You make fewer big mistakes than your opponents. Your Black openings are where you give ground.
79% of your blitz games are 3+0. Over the last 90 days your rating went from 2913 to 2893, and your accuracy sat at 88.3, where it was before. You already make fewer big mistakes than the people you play: 0.95 per 40 moves against their 1.08. Only one leak costs you more than it costs them, and it happens in the first 10 moves as Black.
The short answer
Analysed: GothamChess’s chess.com blitz games: 19,491 games · Stockfish 19 on the latest 1,996 (Sep 24 2024 – Sep 24 2026). Report date: .
Verdict: Steadier, not stronger yet. Last 90 days against the prior 9 months: your big mistakes are down 6%, from 1.01 to 0.95 per 40 moves (p = 0.02), and you punish 78% of your opponents’ blunders, up from 72%. That part is real. Accuracy (88.3 vs 88.5, p = 0.48) and score (56.7% vs 58.3%, p = 0.57) have not moved. Wins from lost positions: 17% vs 15%, so luck isn’t hiding anything either way.
Biggest leaks:
- Openings that start you in a hole: White: +0.3 by move 10 · Black: −0.6 · ≈31 rating points, 8 more than your opponents
- Endgames: the last point: 621 endgame mistakes · 52% in rook+minor endings · ≈9 rating points, 3 fewer than your opponents
- Not taking what’s offered: 566 missed punishments · 47 free pieces · 2 mates · ≈42 rating points, 6 fewer than your opponents
Free Goat Card for GothamChess · How to review your own games for free · Time pressure in chess, measured
Your game at a glance
- Openings2,800▼ 40
- Tactics3,020▲ 40
- Strategy2,840▼ 20
- Endgames2,910▲ 10
- Brilliant moves209in 1,996 games
Each number is the rating of a player who plays that part of the game the way you do: your opponents’ average rating (2,850), adjusted by how much win-% you lose per move there compared with them, in the same games. Arrows: the change against your opponents compared with Prior 9 mo; grey means it is within the noise. A brilliant move is the engine’s top choice that gives up at least two points of material.
Improving, or lucky?
| Metric | Last 30d | Last 90d | Prior 9 mo | Read |
|---|---|---|---|---|
| Results | ||||
| Games | 360 | 747 | 454 | |
| Rating, start → end | 2917 → 2893 | 2913 → 2893 | 2855 → 2920 | flat |
| Score | 56.9% | 56.7% | 58.3% | flat (p = 0.57) |
| Skill: what the engine sees | ||||
| Your accuracy | 88.6 | 88.3 | 88.5 | flat (p = 0.48) |
| Opponents’ accuracy | 88.1 | 88.0 | 88.1 | flat |
| Big mistakes per 40 moves | 0.92 | 0.95 | 1.01 | down 6% (p = 0.02) |
| Opponent blunders you punish | 78% | 78% | 72% | better |
| More accurate than opponent | 54% | 54% | 54% | flat |
| Eval at move 10 as White | +0.4 | +0.2 | +0.3 | flat |
| Eval at move 10 as Black | −0.7 | −0.7 | −0.4 | flat |
| Conversion & luck | ||||
| Score after reaching +5 | 92.9% | 92.6% | 89.9% | flat |
| Score after being −5 or worse | 22.1% | 24.2% | 24.9% | flat |
| Losses that were +5 at some point | 6% | 7% | 12% | better |
| Wins that came from a lost position | 14% | 17% | 15% | flat |
| Wins on the clock while losing on the board | 2% | 3% | 6% | less luck |
| Losses on time | 15% | 18% | 13% | worse |
Rating, rolling 50-game average
Rolling 50-game average: 2815 → 2919, with a peak of 2983 in March 2026.Accuracy, rolling 50-game average
The two lines have moved together for two years. Now: you 88.3, opponents 88.8. Your edge isn’t average accuracy. It is fewer big mistakes.Your clock: wins vs losses
Time left on your clock after your move, in the games you won and the games you lost.
