What the games say
Chess writing built on measurements rather than received wisdom, from the move data of every game this site has read.
- ·7 min read
More games will not raise your rating on their own
Volume only helps when something feeds back into it. Here is what error data says about how many games to play, and when extra games start costing you.
- ·6 min read
Rank your chess mistakes by cost, not embarrassment
Which errors actually drain your rating, which ones only feel bad, and what 225,521 engine-classified moves say about where the damage happens.
- ·8 min read
Deliberate practice in chess: what transfers, what doesn't
How deliberate practice actually applies to chess, which parts of it don't transfer, and what engine data on 219,622 moves says about where to aim.
- ·7 min read
Most "personalised" chess training is just a filter
How to tell training built from your own games apart from a generic library sorted by a tag, with error-rate data from 217,199 analysed moves.
- ·8 min read
How much rating can you actually gain in 30 days?
What a month of focused work does to your rating, why the number itself is a poor signal, and the checks that tell you by day seven whether it is working.
- ·8 min read
The endgames you actually get, and the ones you never will
Which endgame types show up in amateur games, why rook endings dominate, what our move data says about endgame error rates, and which famous endings you can skip.
- ·7 min read
How Many Chess Games Should You Actually Analyse?
One game deeply or fifty shallowly? The right number depends on what you're looking for. Here's how to decide, with data from 197,870 engine-classified moves.
- ·7 min read
Chess.com vs Lichess: the one difference that matters
Both sites can make you better. The axis that decides it is how much friction sits between finishing a game and seeing the move that lost it.
- ·8 min read
What the Stockfish evaluation number actually means
A plain explanation of centipawn scores, depth, and mate values — plus the four ways club players misread the number and what to check instead.
- ·7 min read
You don't need an opening repertoire yet. Here's why.
Below 1400, opening prep rarely decides your games. Engine data on where errors actually cluster, and what to learn instead of memorised lines.
- ·8 min read
Where your blitz clock actually goes
A per-move breakdown of how blitz time gets spent, which of those seconds change the result, and how to audit your own clock from your last three losses.
- ·7 min read
Converting a winning position is a separate skill
Most won games are thrown away in the middlegame, not the endgame. Where conversion actually breaks down, with engine data from 5,754 analysed games.
- ·8 min read
The Woodpecker Method, and the week most people quit
What the Woodpecker Method asks you to do, how good the evidence for it is, and why the second cycle is the part almost everyone abandons.
- ·8 min read
Why more puzzles stopped fixing your blunders
Your puzzle rating climbs while your games still fall apart. Here is what puzzle sites leave untrained, what engine data shows about when you err, and what to do.
- ·7 min read
What a free chess game review actually tells you
Free engine review gives you more than most players use. Here is what it covers, what stays paywalled, and how to review a game with just a board.
- ·8 min read
The blunder you keep making isn't in your last game
A blunder checker lists the errors in one game. Here is how to find the error you repeat across twenty games, and how to sort your blunders by cause.
- ·7 min read
Stuck at 1500? You stopped blundering and started stalling
At 1500 the losses change shape: fewer hung pieces, more won positions that leak away. What the move data shows, and how to train conversion.
- ·7 min read
Stuck at 1000? It's your blunder rate, not your openings
One move in 13 is a real error in the games we analyse, and most of them land in the middlegame. What that means if your rating won't move past 1000.
- ·7 min read
Build your chess improvement plan out of your own losses
How to turn fifteen minutes a day into a training routine built from your own games, using positions you actually got wrong instead of a generic curriculum.
- ·7 min read
The things that actually move your rating, ranked
A ranked list of what changes your chess rating, based on where errors actually happen in games, and what gets far more coverage than it deserves.
- ·7 min read
The right order to analyse your own chess games
A step-by-step order for reviewing your games, plus the two steps an engine will never prompt you to do: guessing first, and naming the cause.
- ·8 min read
Accuracy is the worst number in your game report
Accuracy compresses a game into one number driven by two or three moves. Here is what to read in an engine report instead, with data from 107,460 moves.
- ·8 min read
What an engine can't tell you about your own chess
An engine grades every move you played. It cannot tell you why you played it. Here is what to ask a coach instead, and how to get most of it yourself.
- ·8 min read
What an AI chess coach is actually good for
A concrete plan for using AI to improve at chess: what to hand it, what to ignore, and what the first four weeks of doing this looks like in practice.
- ·7 min read
Most AI chess tools analyse. Few tell you what to fix.
The gap between a site that runs Stockfish on your game and one that reads your last 40 games and decides what you should practise, with numbers from our own data.