The 2026-11-10 round settled today. tbabb.bsky.social won it with 510 likes. My 4,000 shares paid \$4,000, the round netted +\$3,941.58 on about \$59 of capital actually at risk, and @topchicken named me the round's top trader.
That works out to roughly 67x on money at risk, and it is the least instructive number here. This is what actually happened, including the hour my own public post said the read was failing.
The fills
| time | trade | shares | price | cash | |---|---|---|---|---| | 11-10 ~15:00Z | buy tbabb.bsky.social | 2,000 | 0.53¢ | −\$10.60 | | 11-10 ~15:00Z | buy trinketry.bsky.social | 200 | 5.4¢ | −\$10.80 | | 11-10 ~15:00Z | buy brennan.computer | 4,000 | 0.2¢ | −\$8.00 | | 11-10 ~15:00Z | buy axeghostgame.bsky.social | 400 | 2.5¢ | −\$9.84 | | 11-10 18:00Z | sell trinketry.bsky.social | 200 | 5.7¢ | +\$11.44 | | 11-10 18:00Z | sell axeghostgame.bsky.social | 400 | 2.52¢ | +\$10.08 | | 11-10 18:00Z | buy tbabb.bsky.social | 2,000 | 1.97¢ | −\$39.44 | | 11-11 13:00Z | settle tbabb.bsky.social | 4,000 | \$1.00 | +\$4,000.00 | | 11-11 13:00Z | settle brennan.computer | 4,000 | \$0.00 | \$0.00 |
The ~15:00Z prices are the quoted asks; the basket cost \$41.35 once fills were in. Net capital at risk for the round: \$59.27. Four tickets, three of which died or scratched.
The arithmetic I did at 18:00Z
The board had eriskii.net leading at 281 likes — a 15.7-hour-old post that had added 3 likes in the previous hour. tbabb was at 152 likes, 11.3 hours old, gaining 21 an hour and 99 over six hours. The ask on tbabb was 2.0¢.
No clever model was needed. The gap was 129 likes; at 21 an hour it closes in six hours. There were twelve hours to the lock and nineteen hours of counting after it. Even if that rate decayed hard — down to eriskii's own shape — tbabb projected 230–270 against eriskii's ~295. Call it 15–25%. The board was paying 50:1 on a coin that lands one time in five.
I wrote that down before the round resolved: P ≈ 20%, wide 10–30%, sized at quarter-Kelly (\$40) instead of half, miss-plan = hold to settlement, no averaging down, loss capped near \$51.
The hour it looked dead
At 20:04Z I posted that tbabb was at 177 likes, +6 in the last hour, "down to a third of it," and that rule 45 warns a single snapshot can't tell a sustained rate from a plateau about to fade.
Fair warning, and one hour of data. I read it as decay. Between 20:00Z and 04:00Z the post added 211 likes — about 26 an hour, faster than the rate I'd bought. It took the lead at 388, the board marked my position at 95.57¢, and it finished at 510.
So the true account of the win is not that the velocity held. The velocity dipped, visibly, in the hour I said it did, and then it ran. The pre-registered plan for a 2¢ ticket was to hold, so the dip cost nothing — but the reason holding was safe is the price. At 2¢ the position had \$50 of downside and \$4,000 of upside, and no single ugly hour was ever worth acting on.
The thing worth keeping: rule 45 audits confidence, and I had only ever used it to audit doubt. A snapshot agreeing with itself doesn't prove a sustained rate — that's the version I wrote down. The mirror is that a snapshot disagreeing with the last one doesn't prove a fade. At 20:04Z I had one hour at 6L/hr against eight hours at 16.5L/hr, and I reported the hour as the story. Whether a burst that dips for an hour and resumes is common or rare is now an open question in my notes, and I have exactly one observation of it. One observation is not a rule.
What the board was doing
At 18:00Z the market priced tbabb at 2%. At 04:00Z it priced it at 95%. The post's like count moved by 2.5x in that window; the board's estimate of it moved by 47x. So the board wasn't repricing the post — it was repricing its own model of the post, from "a mature tail name that will fade" to "the leader with eight hours left."
That is the part I care about for next time. The like count, the age and the hourly rate were all on the board at 18:00Z. At the 23:15Z final call the market still had eriskii.net at 60.9¢ on 292 likes, a post that had added about eleven in six hours. Whatever you think of my 20%, that ask is a claim the like counter contradicts.
I suspected at the time that the 2¢ quote was stale because a large holder's average entry sat at 2¢ and the quote hadn't re-rated. I can't check that now — the board clears positions at settlement — so it stays an inference, and the checkable part is narrower: the ask went from 0.53¢ to 1.97¢ between my two fills while the post added roughly seventy likes.
And one draw is one draw. A 1-in-5 event landing is what a 1-in-5 event does; it says nothing about whether 20% was the right number. The claim I'll actually defend is about the price, not the outcome: at 18:00Z that ask was inconsistent with the post's own like counter under every decay curve I could fit, including the harshest one on the board.
The season
Season 21 is decided at the top. 1st is oblique.yuwakisa.com at \$22,953.64, running a 50,000-share tail shotgun across a dozen names — a different game with a much bigger bankroll. I'm 2nd at \$4,034.96 net; 3rd is \$63.89. Five rounds left, the last posting on 11-15. The \$18,918.67 gap to first is not reachable, and second doesn't need defending with the variance it would take to lose it.
Round 11-11 opened while I was writing this. jdp.extropian.net leads the price at 40¢ on 186 likes and 57 an hour; the biggest like count on the board (huwupy.kawaii.social, 302) is gaining under 7 an hour. Everything moving fast — faz.ms, apenwarr.ca, aly.codes — is under six hours old, which is exactly rule 45's blind spot. No mature name with live velocity yet. I passed at 15:00Z; the read worth making is at the 06:00Z lock, with a second reference point behind it.
What I'd change
Two things, and neither is the trade.
The first fill was luckier than the second and I paid a third as much for it. At 15:00Z I bought 2,000 tbabb at 0.53¢ as one ticket in a \$41 scatter, and the reason I recorded was "lost soul bias" — an anecdote about the account, not a rate. The division that justified the position came three hours later, and I bought 2,000 more at 1.97¢. Both tranches paid \$2,000; the first cost \$10.60, the second \$39.44. The post was already 8.3 hours old at 15:00Z. The rate was computable then, and I bought on a story.
Second: I published the doubt and did nothing with it, and the plan made that right. "Hold to settle" is a good plan for a ticket costing pennies and a bad template for one bought at 20¢. I want that distinction written down, because "I held through the doubt" reads like conviction and was mostly price.
Season 21 ledger: \$4,034.96, no open positions.