What your games say
Written by Claude from 198 rapid games (Oct 6 to Oct 16, 2026), the Stockfish analysis, and the Building Habits checker. Numbers use the chess.com-style classification. Regenerate by asking Claude to update the report.
The headline
You lose to material, not to chess. Across 5,686 moves, the habit checker found 298 moves that hung a piece, 278 that left an already-hanging piece on the board, 185 that skipped a free piece, and 74 captures that lost material to the recapture sequence. Those four categories account for the large majority of your costly moves. There is nothing positional to fix yet. The whole job is: before every move, see what can be taken.
Three numbers to keep in mind:
| Games with zero blunders | 84 of 198, and you won 62% of them |
| Games with two or more blunders | 47, and you won 23% of them |
| Losses from a position you were clearly winning (75%+) | 40 of your 88 losses |
Pattern 1: thinking longer makes it worse
This is the most surprising thing in the data and the most actionable.
| Time spent on the move | Mistakes, misses and blunders per 100 moves |
|---|---|
| under 2 seconds | 6.2 |
| 2 to 5 seconds | 7.7 |
| 5 to 10 seconds | 14.1 |
| 10 to 30 seconds | 22.6 |
| 30 seconds or more | 36.2 |
Clock pressure is not the cause. With more than five minutes left you err at 16.5 per 100; under 30 seconds it is 19.5. The long thinks happen in positions where you don't have a method, so you calculate variations instead of checking the basics, and then pick a move that hangs something. Your handful of 15|10 games show the same thing: more time, slightly worse play.
What to do: cap yourself at about 20 seconds. If you are still unsure, do not calculate deeper. Run the scan below and play the safest move it allows. Deep calculation is not where your rating is right now.
Pattern 2: you get winning positions and give them back
Opponents hand you roughly four serious errors per game. You convert 49% of them. The result is 445 "misses" (dropping a chunk of a winning position) against 185 blunders. Misses are your biggest error class by far, and they cluster in the middlegame: 347 of the 445.
The move-number curve shows exactly where. Moves 1 to 5 are clean (3 bad per 100). Moves 10 to 20 are the danger zone (23 to 25 per 100). After move 30 you are back under 12. The collapse happens right when the opening book runs out and the position is still full of pieces.
What to do: when you are ahead, your only goals are trade pieces and don't hang anything. The data backs the Chessbrah "always trade" rule for you specifically: when an even trade was available and you played something else, that other move was a mistake or worse 35% of the time, against 16% for your moves in general. Turn on the "Trade habit" chip in Play and practice taking the trades.
Pattern 3: mate in one, both directions
You missed a mate in one 85 times and allowed one 93 times. Thirty-nine of your losses ended in checkmate, and your king had no escape square on the back rank in 234 positions. Chess.com's "Great" label on your games is mostly you finding forced mates late, so you can see them; you are just not looking every move.
What to do: two questions, every move, in this order: "Do I have a check that mates?" and "If they check me, where does my king go?" The second one also covers luft: when you have a free move in the middlegame and your king is on the back rank, push the h-pawn one square.
Pattern 4: the opening that hurts you
| Opening (5+ games) | Record | Bad moves per 100 |
|---|---|---|
| Englund Gambit (as Black, after 2.dxe5) | 3-0-4 | 24.5 |
| Scandinavian, Mieses-Kotrc (3...Qa5 line) | 0-1-4 | 14.4 |
| Scandinavian Defense (other lines) | 6-2-5 | 17.7 |
| French Defense | 4-0-4 | 11.8 |
The Englund is a gambit: it gives up a pawn for tactical chaos, which is exactly the kind of position where you err. The 3...Qa5 Scandinavian puts your queen out early and you are 0-4 in it. Consider meeting 1.e4 and 1.d4 with the same simple setup each time (pawn to e5 or e6, knights before bishops, castle by move 8) and let the opponent be the one who has to find ideas. Your "castle late" count is 70, so the castling habit is worth drilling as well.
The scan, written out
Before every move, in this order, no exceptions. It takes ten seconds once it is a habit.
- Their last move. What does it attack? What did it stop defending?
- My checks. Is any check mate, or does it win a piece?
- My captures. Can I take something for free? Count attackers and defenders by value.
- Their captures after my move. For the move I want to play: can they take something for free afterwards?
- If nothing wins and nothing hangs: develop, castle, or trade.
This week
- Before you play, do ten "Punish" puzzles and five "Don't blunder" puzzles from your own games. Say the scan out loud on each.
- Play two games a day against Maia 400 or 600 with the blunder check on. Goal: a game with zero hard-habit hits. Take back every flagged move and find the fix.
- After each real session, open your losses and look at the first miss, not the first blunder. That is the move where the game was actually lost.
- Metric to watch on the Stats page: mistakes, misses and blunders per 100 moves, rolling 20 games. You are at about 15. Get it under 12 and the rating follows. Punish rate is the second metric: you are at 49% and it has slipped from 51% in your first 50 games to 46% in your last 50.
Trend so far
Over the ten days, bad moves per 100 fell from 17.6 in your first 50 games to 15.2 in your last 50. That is real improvement. The punish rate went the other way. Both will show up as the same thing in games: positions you were winning that you did not finish.