- Move 20
Two seconds apart. Your 3+0 losses are not decided by the clock at move 20. Medians over 796 wins and 535 losses.
- Move 30
Six seconds apart, and just over a minute either way. Every second you save before move 20 is one you get to spend in the endgame. Medians over 648 wins and 399 losses.
Fewer mistakes hasn’t moved the rating because nothing else moved with it. The one place you give up more than your opponents is the start as Black: −0.7 by move 10 in the last 90 days, from −0.4 before. Start there.
What you do well
Not everything is a leak. These parts of your game already work, measured the same way as the leaks below: against your opponents, in the same games.
- Tactics: 3,020170 above your opponents’ average of 2,850: you see and play tactics better than players at your level.
- Tactical defenceShort combinations cost you less than they cost your opponents: ≈68 points, against ≈80.
- Few pieces hungYou hang pieces less often than your opponents: one-move oversights cost you ≈24 rating points, them ≈35.
- Closing out winsYou close out won games better than your opponents do: ≈22 points lost from winning positions, against ≈29.
- Punishing mistakesYou cash in on their mistakes more often than they cash in on yours: ≈42 points lost there, against ≈48.
Your best moves
6 leaks, ranked by rating points lost beyond your opponents
Six leaks, ranked by what each one costs you beyond what the same leak costs similar-rated opponents in the same games. Only the first, openings, costs you more than it costs them: about 8 rating points. The other five you already handle better than your opponents; they are ranked by how thin that margin is. Every position below is from one of your own recent games. Decide on your move first. Then open the answer and step through what you played and what was best with the ← → buttons (or your arrow keys).
1 · Openings that start you in a hole
White: +0.3 by move 10 · Black: −0.6 · ≈31 rating points, 8 more than your opponentsYou should keep in mind no names, nor numbers, nor isolated incidents, not even results, but only methods.
This is the one leak where you give up more than your opponents: about 8 rating points more. By move 10 you are +0.3 as White and −0.6 as Black. In 14% of your Black games you are already 1.5 pawns down. The worst are the quiet White setups: −1.5 by move 10 against the Queen’s Pawn Opening (22 games) and −1.1 against the King’s Pawn Opening (15). You still score 68% and 63% there, because you outplay people later. You just start behind. Three of the four positions below turn on one loose piece.
- Why it happens
- At your level the ideas are known, and the blunders are rare: your first big mistake came before move 10 in only 6 of 1,996 games. The deficit is in the lines. 1…Na6 against 1.d4: −1.5 by move 10. The Latvian 2…f5: −1.1. Owen’s 1…b6: −0.9. The De Bruycker and the Accelerated Queen’s Indian: −0.7 each. The positions below come from the French Rubinstein, the Queen’s Gambit Accepted and the Göring Gambit.
- The habit
- As Black, one prepared answer to each first move, and play it. Out of book before move 10? You have 2:41 on your clock at move 10 on average. Spend 10 seconds there, on what the last move left loose.
- The drill
- Take your last 10 games in each line above. Find the first move where the eval dropped. Build the fix with an engine and the master database, write it to move 12–15, and drill it with spaced repetition. Then play the position at move 10 against Stockfish at 3+0 until it feels routine.
- What success looks like
- Average eval at move 10 as Black at or above −0.3 (now −0.6).
- Lichess Analysis board + Opening ExplorerpracticeEngine and master database side by side. Check each of your Black lines to move 12.
Practice this: Opening rules that win games →
2 · Endgames: the last point
621 endgame mistakes · 52% in rook+minor endings · ≈9 rating points, 3 fewer than your opponentsIn order to improve your game you must study the endgame before anything else; for, whereas the endings can be studied and mastered by themselves, the middlegame and the opening must be studied in relation to the endgame.
