About Chess Statistics Database Viewer
This is a demo for easily viewing and downloading chess statistics for top players
Metrics Explained
TPR (Tournament Performance Rating)
The Tournament Performance Rating measures a player's performance level based on their results against opponents of varying strengths. It answers the question: "What rating would a player need to achieve this score against these opponents?"
TPR is calculated using the formula:
TPR = Average Opponent Rating + 400 × log₁₀(Score / (Games - Score))
CPR (Complete Performance Rating)
For perfect scores (100%) or zero scores (0%), the standard TPR formula produces undefined results. In these cases, we use the Complete Performance Rating (CPR).
The CPR method works by finding the rating at which a player's rating would remain unchanged for their achieved score plus one hypothetical drawn game against an opponent of that same rating. For example, for a player who scores 9/9, we ask: "What rating would they need such that 9/9 plus one draw against themselves (9½/10) is the expected outcome?"
The CPR formula is: CPR = e − ((n+1)/n) × 400 × log₁₀((n + 0.5 − m) / (m + 0.5))
Where m is the score, n is the number of games, and e is the average opponent rating.
Please cite the following paper if you find this helpful:
Ismail, M. S. (2023). Performance rating in chess, tennis, and other contexts. Available at https://arxiv.org/abs/2312.12700
Implementation: github.com/drmehmetismail/Perfect-Score-Performance-Rating
GI (Game Intelligence)
Game Intelligence is a human-centric score that measures a player's ability to weigh risk against potential reward. High-GI players know when to take chances and when to hold firm; they press in complex positions and stay precise when it matters most.
For comparison across events, the average human GI is standardized at 100, with two-thirds of players falling between 85 and 115. Winners of major tournaments often reach 160 or higher.
The scale ranges from 0 to ~200, with 176.1 being the theoretical maximum for a perfect game.
Implementation: github.com/drmehmetismail/World-Chess-Championships
MP (Missed Points)
Missed Points measure how many points a player effectively "left on the table" compared to the engine's top line, expressed in human-understandable terms—points rather than pawns—so that fans can understand what really changed the result.
Fewer Missed Points are better:
- 0 missed points = perfect play according to the engine
- 1 missed point = making a game-losing blunder in a winning position
- 0.5 missed points = making a game-losing blunder in a drawn position
Please cite the following paper if you find this helpful:
Seven, M. M. (2025). Intelligence: Theory and Computation. Available at https://arxiv.org/abs/2302.13937v5
For more information: https://norwaychess.no/en/game-theory/
Practical Reading Guide
- High GI & Low MP: Clinical and effective (precision and conversion)
- High GI & Medium MP: Well-timed risk that paid off under human conditions
- Low GI & Low MP: Solid moves, but not enough practical pressure or conversion
- Low GI & High MP: Risky or imprecise play that didn't yield results
Data Sources
Statistics are calculated from publicly available tournament data. Games are analyzed using chess engine evaluation to compute the various performance metrics.
Data is sourced from major chess tournaments including:
- FIDE events
- National championships and team leagues
- Super tournaments (Norway Chess, Tata Steel, etc.)
- Major open tournaments
Open Source
The statistics are calculated using open-source tools and methodologies.
For questions or feedback, please refer to the project repository.