Week 34: seven chess games without a win
Week 34: 7 rapid games between August 28 and 30, 2024, no wins, 6 losses and 1 draw, and my rating fell from 993 to 945. In six of those seven chess games without a win, the engine's winning-chance estimate reached 80% or more at some point. The positions below show how my advantages changed; the recorded terminations give a separate part of the story.
Seven chess games without a win: the week in numbers
| Stat | This week |
|---|---|
| Rating | 993 → 945 (-48) |
| Games | 7 (#829–#835) |
| Wins / losses / draws | 0 / 6 / 1 |
| Blunders (my moves) | 17 (2.4 per game) |
| Average accuracy | 68.2 % |
| Most played openings | Kings Pawn Opening (1), Scotch Game Goring Double Pawn Sacrifice Variation (1), Queens Pawn Opening Zukertort Chigorin Variation (1) |
The accuracy, 68.2 %, was my lowest for weeks, and 17 blunders in 7 games is almost two and a half per game. I played seven different openings in seven games, so there is no single opening to blame. With Black I lost three and drew one; with White I lost all three.
Six advantages in seven games: what the data says
I went through the seven games looking for a common thread, and three things stand out.
First, the blunders came late. Eleven of my 17 blunders happened in the middlegame, after I had already built an advantage. The opening was not the problem; keeping the advantage was.
Second, I stopped counting defenders. In the first two key moments below, a piece of mine was left with one defender, or none, and my opponent found the capture straight away.
Third, I walked away. Five of my six losses are recorded by Chess.com as "game abandoned": I left the game instead of resigning or playing on. One of them, game 831 against CAVMA9 (969), is the strangest of the week: 96.7 % accuracy, zero blunders, a winning chance above 80 %, and still a loss by abandonment. Whatever happened there, it was not the chess.
The only loss that went to the end was game 835, an Englund Gambit against mansaris1408 (975), where I was checkmated after three blunders of my own.
Three positions where the win slipped
Game 829: a knight defended only by my king
With Black against Goldenstate2023 (984), the engine had me at -4.42, "black is winning". My knight was still at home on b8, and I developed it.
The problem was not the square itself but who protected it. After 21...Nd7 White traded rooks with check, 22.Rxf8+ Kxf8, and position.py spells out the consequence: my king's recapture "stops defending the knight on the queen's side, which can now be captured". White's queen took it on d7 and the engine went from -4.42 to 0.00. The pawn move c6 kept "black is winning". You can replay the game here.
Game 830: from +8 to lost in two moves
Against Nasci77 (980) I had a huge attack with Black: rooks on the h-file, a pawn on h2, my queen on g3. The engine's number was -8.11, "black is completely winning".
Instead of bringing my last rook to the h-file, I took a pawn on the other side of the board. Position.py's warning is short: "the knight can now be captured", and White's queen did exactly that. The real disaster came one move later in the game: 28...Qg2+ simply gave my queen away to 29.Kxg2, and the evaluation swung from -3.69 to +8.51. One greedy move, then one panicked move.
Game 832: an early knight on the edge
The third moment is from the opening. Against InyoureMum (955), after 1.e4 e5 2.Bc4 Bc5 3.Qf3, I put my knight on h6.
The engine's punishment is 4.d4: position.py describes it as a pawn move that attacks my bishop on c5 and my pawn on e5 at the same time, and also "opens a line: the bishop that has not moved yet now attacks the knight". The evaluation jumped from -0.12 to +4.63. My opponent did not find it, the game turned wild, and it ended in a draw by repetition after 6 blunders from me. I was winning at one point there too, which is why it counts among the six.
The lesson of the week
A winning position is not a won game. What I need when I'm ahead is a short checklist before every move: which of my pieces is defended only once, and can a check or a trade remove that defender? And when I'm attacking, does my move add a piece to the attack, or take one away from it? In game 829 a single trade removed my knight's only defender. In game 830 a pawn grab pulled my knight out of play and invited the capture. The third lesson isn't chess at all: if I leave the game, I lose it for certain. Staying and playing the position out is the only way a winning chance can still turn into a win.
All games this week
| Game # | Date | Colour | Opponent (rating) | Result | Opening | Accuracy | Link |
|---|---|---|---|---|---|---|---|
| 829 | Aug 28 | Black | Goldenstate2023 (984) | Loss | Kings Pawn Opening | 71.8 % | game |
| 830 | Aug 28 | Black | Nasci77 (980) | Loss | Scotch Game Goring Double Pawn Sacrifice Variation | 52.8 % | game |
| 831 | Aug 28 | White | CAVMA9 (969) | Loss | Queens Pawn Opening Zukertort Chigorin Variation | 96.7 % | game |
| 832 | Aug 29 | Black | InyoureMum (955) | Draw | Bishops Opening Boi Variation | 52.5 % | game |
| 833 | Aug 29 | White | Vajiraal (982) | Loss | Queens Pawn Opening Accelerated London System | 86.4 % | game |
| 834 | Aug 30 | Black | anmol9841 (989) | Loss | Scotch Game | 53.1 % | game |
| 835 | Aug 30 | White | mansaris1408 (975) | Loss | Englund Gambit | 64.1 % | game |
← Week 33 · Week 35 → · Back to the journey · If you want to know where your own wins slip away, review your own games like this.
For a general method beyond this week's games, see the guide to reviewing lost winning positions.
Frequently asked questions
How did seven chess games finish without a win?
The results were six losses and one draw. Five losses were recorded as abandoned, and the key positions show a removed knight defender and a pawn capture that exposed a knight.
How do you know a position was winning?
Every game was analysed with Stockfish. In six of the seven games the engine gave me at least an 80 % chance of winning at some point, yet I won none of them.
What should you do when you are winning?
My games point to two habits: before each move, check which of your pieces is defended only once, and prefer moves that bring another piece into the attack over grabbing a pawn.