubles formats, individual ratings only tell part of the story. Some partnerships consistently outperform what their individual ratings would predict. Others underperform.
We track synergy scores for team combinations — measuring how a specific partnership's actual results compare to their expected results based on individual ratings alone.
Communication, playing style compatibility, court positioning habits, complementary strengths. Two players rated 1600 might play like a 1700 team together — or a 1500 team. Synergy quantifies that gap.
This is particularly important for mixed doubles, where partnerships can shift between tournaments and the chemistry between partners varies dramatically.
Not all ratings are created equal. A player with 200 matches in our system has a far more reliable rating than someone with 10 matches. We make this explicit with confidence badges:
When you see a prediction involving a Provisional player, take it with a grain of salt. When both players are Stable, our accuracy climbs even higher than the 87.4% overall figure.
At its core, every prediction starts with the Sheet Rating™ gap between two players (or two teams in doubles). A larger gap means a higher win probability for the favored side.
But we don't stop at raw Sheet Rating™. Our prediction engine layers in additional signals:
Win probability is converted to American odds — the format most bettors and fans are familiar with. A 75% win probability becomes roughly -300 / +250. We display both the probability and the odds on every prediction card.
Our ratings are built from the ground up using official PPA Tour match results:
We use official bracket and match data only. No informal matches, exhibition results, or self-reported scores.
Transparency has limits. While we've explained the framework above, certain details remain proprietary:
We show you the outputs — ratings, predictions, confidence levels, odds — not the formulas. This is what makes our predictions unique and what keeps The Dink Sheet the most accurate public model in pickleball.
Questions about our methodology? Get in touch. We're always happy to talk data.