Blog
Writing about improvement that is grounded in what the engine actually says, rather than in what sounds wise.
-
How accuracy is calculated, and why yours differs between sites
Accuracy is not a standard. Here is the exact formula Chess Wizard uses, why it produces a different number than chess.com, and what the figure is actually good for.
· 7 min read
-
Rook endgames: the 5 positions that decide most club games
Lucena, Philidor, rook behind the pawn, the cut-off king, and the rook-pawn exception. Five positions, the one idea in each, and why this is the highest-value endgame study you can do.
· 8 min read
-
Your blunder rate is a clock problem, not a chess problem
Most players lose more rating to time pressure than to any opening or endgame weakness. Here is how to measure whether that is you, and what to do about it.
· 6 min read
-
The 10 openings that lose the most rating at 1200–1600
Named five-move sequences from 20 million Lichess games in 2024, in the rating band where club players actually live. Expected score, sample size, and what to play instead.
· 8 min read
-
How to read an eval graph without lying to yourself
The eval graph is the most-looked-at and least-understood part of any game review. Four shapes, what each one means, and the trap almost everyone falls into.
· 5 min read
-
What centipawn loss actually measures (and when it lies)
Average centipawn loss looks like a skill number. It is mostly a record of what happened after the game was already decided. Here is what it measures, when it lies, and what we use instead.
· 6 min read
-
Best move, good move, inaccuracy: where the thresholds come from
Chess Wizard labels every move best, good, inaccuracy, mistake, or blunder. Here are the exact centipawn bands, why they are thresholds rather than a curve, and when a "blunder" did not actually throw the game.
· 6 min read
-
Opening repertoire for people who hate studying openings
At club level a repertoire is a clock strategy, not a theoretical one. Three systems, the one job they have to do, and why a worse opening you know beats a better one you have to invent.
· 6 min read
-
How many games do you need before your stats mean anything?
A 90% score in five games is not a repertoire. Chess Wizard stays quiet until roughly 15 games, and some charts need more. Here is the arithmetic, and why silence is a feature.
· 5 min read