Week 12: how to get better at checkmating in chess, starting with a mate I missed

Week 12 (March 25–29, 2024) was the cleanest week of my first three months: 13 games, 8 wins, 5 losses, rating 999 → 1036 (+37), with five wins in a row and only 12 blunders, less than one per game for the first time. Three of the wins were checkmates. And yet the position I learned most from was a checkmate I didn't find: a forced mate in three that I walked past. If you're trying to figure out how to get better at checkmating in chess, that single position is the whole lesson.

The week in numbers

The accuracy (82.7%) was my best weekly average so far, and five games were above 95%: #315, #317, #321, #323 and #325. The streak ran from game #315 to #319, Tuesday and Wednesday, and took me from 991 to 1035. With White my 1.d4 games labelled Zukertort-Chigorin went 3 wins and 1 loss; with Black, the Slav (two games, both wins) worked well.

The losses were mostly ordinary, with one exception. Game #320 against janpoklop (1202) lasted seven half-moves: White's knight took my pawn on d5 and could simply have been captured by my queen. I played ...c6 instead, was a pawn down, and resigned on the spot. That's not a chess lesson, it's an attention one.

The mate I missed, and a capture with the wrong piece

Evaluations are from White's side. "M3" means the engine sees a forced checkmate in three moves.

Game #316: rook takes on f7, check

With White against Legionista696 (977) the position on move 20 was a forced mate in three. Black's king had walked up to f6. My rook was on h7, my knight on g5, and my queen on c2, which looks far away but sits on the diagonal c2–d3–e4–f5–g6 with nothing in between. I took Black's rook on h8 with my rook: a good, safe trade that left me winning (+3.47). The mate was rook takes the pawn on f7, with check.

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

Stockfish's line: rook takes on f7 with check, the king takes my knight on g5, another rook check from g7, and the queen comes to g6 for checkmate. Why didn't I see it? Because Rxf7+ looks like it gives something away: the king can take my knight. I never looked at a check that "loses" a piece. I still won by checkmate nineteen moves later, but it should have been over on move 22.

Game #318: the recapture that hangs the rook

With Black against olivierpollop (1173), I was winning (-5.89). I took the pawn on d4 with my rook. The engine shows the problem: my rook on d4 was protected only by my bishop on c3, and White can give up a rook for that bishop and then take my rook (+1.85). Taking on d4 with the bishop instead keeps everything protected and also pins the f2 pawn to White's king.

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

White missed it, and I won by checkmate eight moves later.

The lesson of the week

Getting better at checkmating, for me, came down to one habit: when the enemy king is exposed, look at every check first, even the ones that seem to lose material. Rxf7+ "loses" a pawn-protected knight, but every reply runs into the next check. I'd been filtering checks by whether they looked safe, and that filter is exactly what hides mates.

The second habit from game #318 is the same one as week 9: when two pieces can capture, compare them. One leaves a piece protected by a single defender that can be removed; the other doesn't.

Twelve blunders in thirteen games felt like progress, and it was. But the best moment of the week was still the one I missed.

All games this week

Before: week 11, three mistakes to avoid. After: week 13, beating the London as Black. Every week is listed on the journey page. To have an engine point out the checks you skipped in your own games, see how Chess Coach analyzes a game.