Lost by abandonment in a winning position: game 1330

I lost game 1330 by abandonment in a winning position: after White's 31st move the saved evaluation was -7.01, completely winning for me as Black, and there was exactly one move that kept it that way. The game record doesn't show a reply. It says Wipp690012 (1369) won, game abandoned.

The recorded game

I played this game on January 20, 2025 as Black, on my Chess.com account nacsol3. My recorded rating was 1327. It was a ten-minute rapid game without increment, one of the games in the From 631 to 1385 journey.

I call this a collapse, but not because I was winning the whole way. The evaluation swung back and forth many times. Both of us made eight blunders. What makes it "the win I threw away" is the last position: White had just made a blunder, I had a forced winning move, and the game ended there with a loss for me.

Context: two wins, then this

I came in after two wins in a row. Game 1328 was a win with the Pirc against arjuna1001, and game 1329 a win with White against arabina. My recorded rating after game 1329 was 1334. This game took it to 1327, and the next two games were losses too: 1320 after game 1331 and 1312 after game 1332. The weekly diary for week 55 has the full list. That week several of my games ended by abandonment, in both directions. The record doesn't say why, and I won't guess.

The opening: the e-pawn that waited

White played a slow set-up: 1.d4 d6 2.e3 g6 3.h3 Nf6 4.a3 Bg7 5.Nc3 O-O. My first big mistake was 8...c6, a blunder in the record (-0.20 to +1.96). Stockfish wanted 8...e5, the central pawn push. White gave the advantage back at once with 9.Nge2, and the engine asked for e5 again on moves 9 and 10. I played a queen move and a rook move instead. I had spent only about two minutes by then, so this was not time trouble. I just didn't see the centre push as urgent.

Then the game turned my way. White pushed 12.d5, my knight jumped to e5 and took White's bishop on f3 with check, and after 14.gxf3 my bishop took the pawn on h3. White's king had lost two of its pawn shields. The analysis marks 16...exd5 as a good moment of mine: a correct recapture that kept a clear advantage (-1.80).

Position 1: move 19, the pawn that attacks a bishop

White's 19.Bf4 was a blunder, and before my move the evaluation was -2.19, clearly better for me. I played the quiet 19...b6.

. . r q . r k .
p p . . . p b .
. . . p . n p p
. . p P . . . .
. . . . . B . .
P . N . . P N b
. P P . . P . .
R . Q . R . K .

Stockfish wanted 19...g5. The checker describes it in one line: the pawn attacks the bishop on f4. The engine's line is 20.Bd2 Nh7 21.f4 gxf4 22.Bxf4, keeping Black clearly better. After my pawn move to b6 the evaluation was 0.00. I had 281 seconds left.

Position 2: move 20, taking towards my own king

White's bishop took my pawn on h6 (20.Bxh6). That bishop was now attacking my bishop on g7. I took it.

. . r q . r k .
p . . . . p b .
. p . p . n p B
. . p P . . . .
. . . . . . . .
P . N . . P N b
. P P . . P . .
R . Q . R . K .

My capture looked natural: a bishop for a bishop, and my bishop even attacked White's queen. But the checker adds "they can take the bishop back", and White did, with the queen. After 21.Qxh6 White's queen stood right next to my king. The evaluation went from -0.10 to +4.36, better for White.

Stockfish preferred 20...Nh7, which keeps my bishop on g7 and lets White take it there instead: 21.Bxg7 Kxg7, and my king, not White's queen, ends up on g7. The difference is where the enemy queen lands. I'll remember that a trade is not equal if it brings their queen next to my king.

Position 3: move 22, the same knight move again

White then made its own blunder, 22.Re3, and the evaluation was down to +2.56. White was still better, but much less so. I played 22...Re8, attacking White's rook on e3.

. . r q . r k .
p . . b . p . .
. p . p . n p Q
. . p P . . . .
. . . . . . . .
P . N . R P N .
. P P . . P . .
R . . . . . K .

The engine's first choice was the same knight move as two moves earlier, 22...Nh7. In my check it keeps White only clearly better (+2.80). The other moves it looked at, a bishop move to g4 and a rook move to c7, were already "white is winning". My rook move scored +6.48 in the saved analysis. The knight on h7 blocks the queen's path and guards the squares around my king. Twice in three moves the defensive knight step was the move, and twice I chose a different one.

Position 4: move 29, the turning point

The game went on with more swings. White let a big advantage go on move 28 (from +4.01 to +0.42; the record's best move there was Qxe8+), and the rooks came off. Before my 29th move we had the queens, White's two knights against my knight and bishop, and pawns, and the evaluation was +0.41, about equal. I had about two minutes.

. . . . . . k .
p . q b . p . n
. p . p . . . .
. . p P . . p .
. . . . . . . .
P . N . Q P N .
. P P . . P K .
. . . . . . . .

I pushed my f-pawn two squares. That was the move the analysis picks as the turning point: from +0.41 to +6.43. Stockfish wanted 29...Qd8, a quiet queen move; in my check its line stays about equal. My pawn push left my king with one less pawn in front of it, and after 30.Qe7 White's queen was inside my position. I pushed the same pawn again (30...f4, attacking White's knight on g3) with 108 seconds left.

Position 5: move 31, the move I never played

Then came White's last move, 31.Nh5. The checker describes it as a knight move that threatens checkmate on the next move. It was also a blunder: from +7.58 to -7.01. Now I was the one completely winning, with a single move.

. . . . . . k .
p . q b Q . . n
. p . p . . . .
. . p P . . p N
. . . . . p . .
P . N . . P . .
. P P . . P K .
. . . . . . . .

The winning move is 31...Bh3+. My bishop on d7 was standing between my queen on c7 and White's queen on e7. Moving it away with check is a discovered attack: one piece moves and uncovers an attack by another piece behind it. White has to deal with the check first, and then my queen takes White's queen. The engine line is 32.Kxh3 Qxe7, with Black completely winning. The checker also notes that the bishop can be captured, "but the line shows why it still works".

The engine's other candidates show how narrow this was. A bishop move to g4 scored +7.93 for White, and a bishop move to e6 +10.08. Only the check won.

There is no 31st move of mine in the record. The game ended there, and the result is a loss by abandonment.

The lesson of game 1330

Two habits come out of this game, and neither one is about the abandonment itself, which the record can't explain.

First, when the enemy queen comes close to my king, the defending move is often a quiet knight step (Nh7 on moves 20 and 22). I kept choosing captures and rook attacks instead, and both times the engine said the quiet move was better.

Second, when my opponent makes a threat, I look at my checks first. In the last position White was threatening checkmate, and the only answer was a check that also uncovered an attack on the queen. Checks, captures and threats, in that order, is the simple routine that finds such moves. The answer on why winning positions get lost goes into this pattern more broadly.

Where this sits in the journey

The recorded ratings are 1334 for game 1329, 1327 for this game and 1312 for game 1332. They describe a short slide, not a cause. If you want to see your own games the same way, with each key moment on a board and arrows for the move played and the better one, you can use Chess Coach in Claude or ChatGPT. The guide to reading Stockfish analysis explains the plus and minus signs, and the journey hub has the rest of the series.