Week 42: how to stop blundering in chess? Look at my wins

Week 42 was small and, for once, fairly clean: 9 rapid games from October 21 to 27, 2024, 5 wins and 4 losses, and my rating went from 1001 to 1009. I made 13 blunders, about 1.4 per game, my lowest rate since early August. And the split tells the whole story of how to stop blundering in chess, or at least why it matters: one blunder in my five wins, twelve in my four losses.

The week in numbers

Four of my five wins had no blunder at all. Game 1040 against Bishop_Goes_PewPew (1029) was the best: 97.0 % accuracy, no mistakes or inaccuracies, and my rating peaked at 1025. Game 1038 against tomerd9991 (999) was the cleanest game with Black: 96.0 %, not a single inaccuracy.

Then October 27 ended badly. After that 97 % win I lost twice in a row with Black, at 46.5 % and 49.4 % accuracy, both by checkmate, and both from positions where the engine had given me an 80 % or better chance of winning. Same day, same player, two very different games.

How to stop blundering in chess: what separated my wins from my losses

I looked at what the four blunder-free wins had in common, and the honest answer is: nothing fancy. They were games where nothing got captured without me checking the recapture. Three of them were my usual 1.d4–Bf4 setups with White, one was a Giuoco Piano with Black. My losses came from the same kind of openings (two of them with the same 1.d4–Bf4 setup); what changed was whether I made a capture or a "threat" without asking what came back.

The three key moments of the week are exactly that kind of move.

Game 1034: a threat that cost my opponent nothing

With White against Sco24 (1003), I was clearly better (+1.93). My bishop on b5 could go to c6 and attack the rook in the corner.

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

Bc6 does attack the rook, which had no defender. But a rook that's attacked can move, and the engine's reply is Rb8, which "attacks the pawn on the queen's side, which has no defender" — my b2 pawn. The other option for Black, Nxc6, simply trades off my bishop. Either way the advantage was gone (-0.63). The quiet Bd3 kept "clearly better for white". I lost the game, recorded as abandoned.

Game 1036: the trade that let a pawn hit my bishop

With Black against RED10OUANE (985), the position was about equal (+0.32). White's knight had landed on e5 and I took it.

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

A knight for a knight looks like an even trade. But White recaptures with the f-pawn, and position.py describes that recapture as "Pawn takes the knight in the center: attacks the bishop on the queen's side", my bishop on d6. The engine's verdict after fxe5 is "white is winning" (+3.77 at the time, +4.83 in its own line). Chasing the knight with f6 kept "about equal". My opponent didn't find the best follow-up, and I won the game by checkmate, but the trade itself was a blunder.

Game 1037: the Englund Gambit capture that opened lines

With White against jie_naphie313 (1011), in the Englund Gambit (1.d4 e5), I was a pawn up and better (+1.15). Black's pawn on f6 was attacking my pawn on e5.

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

exf6 attacks the queen, but Black simply recaptures with the knight, developing it with a capture, and the engine rates that "slightly better for black". The game's analysis preferred Qd3, a queen move that attacks the bishop on d7 (when I re-ran the position, the engine put the similar Qd2 on top, "slightly better for white"). In the game it got worse: after 9...Nxf6 I grabbed a pawn with 10.Bxc7 and the evaluation went to -5.31. All four of my blunders in that game came in the opening, and I abandoned it after 30 half-moves.

The lesson of the week

This week gave me the clearest evidence so far that, at 1000, blunders decide games: 1 in my wins, 12 in my losses. Two of the three key blunders were captures where the recapture was the problem: fxe5 hit my bishop, Nxf6 developed Black's knight for free. The third was a "threat" that let the attacked rook move with a threat of its own. My rule from week 42: before I capture, I look at the recapture; before I threaten, I look at where the threatened piece goes. It's not a cure for blundering, but it would have removed all three of these.

All games this week

Previous: week 41 · Next: week 43 · Journey hub · Review your own games like this