Week 99: how to convert an advantage in chess, after ten wins and nine losses

Nineteen games in six days, ten wins and nine losses, and my rating went from 1313 to 1328 (+15). The week from December 16 to 21, 2025 started with four wins in a row, and three of my games went above 95 % accuracy: 97.6 %, 96.8 % and 95.5 %. But when I went through the engine review, the theme wasn't brilliance. It was the question every improving player hits: how to convert an advantage in chess once you have it. Five of my nine losses came from positions where I'd been winning, and all three key moments the engine picked for this week happened in games I won anyway.

The week in numbers

The split is telling: my wins averaged 84.2 % accuracy and my losses 65.0 %. With White I went 5–4, with Black 5–5. Thirty-one of the forty blunders came in the middlegame. On the first day alone I played six games — four wins, then two losses — and my rating touched 1345 before drifting back.

How to convert an advantage in chess: three wins that wobbled

Game 1890: the rook I didn't take

As Black against tolivetolove (1325), a Pirc Defense, I reached an endgame where I was completely winning. White's rook on f6 was attacked by my bishop on d4.

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

I played 42...Kg5, attacking the rook with my king. The rook just stepped away with check (43.Rf5+) and my position went from "completely winning" (−7.31 after Bxf6) to "clearly better" (−1.85). I still won by resignation six moves later (game), but that's the purest example of the week: when a capture wins a whole rook, take it. Attacking a piece gives it a move to escape; capturing it doesn't.

Game 1891: a quiet rook move with my queen far away

With White against oliveticon (1325) my queen had raided all the way to a8, on Black's back row. The review had me at +5.30 before move 22.

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

After 22.Rac1 the engine's answer is 22...Qa6, which attacks my rook on f1 and my queen on a8 at the same time — a double attack — and the evaluation turns clearly better for Black (−2.09). The pawn push 22.b5 kept me clearly better (+2.70). My opponent found 22...Qa6, the queens came off, and I added three more blunders in the next seven moves (23.b5, 26.Rfc1 and 29.dxc5). The game still ended in my favour when my opponent abandoned on move 41 (game). Lucky, not good.

Game 1889: the pawn on a6

Against Oswaldo_00 (1323), a Modern Defense, the position was level on move 29 (−0.07). I moved my knight from d7 to f6, and the engine's replies include 30.Rxa6, simply taking my a-pawn (+2.05); in the game the pawn fell a move later, 31.Rxa6. From b8 the knight would have kept an eye on a6 and c6, and the engine rated 29...Nb8 about equal (−0.20). I won that one too, when my opponent abandoned on move 45 (game).

The losses where I was winning

The other side of the week hurts more. Against J4sure (1357) I was −3.64 when 27...Rxf6 gave White +9.28. Against harryajm (1359) I was −6.22 before 17...g5 handed over +3.65. And against saskaroni (1354) I had +7.74 on move 42 with one second on my clock; I played 42.Bh5 and lost on time. That last one isn't a chess lesson, it's a clock lesson: in a 10-minute game I had 244 seconds left by move 23 and 77 by move 35.

The lesson of the week

Converting a winning position is mostly about doing less. Take the free piece. Defend the loose pawn. Don't start a second plan while your queen is far from home. When I'm clearly winning, my job is to remove my opponent's chances, not to find the most beautiful move, and to keep enough time on the clock to finish the game.

All games this week

Previous: week 98 · Next: week 100 · The whole journey · Review your own games like this