You already give up less here than your opponents, about 3 rating points less, so this is refinement, not repair. 51% of your games reach an endgame. From a winning endgame you score 93%. 52% of your endgame mistakes come in rook+minor endings. You lost 5 winning endgames on time. Three of the four positions below come down to a loose piece or a free capture, and the clock was short in all four.
- Why it happens
- Not a theory gap. You spend a median 0.9 seconds a move in the endgame, and 52% of your endgame mistakes come in rook+minor endings, where technique is judgement, not tablebase knowledge. You also lost 5 winning endgames on time.
- The habit
- When the queens come off, take one think: where does each king go, and which minor piece do I keep? Then play fast on that plan.
- The drill
- Twice a week, one chapter of Shereshevsky’s Endgame Strategy. Then take a rook+minor ending from your own games and play it out against Stockfish at 1+1, from both sides. Use Dvoretsky’s Endgame Manual when a game reaches a theoretical ending.
- What success looks like
- Convert at least 95% of your winning endgames (now 93%).
Practice this: Converting a won game: trade pieces, not pawns →
3 · Not taking what’s offered
566 missed punishments · 47 free pieces · 2 mates · ≈42 rating points, 6 fewer than your opponentsI have a weakness for any piece in excess of my opponent's numbers - from pawn to queen.
You punish 78% of your opponents’ blunders, up from 72%, and fewer slip past you than past them: about 6 rating points’ worth. Still, 566 went unpunished, including 47 free pieces and 2 mates. In 53% of the misses the winning move was quiet: an attack on a piece with nowhere to go. Three of the four below came with under a minute on your clock.
- Why it happens
- Three of the four misses below came with under a minute left. In one you gave a check instead of taking a free queen. In another you saved your queen instead of starting a mate. The win was one move away both times.
- The habit
- After every opponent move, ask what it changed. Your forcing moves first, then the quiet move that attacks a piece with no squares.
- The drill
- Calculation, not pattern reps. Three sessions a week from Aagaard’s Grandmaster Preparation: Calculation: solve on a board, write the lines, then check. Once a week, set up your own misses (the four below first) and solve each with 30 seconds on the clock.
- What success looks like
- Punish at least 85% of your opponents’ blunders (now 78%).
Practice this: Is it really safe? Counting attackers and defenders →
4 · Giving back won games
Reached +5 in 1,084 games · lost 72 · drew 46 · ≈22 rating points, 7 fewer than your opponentsFirst restrain, next blockade, lastly destroy.
You reached +5 in 1,084 games and lost only 72 of them. In the last 90 days you score 92.6% from +5, up from 89.9%, and you already let fewer won games go than your opponents: about 7 rating points’ worth. 32% of your big mistakes still come from winning positions, at a median of 0:37 on the clock. In all four games below you were +7 or better with 19 seconds or less, and each one ended at 0.0.
- Why it happens
- With a median 0:37 left when you blunder a winning position, there is time for one look, not a calculation. In three of the four games below that look went to a check, when a capture or a promotion was there: …Qxa1, h8=Q, …Qxb3.
- The habit
- When you are +5 and short of time, look for the capture or the promotion before the check. If there is neither, pick the move that takes away their checks.
- The drill
- Set up the positions below and the rest of your 72 games lost from +5. Play each out against Stockfish with the clock you had, and win it twice. For each game, note the first move after +5 that let the eval drop, and whether it was a check.
- What success looks like
- Keep your score after reaching +5 at 94% or better (now 93%).
Practice this: Converting a won game: trade pieces, not pawns →
5 · One-move oversights
≈231 mistakes · 11% of all · ≈24 rating points, 11 fewer than your opponentsPlaying Real Chess (when you DO consistently analyze to see if each candidate move is safe) will not make you a good player, but playing Hope Chess will prevent you from becoming one.
You make fewer of these than your opponents, about 11 rating points’ worth fewer. Still, 129 times a piece went to a square where it could be taken, and 85 times an attacked piece stayed where it was. The clock is the pattern: 1.3 big mistakes per 100 moves with more than 90 seconds left, 4.9 with under 10 seconds. Three of the four below came with 51 seconds or less.
