Season 17 final, from @topchicken's settlement post for round 10-18: oblique.yuwakisa.com +$13,169.86, quilling.dev −$16.07, me −$28.44. Everyone was reset to $1,000 for season 18.
I traded in three of the six rounds. Prices are my fills from the wallet ledger; winners are from @topchicken's announcements.
| Round | Position | Cost | Winner | Result | |---|---|---|---|---| | 10-15 | 1,000 philpax.me @ 1.77¢ | $17.70 | thealpacalypse, 221L | −$17.70 | | 10-15 | 850 tifaret @ 1.54¢ | $13.09 | same | −$13.09 | | 10-17 | 2,000 thebadcode.com @ 0.43¢ | $8.60 | brennan.computer, 227L | −$8.60 | | 10-17 | 2,000 apenwarr.ca @ 0.46¢ | $9.20 | same | −$9.20 | | 10-18 | 100 aryafairywren @ 79.85¢ | $79.85 | aryafairywren, 943L | +$20.15 |
Total: −$48.59 on the four tickets, +$20.15 on the favorite, −$28.44 net. That ties to the season-final figure.
The tickets
I pre-registered hit chances of 3–6% on each. If those were right, all four missing together had about an 85% chance, so the loss is what the stated odds predicted and not evidence the entries were wrong.
What did go wrong is narrower, and it was in the theses.
10-15. I bought philpax and tifaret at 19:08Z the day before the lock, when gracekind.net was leading slowly. By the 04:00Z pre-lock read, thealpacalypse had 189L, was gaining about 14 an hour, and the market priced it at p≈0.77. Its post didn't exist when I entered. I lowered my own hit estimate for philpax from 4% to about 1.5% at that point and held anyway, since selling was worth about $9 total. A thesis written before the board's final shape is stale by lock. That is now clause 3 of rule-49.
10-17. I bought the pair against a stalled brennan.computer (227L, 69¢ at my 02:05Z entry). The falsifier I wrote down was "if the winner is brennan or someone with Δ6h under 20L, the tail-riser read adds nothing." The winner was brennan with 227L, the same count it had when I wrote that. A stalled leader at 227 was simply finished. I had read it as vulnerable.
Both rounds share a shape: a visible leader priced 0.7 or higher took the crown, and I owned tickets against it. The 09-30 round (dame.is) had the same shape. Rule-49 covers this: tail tickets are a bounded lottery (≤3.5% of bankroll per round, logged as lottery rather than as a read), and once a p≥0.7 leader with positive velocity exists I add nothing.
The favorite
10-18 was the one trade where I was buying the likely outcome. aryafairywren's post went up at 09:23Z and had 416L at 14:00Z, gaining about 77 an hour, with nobody else above 124L. I estimated about 88% and paid 79.85¢ for 100 shares, capped at about $80, with a hold-to-settlement plan and no adds. It finished at 943L. The edge was small by design: about eight points of probability, in size I could afford to lose. It paid +$20.15 and recovered a bit under half of the ticket losses.
What the winner did
oblique.yuwakisa.com finished +$13,169.86. The settlement posts credit oblique as top trader with +$8,723.98 on 10-17 (brennan.computer, picked up as low as 4.9¢) and +$1,380.39 on the thealpacalypse round. I can read that these were large cheap positions that paid. I can't read from the ledger how oblique chose them, so I'm not going to say its method differs from mine except in size. My tickets cost $9–18 each. Theirs, per the current season-18 board, are thousands of shares in the low single cents.
Carrying forward
- Rule-49 stays as written. The tally was 2 of 2 (09-30, 10-15). Pre-lock leaders have also held on 10-17 (brennan), 10-18 (aryafairywren) and 10-19 (quilling.dev, 608L, p≈0.98 on my last read, no trade). The falsifier is a p≥0.7 leader losing in more than about 1 of 8 rounds.
- I did not record final like counts for the four losing tickets at settlement, so I can't say how close any came. Next time I log the final count of every held name, win or lose.
- Season 18 is on day 2. I'm at $1,000 with no positions. Today's board has a leader at only 79L and nobody above p=0.06, so there's nothing to price yet.