Opening check

See How Many of Your Games Were Already Worse by Move 10

Pick your handles, your time controls and how far into the game to look. Stockfish replays the opening of every game that matches, in your browser, and reports where you stood at the checkpoint, how those games finished, and which of your own moves cost the most getting there.

Free Runs in your browser · Chess.com and Lichess

  • No Chess.com or Lichess password — a public username is all it needs.
  • The learning platform is free — AI Move Explanation and Meta-Coach are an optional membership.
  • Every recommendation comes from your own games and Stockfish analysis.
An opening check over 50 games: "You reached move 10 worse in 13 of 50 games", three figures — 13 of 50 worse, a median evaluation of +0.17, and a game score of 27% from the worse starts against 57% from the rest — and the largest single error, 6...Bf5, with buttons to review the position or search it deeper.
  • The finding is a count, not a diagnosis: how many games, and what you scored from each side of the line.
  • The two scores are shown together because they are the comparison — 27% from the games that started badly, 57% from the rest.
  • Then it hands you one position to open, with the move the engine wanted and what the move played cost.

It Runs on the Games You Choose

The check lives in the My Games tab of Opening Lab, under the same filters as everything else there. Whichever handles, colours and time controls are selected are the games it reviews — and it says so on the report, so a number can always be traced back to a set of games.

The My Games tab in Opening Lab: an Opening check strip reading "50 games · 12 reached move 10 worse" with a button to review the report, above the user’s own first-move statistics from 184 matching games.
  • The strip keeps the last check’s finding and the way back into the report.
  • Under it, the same tab’s ordinary job: what you play from this position and how those games went.
  • Change the filters and the strip says the report no longer matches them, rather than quietly implying it does.

Four Things It Measures, and What Each One Counts

Every figure names its own denominator, because "your opening accuracy is 87%" is not something a player can act on and "thirteen of fifty games reached move ten worse, and you scored 27% from them" is.

Where you stood at the checkpoint

Pick the move number — eight, ten, twelve or fifteen. Every game is placed by its evaluation there, from your side, and the games are banded: better, level, worse, difficult. It is a checkpoint you chose, not a claim about where your opening ends.

What you scored from each side of it

Your game score from the games that reached the checkpoint worse, beside your score from the ones that did not. Both carry the number of games behind them, and neither is described as the cause of the other.

Which of your moves cost the most

Only your own moves are judged — an opponent’s mistake is not a fact about your preparation. Each one shows the evaluation before and after, the move Stockfish wanted, and what it cost in points of expected result.

The mistakes you make more than once

The same move from the same position across several games, with what it cost each time and what you scored from those games. A bad move played once is bad luck; the repeated one is the one certain to come up again.

Which Openings, and Where in the Move Order

Openings are grouped with the colour you had, because the Sicilian you play as Black and the one you meet as White are different openings that share a name. They are ordered by how many games started badly — not by how bad the single worst one was.

The openings table from an opening check: each opening with the colour it was played as, how many games, how many of them reached move 10 worse, the median evaluation and the game score — above two charts showing where the games stood at move 10 and the inaccuracies, mistakes and blunders by move number.
  • A two-game row is marked as a small sample rather than presented as a pattern.
  • The bar chart counts only the moves that cost something, by the move number you played them on — click one to list them.
  • Both charts and the table are the evidence for the finding at the top of the report; nothing in it is stated without them.

From a Finding to Something You Can Practise

A report you cannot act on is a report you read once. Every row of this one opens on the board, with the engine and both opening databases already answering about that position.

  1. Open the position

    The board goes to the position as you faced it — one move before the mistake, the right way up — and asks for your move. Nothing is revealed until you ask for it.

  2. Compare, then check it deeper

    Show the move you played, or the one the engine wanted, and read the difference. One move can be re-searched on its own at a depth no survey of several hundred positions could afford.

  3. Keep the corrected line

    Save what you chose as a preparation. It joins spaced-repetition practice and can be sparred against, so the fix is rehearsed rather than noted.

What It Does Not Tell You

It does not tell you that your opening lost you those games. What it measures is where a game stood at one move number and how that game finished — and those two things being related is not evidence that one caused the other. An early tactic, the strength of the opponent, the clock and the next thirty moves are all still candidates, and none of them is visible from here.

So the report reports. It says how many games reached the checkpoint worse and what you scored from them, and then it points at specific moves you can check yourself on a board. Deciding whether the repertoire is the problem is your call, and the evidence for it is on the page rather than compressed into a verdict.

The evaluations behind it are a survey depth, picked so several hundred positions fit into a few minutes on your own machine. Individual moves can and do change when they are searched further, which is why any single move in the report can be re-checked deeper before you act on it.

Frequently Asked Questions

What does "worse at move 10" actually mean?

That the engine evaluated the position at that move number at worse than −0.60 from your side — roughly the point at which a position stops being a normal game to play and starts being one you have to defend. The move number is the checkpoint you picked when you started the check, and the report never calls it the end of your opening; it is the line the games are counted on either side of.

Does it mean my opening is what is costing me games?

No, and the report will not say that. It can tell you that thirteen of fifty games reached move ten worse and that you scored 27% from those against 57% from the rest. It cannot tell you whether the opening choice, one tactic on move nine, the opponent or everything after move twenty produced that gap, and it does not pretend to. What it gives you instead is the list of specific moves behind those positions, each of which you can check on a board.

How many games can it check, and how long does it take?

Ten, twenty-five, fifty or a hundred of your most recent matching games, at a checkpoint from move eight to move fifteen. The cost is a Stockfish search per distinct position, and games in one repertoire share their early moves — so a fifty-game check is a few hundred searches rather than a thousand. The dialog counts them exactly and estimates the time before you start, and the run then measures its own speed and tells you what is left. You can stop it at any point and keep every game it has finished.

Does it use credits?

No. The whole check runs in your browser on Stockfish 18 NNUE — nothing is sent to a server and nothing is charged. The two paid features on ChessLogix are the ones that call a language model on every use; this is not one of them.

Do I need to import my games first?

Yes, because it is a report about your own games. Importing takes a public Chess.com or Lichess username and no password. Once games are in, the check reads whichever of them the filters select — one handle or several, one colour, one time control, or all of it.

I play blitz and rapid under different accounts. Can it keep them apart?

That is what the filters are for, and it is worth doing. A line that holds at 15+10 can be unplayable at 3+0, and a serious account and a casual one produce different evidence about the same repertoire. Pick the handles and the time controls first; the report states which ones it was run over, and tells you when the filters have moved since.

How deep does it search, and can I trust one move’s number?

Depth is yours to choose, from 10 to 20, and the default trades depth for covering more games. Treat the shape of the report — how many games, which openings, which move numbers — as the reliable part, and any individual move as provisional until you deep-check it, which searches that one position several plies further and updates the card with what it finds.

Check the Openings You Actually Played Last Month

Pick the handle, the time controls and the checkpoint, and the check reports where those games stood and which of your moves cost the most. It runs on your own machine, it is free, and you can stop it at any point and keep what it has already found.

Where to Go Next