Week 49: missing forcing moves in winning positions

Week 49 covers December 9–15, 2024: 18 rapid games, 10 wins and 8 losses, with my rating moving from 1099 to 1120 (+21). Six of those eight losses came after the engine's winning-chance estimate reached at least 80%. This entry looks at missing forcing moves in winning positions, through three decisions from that week.

Missing forcing moves in winning positions: the week in numbers

The split between wins and losses is striking. My average accuracy in the ten wins was 83.2%; in the eight losses it was 66.8%. The middle of the week was the best part: six wins in a row between games 1173 and 1178, with accuracies of 95.8, 83.9, 81.4, 92.9, 80.4 and 93.3. Right after that came four losses in a row, and three of them (games 1180, 1181 and 1182) were among the six "winning positions" I let go. Of my 44 blunders, 32 happened in the middlegame and only 6 with less than a minute on the clock, so I can't blame time trouble. Five of the eight losses ended with Chess.com marking the game as abandoned rather than finished on the board.

Three moments where the win was right there

All three positions below come from the first three games of the week, and all three are what the engine calls a blunder (a move that changes the evaluation by a lot).

Game 1169: a knight fork I didn't see

I had White against fredbaho (1091) in a Queen's Pawn Opening, Zukertort Variation (see the game on Chess.com). The engine rated the position +5.57 for me, which means I was clearly winning.

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

I played Nf3, a normal-looking move that attacks a loose pawn. The engine wanted Nd6+, a fork: one piece attacking two things at once. The knight gives check, attacks the rook on c8, and by leaving c4 it also lets my queen on c2 look at that rook. The engine line is 18.Nd6+ Ke7 19.Nxc8+ Qxc8: I win a rook for a knight. After my Nf3 the evaluation dropped from +5.57 to −0.04, basically equal. I was lucky: my opponent answered 18...e4, the same knight check worked one move later, and I won by resignation.

Game 1171: the capture with a threat

With Black against kaunghtat (1099) in a Pirc Defense (game link), I was the one with the big advantage: −4.82, winning for Black.

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

Nxe4 grabs a pawn and attacks the white queen at the same time, so White has no time to fix anything. The engine line continues 15.Qd3 Bxb2 16.Na3 Bxa1: once my knight leaves f6, the bishop on g7 sees all the way down the long diagonal, takes b2 and then the rook in the corner, which has no defender. I played the quiet Nd7 and the evaluation fell to −0.95. My very next move, 15...Ne5, was another blunder, 19...Nf3+ (+0.20 to +5.83) was the last one, and the game ended as abandoned. Peak winning chance in that game: 85.5%. Final result: zero.

Game 1170: a pin that left my knight loose

This one is an opening mistake rather than a missed win, but it belongs here because it shows the same blind spot. With Black against darkside_dv (1121) in the Modern Defense (game), the position was −1.35, slightly better for me.

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

Bg4 creates a pin: the knight on f3 can't move without exposing the queen behind it. It looks active. But the engine's answer for White is Bd5, a bishop move to the center that attacks my undefended knight on e4 and the pawn on b7 with the rook behind it. Two targets at once again. The evaluation went from −1.35 to +2.22, and I later lost that game with 7 blunders.

The lesson of the week

Looking at these three moments together, the pattern is simple: when I was better, I played the calm move and skipped the forcing one. Nd6+ was a check. Nxe4 was a capture that attacked the queen. In both cases I chose a move that "looked fine" and gave the advantage back. For the next weeks my rule is: when the position feels good, first list every check, every capture and every move that attacks something, for both sides, and only then pick a quiet move. If the engine's winning chance hits 80% and I still lose, the problem isn't the opening.

All games this week

The week before is week 48 and the next one is week 50. Every week of the series is on the journey hub, and if you want to review your own games like this, with the key moments explained in plain words, see how Chess Coach analyzes a game.

For a general method beyond this week's games, see the guide to reviewing lost winning positions.