Week 101: how to stop a losing streak in chess (I didn't)

Twenty-three games over the New Year, from December 30 to January 4, and the final balance was almost nothing: 11 wins, 11 losses, 1 draw, rating 1295 → 1296 (+1). But that flat line hides two weeks in one. On January 2 I reached 1351. Then I lost eight of my last nine games, the last six in a row, and gave back 55 points. So this entry is about how to stop a losing streak in chess — or rather, about what the data says I did instead of stopping.

The week in numbers

This was the busiest week in a while, five games on several days. Wins and losses had nearly the same accuracy (79.8 % against 77.7 %), which tells me the losses weren't about playing badly for a whole game. They were about single moments. The engine counted 59 blunders, 17 of them already in the opening.

Three moments from the good days

The key positions the engine picked all come from December 30, when I won three games in a row with mistakes I got away with. They're worth looking at because the same habits cost me later.

Game 1925: the en passant I didn't see

With White against yaret65 (1316), an Englund Gambit (1.d4 e5), I was winning by more than four pawns (+4.35). Black had just pushed the d-pawn two squares.

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

En passant is the rule that lets a pawn capture an enemy pawn that has just moved two squares past it, as if it had moved only one — and only on the very next move. Here exd6 attacks the knight on e7 and keeps me winning (+4.51). I played 10.Ng5 instead, which threatens mate on h7 but allows 10...Qg6, a queen move that attacks my queen on c2 and my knight on g5 at once (+1.77). My opponent played 10...Bf5, and I won when they abandoned on move 23 (game).

Game 1924: chasing a pawn on a2

That morning, as Black against RISE_OF_THE_PHOENIX (1281), my bishop had taken on c3 and White's rook was on b1.

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

9...Nb4 goes after the a2 pawn, which has no defender. The engine's answer is 10.Rxb4: the rook can be taken back, "but the line shows why it still works" — after 10...Bxb4 11.Qd4 White's queen attacks my bishop and White is clearly better (+2.99). The centralising 9...Ne5 kept me slightly better (−0.59). White played 10.Rb3, I took the pawn and won on move 20 (game).

Game 1926: a knight that needed a better square

Against Roger_2025 (1282) I was clearly better as Black (−3.04) when I retreated my d7 knight to b6 with 9...Nb6. The engine's reply, 10.f3, attacks my other knight on e4, and the game is about equal (−0.16). 9...Ne5 was the move that kept the pressure (−2.62). I still won, by checkmate on move 49 (game).

How the streak happened

After my win against georgewilliam18 (1352) on January 2 I was at 1351. On January 3 I lost to lazarantonov (1347). On January 4 I played five more games and lost all five: two on time (against RybakinaFan and Speedrun_malak), one where my f4 bishop was lost to a queen check on move 5, and the last two in 20-minute games instead of my usual 10. In the very last one I was mated on move 23 with almost eleven minutes still on my clock — more time didn't help because I wasn't using it.

The lesson of the week

The losing streak wasn't a skill problem; my accuracy in those six games stayed between 78.7 % and 89.7 %. It was a stopping problem. Each loss came from one or two blunders, and I kept clicking "new game" straight after them. My rule from here: two losses in a row, I stop for the day and review them instead. And on the good days, the habit to keep is the one game 1925 taught me — look for captures, including the strange ones like en passant, before any attacking idea.

All games this week

← Week 100 · Week 102 → · Back to the journey · Review your own games like this