← nuc.grChess · auto-written weekly
The slow bleed 2026-06-04 → 2026-06-10 chess.com/nucos

I drifted to rapid; blitz stayed at 900.

What works showed up too: mornings (53%), the 800s I keep crushing (85%), and a White edge (50%) that flips the rapid script. Then the rest — 34 games to June 10, 14–18–2, minus-20 to 902, before I set blitz down and drifted to rapid. My own 900s are a coin flip (41% over 150), the 1000s a wall (28%), and the Caro-Kann that flies in rapid leaks to 38% here, though it still flashed two clean mates. The bleed is slower than last week's, but the number's stuck at 900. The next mountain is my own bracket.

Record
14–18–2
34 games
Rating
902
-20 · off a 925 week-high
Best streak
3
in a row
Accuracy
85.3
mean
Win rate
41%
this stretch
14 won 2 drawn 18 lost
Rating · the arc · hover to read it
start 922 peak 925 now 902

Thirty-four games and the line just saws sideways — 922 down to 902, never out of the 899–925 band. No run to ride, no collapse to fear. The blitz number is simply stuck.

Openings — what's working
London Systemwhite · 62g · 51.6%
The daily driver from White — 51.6% over 62. Solid, unspectacular, the one line that stays above water all stretch. model game ↗
Caro-Kann Defenseblack · 60g · 38.3%
Most-played as Black and the week's leak — 38.3% over 60. The defense that wins 83% in rapid; on the fast clock it bleeds, yet still flashed two checkmates. model game ↗
Queen's Pawnwhite · 12g · 58.3%
A fine second White — 58.3% over 12. When the London isn't on the board, this picks up the slack and scores better. model game ↗
Englund Gambitwhite · 8g · 62.5%
When opponents meet 1.d4 with 1...e5, I'm ready — 62.5% over 8, the best return of any line this stretch. model game ↗
Games worth seeing
Win rate by opponent strength
800–899 · 20g 85%
900–999 · 150g 41%
1000–1099 · 29g 28%
Tendencies the coach noticed
White's the better half

White scores 50.5% over 97; Black just 37.9% over 103. The mirror of rapid — there the Caro-Kann flies, here the same defense drags the Black score down.

I climb by beating down

The shape of the slump: 85% over the 800s, a 41% coin-flip with my own 900s across 150 games, and a 28% wall at the 1000s. The wins come from below me, not above.

How it ends

When I win I finish it — 39% by checkmate. When I lose it's resignation (51%), outplayed not out-clocked; only 9% of losses fall on the flag. Time isn't the killer.

Mornings, then a fade

The one bright window is early — 53% in the morning over 55 games, easing to 42% in the afternoon and 39% by evening. Late-day blitz is where the rating leaks.

Season to date
2026-05-03 → 2026-06-10 200 games 88–105–7 peak 1040 streak 4 scalp 1143

Five weeks, 200 games, 88–105–7 — a sub-.500 grind that's bled a 1040 peak down to 902. My own 900s are a coin flip, the 1000s a wall, and the rating keeps finding new floors. This is the one format that won't climb.

Past editions
Where it climbs · "Give me a real clock and the wall disappears."The relapse · "The wall came down to meet me."The correction · "The wall pushed back."The opening stretch · "The Sicilian is carrying me."