Should You Analyze Wins or Just Losses? Both. Your Mistakes Hide in Wins Too

Analyze both. Your recurring mistake hides in wins just as much as losses

Review your wins and your losses, not just your losses. That is the direct answer. The same mistake that costs you a loss can sit quietly inside a game you won, unnoticed, because the result never forced you to look back.

This matters most if you are stuck on a plateau. A plateau is a stretch where your rating sits near the same number for weeks or months, without going up or down. Say you keep hanging pieces. A hanging piece is a piece left with no defender, so your opponent takes it for free. If your opponent misses the free piece sitting there, you still win the game. The mistake happened. It just did not get punished that time.

The same goes for forks. A fork is one move that attacks two of your pieces at once, so you can only save one of them. Getting forked in a losing game is obvious. Getting forked in a game you still managed to win, because your opponent made a bigger mistake right after, is just as real a leak. If you only look at your losses, you never see it.

BlackBox Chess is built around this exact idea. It looks across your last 10 to 100 games. It ranks the mistake type that repeats most. That holds no matter how any single game ended, in a win, a loss, or a draw. The result of any single game does not change whether the mistake happened. Only looking at every game does.

A win can bury a hanging piece or a missed fork your opponent never punished

Say you hang a piece in the middlegame. The middlegame is the part of the game after both sides finish developing their pieces. Your opponent has a free capture sitting right there. But they do not take it. Maybe they are focused on their own plan. Maybe they simply miss it. A few moves later, they make a bigger mistake themselves. You go on to win the game.

What just happened to your hanging piece? Nothing, as far as your memory of that game is concerned. You won. You feel good about it. You move on to the next game without a second thought. But the mistake still happened. You still left a piece undefended. Your opponent just failed to collect it that particular time.

The same thing happens with forks. A fork is one move that attacks two of your pieces at once, so you can only save one of them. Say a fork was sitting on the board for your opponent to play against you, and they walked right past it because they were chasing a different plan. You survive that moment untouched. The game ends in your favor. But the position that made you vulnerable to that fork in the first place is still something you created, and nothing about winning the game addresses it.

This is the real problem with judging your games by result alone. A win tells you the game ended well. It tells you nothing about whether you actually played the position correctly along the way. The leak stays exactly where it was, quietly waiting to show up again in your very next game, whether that one ends in a win or not.

Losses-only review shows you what got punished, not what you actually do most often

Reviewing only your losses sounds efficient. It is not the full picture. When you only look at games you lost, you only ever see the mistakes that were bad enough to actually lose the game. That is a narrow, skewed slice of everything you actually do at the board.

Think about what that filters out. A small leak, like a hanging piece your opponent overlooks, never shows up in a losses-only review, because the game did not end in a loss. A pin can quietly restrict one of your pieces for ten moves. A pin is when a piece is stuck in place because a more valuable piece sits right behind it. Say a pin like that never directly costs you material. It will not show up either, because the game's result does not reflect it. None of these smaller, more common leaks make the cut when the only games you open are the ones you lost.

What you are left with, if you only study losses, is a pile of your biggest blunders. A blunder is a bad move that hands your opponent a big advantage. Big blunders matter, but they are often one-off, dramatic moments, not the pattern that repeats across most of your games. The mistake type that actually costs you the most rating over time is usually small. It happens constantly. It shows up in wins and draws too, not just the rare catastrophic loss.

So a losses-only sample tells you what got punished hardest. It does not tell you what you actually do most often. Those are two different questions, and only one of them tells you what to fix first.

BlackBox's own recurrence numbers only work because they count every game, not just losses

Here is a real number that only makes sense once you understand where it comes from. In BlackBox Chess's data, among players rated 1200 to 1399 who get forked, that mistake recurs an average of about 27 times across their games. Hanging pieces recur about 7 times per affected player in that same band. A hanging piece is a piece left with no defender, taken for free. A fork is one move that attacks two of your pieces at once, so you can only save one of them.

Think about what those numbers would look like if BlackBox only counted losses. A fork that shows up but does not end the game does not always cause a loss. Neither does a single hanging piece, especially early on when the game is long from over. Say the count only included losses. Both numbers would drop, and drop hard. Not because the mistake happens less. Because fewer of those instances happened to end in defeat.

That is exactly why the recurrence count has to come from every game checked, wins and draws included, not just the losses. A count built only from losses would undercount the real pattern. It would tell you the mistake happens rarely, when really it happens constantly and simply does not always decide the result.

This is the practical reason "analyze everything" beats "analyze only losses." The whole point of finding your recurring mistake is knowing how often it actually happens. A losses-only count answers a different, narrower question. It gives you the wrong number for the one you actually need.

Let a 10-game scan cover your wins and losses so you don't have to sort them yourself

Sorting your own games into wins and losses, then deciding which pile actually matters, takes real time. BlackBox Chess was built to skip that sorting entirely.

It runs Stockfish across your last 10 games. Stockfish is a strong chess engine, a program that scores every position and finds the best move in it. It checks every move you made, in every one of those games, against what Stockfish would have played instead. It does not care whether the game ended in a win, a loss, or a draw. It just looks for the mistake type that keeps showing up across all of them. Then it names your single most common recurring mistake. This check is free. It takes about 60 seconds. It needs no account. You just connect your Chess.com or Lichess account. Or you can upload a PGN file instead. A PGN file is just a text file. It lists every move made in a game.

That answer covers your whole set of games at once, wins included. You do not have to guess which of your recent wins might have hidden a leak. You do not have to decide on your own which losses were real patterns and which were flukes. The scan treats every game the same way.

If you want the deeper picture, the paid report costs $10, one time. It covers 100 of your games instead of 10. It ranks every recurring mistake pattern you have by severity, using BlackBox's 1-8 damage scale, and by how often it happens. Then it builds a fix course around what it finds across your full set of games, not just the ones you lost.

Other tools exist in this space too. Aimchess offers a free tier along with a paid plan for ongoing game analytics. BlackBox is built to answer one question fast, across every result you have: what is the single mistake actually repeating in your games right now.

Run your last 10 games, wins and losses both, through https://blackboxchess.app/ and see what has been hiding there.

Why wins hide mistakes too

The numbers make this clear. In BlackBox games from players rated 1000 to 1199, a hanging piece showed up in about 82 of every 100 players. Getting forked showed up in about 74 of every 100. At 1200 to 1399, hanging pieces hit about 77 of every 100 and forks about 70 of every 100. Those leaks do not vanish in the games you win. Your opponent just missed the free piece. Review the win and you still find the pattern.