An engine called Maia focused on predicting human moves accurately instead of just being the strongest in chess, resulting in a more meaningful impact, especially for club-level players.
By individualizing chess engines to predict moves of specific players, accuracy can be increased by 4-5% and players can be identified with 98% accuracy from a pool of 400, based on their game patterns.
Identifying players through their mistakes is a crucial aspect - as mistakes are unique to individual players, understanding and fixing them can greatly aid in chess improvement.
Not all equal chess positions are the same, a sharpness score helps quantify how easy it is for a position to go wrong for either side.
The sharpness score can help in comparing different openings, understanding the impact of piece exchanges on position sharpness, and providing additional insight into the evaluation of a position.
By combining evaluation with sharpness score, it's possible to differentiate between positions that have the same numerical evaluation but are different in nature.
Analyzing chess games using LC0's WDL can provide a more insightful overview of the game compared to centipawn graphs.
Increasing the number of nodes per move in analysis results in spikier graphs, showing more extreme evaluations; finding a balance between accuracy and relevance to human play is important.
Using WDL contempt values in LC0 analysis can adjust the winning probabilities based on player ratings, offering a new perspective on game outcomes.
Average centipawn loss (ACPL) gives an overview of the quality of a chess game by indicating the average number of centipawns a player loses during a game.
Looking at the distribution of centipawn loss (CPL) in a game can provide more insightful information than just relying on ACPL, especially in drawn games.
Analyzing CPL distribution for multiple games by the same player can show patterns in their play, including frequency of blunders and small mistakes compared to other players at their rating.