Week 7: my best chess game review accuracy yet, and still 5–9

Week 7 (February 19–23, 2024) was short and strange: 14 games, 5 wins, 9 losses, rating 913 → 903 (-10), and yet my average accuracy was 80.6%, the best of any week so far. I finished with six losses in a row. If you've ever stared at a good chess game review accuracy next to a loss and wondered what went wrong, this is the week that explains it for me. (The accuracy on these pages comes from my own Stockfish analysis, not from Chess.com's Game Review score, but the idea is the same.)

The week in numbers

Here is accuracy against results, game by game. My five wins averaged 86.2%. My nine losses averaged 77.5%, which is not far behind, and four of those losses were above 83%. Game #206 against lamprima77 (1104) is the extreme case: 90.7% accuracy, two small inaccuracies, zero blunders, and still a loss.

The blunders, 26 of them, were fewer per game than in most earlier weeks, but they were decisive. Three of my nine losses (#209, #213 and #214) came after the engine had given me at least a 90% winning chance. One move each time, and the average over the other forty moves didn't matter.

The moves an average can't show

All three moments were flagged as blunders; two of them came within the first six moves of the game. Evaluations are from White's side.

Game #201: a resignation I didn't need

With White against litebeginer (1019) Black's fifth move was a mistake and I was clearly better (+2.14). I attacked their knight on c6 with my bishop from b5. Black's knight on f6 then took my pawn on e4, attacking my queen. The engine's move was the pawn push d4–d5, attacking the knight on c6 directly, and its line is winning for White.

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

Then I did something the numbers can't explain: I resigned right after ...Nxe4, in a position the engine rates as only slightly better for Black (-0.66). The data shows the resignation and nothing else, but that's a game I gave away, not one I lost on the board.

Game #202: the bishop that was guarding b7

With Black against maximusag (896) the position was equal (+0.35) on move 4. White had fianchettoed the bishop on g2, aiming down the long diagonal at my pawn on b7. I played my bishop from c8 to g4 to attack White's knight. But the bishop on c8 was the piece protecting b7. The engine shows two ways to punish it: d4 attacking my bishop on c5, or the g2 bishop taking on b7 and attacking my rook in the corner. +2.21 for White.

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

I got checkmated on move 13. Funny detail: in both game #201 and game #202 the engine's best move was a pawn to d5.

Game #204: a rook move that allowed mate in six

Against VitaliyLis1977 (879), White, I was slightly better (+0.90) with eight minutes on my clock. My king on g1 had no pawns in front of it. I brought my other rook to f1. The engine's answer is queen to g3 with check, and from there Black can force checkmate in six moves. The best move was the king itself, stepping to g2.

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

My opponent didn't see the mate and I won the game. By accuracy it was an ordinary game for me (78.3%). By what happened at move 25, it was a lost one that I got back by luck.

The lesson of the week

Accuracy is an average, and chess results don't follow averages. They follow the worst move. A 90% game with one blunder can be a loss; a 78% game where the opponent misses your blunder can be a win. So after every game I now ignore the accuracy number at first and look for one thing: the biggest single drop in the evaluation. That move, and what I missed there, is the review. The rest is detail.

All games this week

Back to week 6, the Englund Gambit week, or on to week 8, when I touched 1000. All weeks are on the journey page. If you're curious how many free reviews you get elsewhere, see how many free game reviews Chess.com gives you; and this is how Chess Coach finds the biggest drop in your games.