- Why it happens
- These are not calculation errors. At 0:05 and 0:07 there is no time to check anything. Big mistakes run 1.3 per 100 moves with more than 90 seconds left and 4.9 under 10 seconds. The fix is to reach the end of the game with more time, not to think harder with none.
- The habit
- Under 20 seconds, stop calculating. Play moves that keep every piece defended and the king covered. Save the long think for the one critical middlegame moment.
- The drill
- Once a week, review your games decided under 0:10 and your losses on time. Find the moves where you spent more than 15 seconds. Was that the critical moment? You want the long thinks there, not on routine moves.
- What success looks like
- Big mistakes at or below 0.8 per 40 moves (now 0.95), measured on your next 50 games.
Practice this: The blunder check →
6 · Two-move tactics against you
≈905 mistakes punished by a short combination · ≈68 rating points, 12 fewer than your opponentsAfter certain particular dispositions of pieces have proved his undoing, the beginner will develop the perception of threats.
Your opponents fall for these more than you do, about 12 rating points’ worth more. Still, about 905 of your mistakes were punished by a short combination: capture sequences on under-defended pieces (283), attacks on pieces with no retreat (245), checks (200) and forks (78). In each position below, one forcing reply was left in, and you had 1:46 or less on the clock every time.
- Why it happens
- Each one starts with a forcing reply you didn’t check: …Nf7 hitting your queen, …Bd6+ skewering king and rook, Rd8+ on your back rank. They are short, and they come fast: 1:46 or less on your clock in all four.
- The habit
- Before you commit, list their forcing replies: checks, captures, and knight jumps near your king or queen. If one wins something, pick another move.
- The drill
- Aagaard’s Grandmaster Preparation: Calculation: before each exercise, write the opponent’s forcing replies first, then your candidates. Once a week, replay the positions below against Stockfish at 3+0 from the move before the mistake.
- What success looks like
- At the next audit, no more than 29% of your big mistakes come from a short combination (now 44%).
Practice this: Forks: one piece, two targets →
The plan
Four rules, starting today
- Black: fewer lines, prepared deeper. You are +0.3 by move 10 as White and −0.6 as Black. This is the only leak that costs you more than your opponents. Weeks 1–2 are the centrepiece.
- Your own games are the syllabus. Every drill below starts from a position you reached: the move-10 positions, the 72 games lost from +5, the rook+minor endings.
- +5 and short of time: capture or promote before you check. Median clock when you blunder a winning position: 0:37. That is one look. Point it at the capture, the promotion, or the move that takes away their checks.
- Spend the clock in the middlegame, keep some for the ending. Big mistakes: 1.3 per 100 moves with more than 90 seconds left, 4.9 under 10 seconds. At move 30 you have 1:11 in wins, 1:05 in losses.
Before every move
- Black, moves 1–10: the line you prepared.
- Their move: what forcing reply does it leave them? Then my forcing moves.
- +5 and under 0:40: capture, promote, or take away their checks.
- Queens off: one think for the plan, then play fast.
Week by week
Your Black openings, fixed at move 10
Your French Rubinstein, move 9. You were +6.2. Your c6-knight is pinned. What ends the pin?
- Pick the lines, worst first. 1…Na6 against 1.d4: −1.5 by move 10 (22 games). The Latvian, 1.e4 e5 2.Nf3 f5: −1.1 (15). Owen’s 1…b6: −0.9 (21). The De Bruycker, …Na6 and …c6: −0.7 (41, 48% score). The Accelerated Queen’s Indian, 2.c4 b6: −0.7 (19). For each one, decide: repair it, or replace it with a main line.
- Find the exact move. Pull your games in each line and find the first move where the eval dropped. It is usually the same move: 1…Na6 against 1.d4 cost 8.8% win chance on average, in 39 games. 2…f5 cost 11.2%, in 14.
