The bounce that refused to come last week showed up: thirty-two games, a dead-even 16–16, and the rating up 15 to 903 anyway. The shape tells it, a slow bleed to an 864 floor mid-week, then a five-win streak, the season's best, hauling it to 911 before settling. Checkmates from both colors led the reel; the ugliest loss had mate on the board and fell on time. The season sits at 94–98–8 over 200 games, a 47% grind below the 995 peak. The 800s still fall (70%), the 900s stay a coin flip (40% over 144), and mornings (53%) beat evenings (40%). The work is unchanged: keep the won ones.
A slow bleed to an 864 floor mid-week, then five straight wins to 911 before settling at 903. The dip reversed this time; the late games finally paid the rating back.
White scores 48% over 98, Black 46% over 102. The Black half caught up a step this season; the Caro-Kann keeps it level rather than ahead.
Below 900 the wins come easy (70% over 53); at 900–999 it is 40% over 144. The 1000s barely feature yet, three games all told.
Mornings run 53% over 58, evenings 40% over 48. The pattern has held for weeks: the early games bank the points the late ones spend.
A 38% season throwaway rate and 2.52 blunders a game. The five-win streak showed the ceiling; the giveaways still set the floor.
Two months, 200 games, 94–98–8 at 47%, the 995 peak the mark to chase. The 800s fall (70%), the 900s are a coin flip (40% over 144), and a 38% throwaway rate is the standing tax. The climb resumes when the won ones stay won.
Bullet got actual reps this week: ten games, 6–4, the rating up 7 to 843 and brushing the 849 season peak. The middle sagged to 811 before four straight wins pulled it back, vintage bullet whiplash. The season now reads 35–28–2 over 65 games at 54%, and the patterns hold: Black is the better half (59% over 32), the Sicilian keeps working at speed (53% over 15), and most results still land on the clock, with 57% of wins by flag. Evenings are the bullet hours, 65% over 23. Fast hands earned their keep this week; the think-slower lesson can wait until they don't.
Ten games of classic bullet whiplash: an early 845, a slide to 811, then four wins in a row back to 843. The sawtooth cuts both ways; this week it cut upward.
57% of season wins come by flag and 43% of losses go the same way. Bullet here is a flagging contest first and a chess game second.
Black scores 59% over 32, White 49% over 33. Bullet stays the one format where the Black pieces are the stronger half.
65% over 23 in the evening block, the best late-day number on the page. The fast games like the warmed-up hands.
Three months, 65 games, 35–28–2 at 54% with the 849 peak two points away. Black leads (59% over 32), the Sicilian works at speed (53%), and the flag still decides most results. Quietly the healthiest fast-format ledger.
Rapid finally got volume and paid the toll: twenty-four games, 9–15, the rating down 41 to 1180. The slide was a sawtooth, not a collapse, every rally back to 1220 handed straight back, and the culprit is plainly marked: at 1200+ the season score is 37% over 60 games, while the 1100s keep falling at 72% over 36. The Black side and the Caro-Kann still carry the format (56% and 54%), and the week's best wins were real scalps, a 1217 ground down from the Black side the pick of them. The 1250 peak keeps its distance for now; the route back runs through the 1200s.
Twenty-four games of sawtooth drift from 1221 to 1180, with three separate rallies to 1220 that never held. Not a collapse, a toll: the 1200s charging admission.
Below 1200 the wins pour in, 72% over 36 in the 1100s; at 1200+ it is 37% over 60. This week was played hard against that line, and it showed.
Black scores 56% over 54, White 48% over 46. The Caro-Kann remains the most reliable tool in the rapid bag.
Afternoons run 56% over 43, evenings 46% over 35. The late-day sag that taxes blitz shows up here too once the volume rises.
A hundred games in a single month, 52–42–6 at 52%, with a twelve-game streak and a 1253 scalp on the books. The 1100s fall at 72%; the 1200s hold at 37%. The 1251 peak waits on the other side of that wall.
The daily board went quiet, nothing finished since late June, so the season remains the story, and it is still the best one on the page: 18 games, 13–3–2, a 72% clip at 1009 with 86% accuracy. The pattern behind it has not moved either: with days per move the blunders that sink the fast formats simply never get played, Black scores 82% over 11, and even the London holds even. Small sample, uneven opposition, same conclusion every format keeps teaching: more clock, better chess. The queue just needs feeding.
Eighteen games from a 956 floor to a 1052 peak, resting at 1009. The arc has not moved in weeks because nothing new has finished; the shape it left behind still points up.
Black scores 82% over 11, White 57% over 7. With days per move, the Black openings finally get the room they need to work.
86% accuracy and a 72% score. The blunders that decide bullet and blitz never make it onto the slow board.
A clean sweep below 900 and 60% over 5 at 1000–1099. The sample is small and it flatters, but the trend keeps agreeing with it.
Two and a half months, 18 games, 13–3–2 at 72% with 86% accuracy. Black wins 82% and patience does the rest. Still my best chess, whenever the moves actually get made.