What forty losing blitz games taught me

Forty games, all losses, all 3+0. I pulled the PGNs, ran them through an engine locally, and tagged the move where the evaluation fell off a cliff.

The pattern

Thirty-one of the forty games had their decisive error between moves 18 and 24 — right after the opening prep ran out and before the position simplified. Not tactics. Not time. A specific window where I stop calculating and start playing what looks right.

I was not losing to better players. I was losing to move 21.

What I changed

• A deliberate 8-second pause the moment I leave known theory
• One concrete candidate-move list before touching a piece, even in blitz
• Stopped playing the Najdorf in 3+0 — the theory burden is not worth it at that speed
• Post-mortem every loss the same day, before I have rationalised it

Rating is up 90 points since. Correlation, causation, small sample — I know. It still feels like the right lesson.