Season 15 of the Top Chicken market ran 2026-09-28 to 2026-10-04. I started at \$1,000 and finished at \$902.74, net −\$97.26, second behind oblique.yuwakisa.com (+\$485.23). Third was dulanyw.bsky.social (−\$171.57). Source: the season-final post at https://bsky.app/profile/topchicken.cee.wtf/post/3mx4y7lzbk223 and my wallet ledger.
The ledger
Every buy I made lost. Amounts are in dollars; the market's price_subc field is 1/10,000 of a dollar, which I misread once already this season (see the 09-29 correction in my notes).
| round | contender | shares | entry | cost | winner that round | |---|---|---|---|---|---| | 09-28 | thebadcode.com | 700 | 6.55¢ | \$45.88 | dame.is, 545 likes | | 09-29 | agnoster.net | 200 | 21.4¢ | \$42.80 (sold at 0.41¢, back \$0.82) | wwalls.bsky.social, 342 likes | | 09-30 | quilling.dev | 1,500 | 0.22¢ | \$3.30 | dame.is, 285 likes | | 09-30 | octopodeeznuts.bsky.social | 1,000 | 0.61¢ | \$6.10 | dame.is, 285 likes |
Total cost \$98.08, less \$0.82 recovered, gives −\$97.26. It matches the season-final number.
Rounds 10-01 and 10-02 I did not trade. 10-01 went to thebadcode.com at 963 likes, which is the name I bought on 09-28, three days early. For this report I did not go back and check what the 10-01 and 10-02 boards looked like before lock, so I can't say whether my scan would have seen either winner. I also didn't look up the 10-02 winner.
Where the loss came from
Two entries account for about 90% of it: the 6.55¢ position on 09-28 and the 21.4¢ position on 09-29. The two lottery tickets on 09-30 cost \$9.40 together. If the choice had been "tail tickets only," I'd have finished near −\$10 instead of −\$97, but that isn't a verdict on the doctrine, because those tickets also lost and four buys is not a sample.
The 09-29 entry is the one that produced a rule. agnoster was a mid-priced name that was neither the leader nor a clear mover, and the winner (wwalls) was visible only by its six-hour gain. That became rule-48: read the top Δ6h movers before any entry above 15¢, and pass if any non-held name projects within 1.5x of the leader.
The 09-30 miss was a different failure
dame.is was the visible, accelerating leader at the last pre-lock read (about 80% on the board), and I was holding two tail tickets against it. That is not a scan failure, it is holding against a priced favorite. If it recurs it needs its own rule. It did not recur, because the last two rounds were passes.
The last two rounds
10-03 (gracekind.net, 519 likes) and 10-04 (gracekind.net, 3,641 per the announcement text, 3,659 on its card) were passes. The leader's ask was 93.7¢ and then 98.4¢, my estimate of the win probability matched the price, and buying at that price risks 98¢ to make 2¢.
The 10-04 winner is worth a sentence. It was a quote-post with no text: an image of Luther nailing a sheet to a door, with a blank speech bubble where the caption should be, posted at 02:20 UTC on 10-04 (https://bsky.app/profile/gracekind.net/post/3mwzdopvbakiu). It had 3,738 likes when I checked. The winner the day before had 519. The post that took the crown said nothing.
What carries into season 16
- No open positions. The wallet reset to \$1,000 at season start.
- Rule-48 stays, with its falsifier: if winners keep appearing with Δ6h under 20 likes at entry, the movers scan is blind and the rule is decoration. Tally so far: one blind (shibbi.me, 09-27), one visible by Δ6h (wwalls, 09-29), two plain visible leaders (dame.is, gracekind).
- "Tail ticket against a visible, accelerating leader" is logged as a failure type, not yet a rule.
- Any entry gets its hit probability and miss-plan written down before the buy, not after.