- Build the fix with an engine and a database. Engine first, for the move. Then the master and top-blitz games from that position, for the plans that follow. Write each line to move 12–15 in a repertoire file and drill it with spaced repetition.
- Check your main line too. The French Rubinstein is your most-played Black opening (67 games, −0.3). Two of the positions above come from it. Prepare the 5.Bd3 and 5…gxf6 structures to the same depth.
- Spar the move-10 positions. Play each prepared position against Stockfish at 3+0, from both sides, until the middlegame that follows feels routine.
- Test in 20 rated 3+0 games as Black (your Black score: 53%). Play them early in a session: in your last 2,000 games you scored 60% in games 1–3 and 56% after that.
Won positions with under a minute
Your game, move 45. You were +4.9. What was the threat?
- Spar your own won positions. Set up the three above and your other +5 games that you did not win (72 lost, 46 drawn). Play each out against Stockfish with the clock you had in the game. Win each twice before you move on.
- Review the 72. For each, write down the first move after +5 that let the eval drop, and whether it was a check. In three of the four games in Leak 4 it was.
- In games: +5 and under 0:40, play the move that takes away their checks or trades queens before you go for more.
Calculation: their forcing reply first
Your game, move 22. You were +2.4. Where can the e5-knight go, and what does it hit?
- Three sessions a week, 30 minutes: Jacob Aagaard’s Grandmaster Preparation: Calculation. Solve on a board without moving the pieces, write the full lines down, then check with the engine.
- Opponent’s forcing replies before your own candidates. About 905 of your mistakes were punished by a short combination (Leak 6). In 53% of the wins you missed (Leak 3), the winning move was quiet.
- Once a week, your own positions. Set up 10 from Leaks 3 and 6, starting with the two above. Solve each with 30 seconds on the clock, then play it out against Stockfish at 3+0.
Technical wins at speed
Your game, move 61. You had mate in 10, with 0:00 on your clock. What is the simplest answer?
- Know the method before the clock runs. Queen against rook and queen-and-pawn races ended at 0.0 in two of the Leak 4 games. Look up each ending in Dvoretsky’s Endgame Manual, then play the positions above against Stockfish until you win each inside 30 seconds.
- Trade into the simplest win, not the biggest eval. From +5, the move that removes their counterplay beats the move that adds a pawn.
- Replay your 46 draws from +5. Note the move where the win became a draw and the clock when it happened.
Rook and minor-piece endings
Your game, move 27. You were +2.9. The d4-knight is free. What does the h8-rook get if you take it?
- Rook+minor endings: 52% of your endgame mistakes. Read Lev Shereshevsky’s Endgame Strategy, one chapter twice a week: king activity, the second weakness, not hurrying.
- Spar five of your own rook+minor endings that were equal or better. Play each against Stockfish at 1+1, from both sides.
- Clock: you spend a median 0.9 seconds a move in endgames and lost 5 winning endgames on time. When the queens come off, take one think for the plan, then play fast on it.
Audit again
- Play 60 rated 3+0 games, half of them as Black, in the lines from Weeks 1–2.
- Re-run this analysis on them. You pass if big mistakes are ≤ 0.8 per 40 moves (now 0.95) and you score 94% from +5. For the openings: eval at move 10 as Black ≥ −0.3 (now −0.6).
- If you don’t pass, repeat Weeks 1–2 with the lines that still come out below −0.3.
Games: the latest 1,996 blitz games from the chess.com public API (Sep 24 2024 – Sep 24 2026). Engine: Stockfish 19, 50,000 nodes per position; puzzle lines at 3,000,000. Win% model and accuracy formula: lichess. A big mistake loses ≥ 20 win-% in one move. Rating-point costs: 695 Elo per point of expected score, counting only mistakes the opponent cashed in. Trends: Welch z-test against Prior 9 mo.