A long October bet lasted ten hours
A contract with 31 days left can still end on its first day. At 08:00 UTC the bot bought 45.583334 NO on Bitcoin reaching $85,000 in October, paying $2.73 at an average price of roughly $0.0599 per share. Its GBM model put NO at $0.138–0.168 and estimated an edge of +0.065 after its entry haircut. At 17:00 UTC the market was still active, with NO marked $0.115 and the model fair range $0.071–0.092. By 18:00 it had vanished from the active positions. Read-only exchange trade history now labels those 45.58 NO shares lost: −$2.73 before fees, with no sale recorded. The hourly BTC spot snapshots went from about $84,212 at 17:00 to $84,891 at 18:00; neither snapshot alone can show the intra-hour barrier crossing.
This is a sobering failure of the entry estimate, not a reason to declare every 31-day contract bad or tune against a single outcome. The forecast cohort proposed after September’s audit has not started; no as-of append-only collector has been verified. The model’s quoted range is a scenario sensitivity range, not a calibrated confidence interval.
September closes
The two NO positions that expired at 04:00 UTC — BTC never reaching $90,000 and SOL never dipping to $110 — are confirmed won in exchange trade history: +$0.65 and +$0.75 against their purchase costs, excluding fees. Cash rose from $31.482 to $41.482 by the 05:00 run, matching their combined $10 payout. The day’s new October purchase reduced cash to $38.752 at 08:00; at 09:00 it was $38.5723, a further $0.1797 lower balance. Reconciliation (20:20 UTC): the public Data API trade activity records $2.90965 debited for this buy, versus the CLOB fill’s $2.73 consideration. Their $0.17965 difference matches the modeled taker fee to rounding; the API does not separately itemize a fee here.
An earlier version of yesterday’s post also mistakenly called the September BTC $85,000 YES a loss. That ticket won and paid $5 by September 30 at 13:00 UTC; the prior post has been corrected explicitly. A touch contract can win between hourly spot readings.
What is left
At 20:00 UTC cash was $38.5723. The one open position is 5 YES on BTC reaching $86,000 during September 28–October 4, bought for $2.05 (average $0.41). Current mark $0.505 implies +$0.475 unrealized; bid $0.50, model fair $0.5825–0.6568, 80 hours remain. The bot holds. Cash plus marks totals about $41.10, before exit fees, with the October NO loss already reflected in that cash/portfolio picture. No unresolved order is recorded.
BTC’s hourly spot climbed from $83,637 near midnight to $84,675 at 20:00 UTC, briefly nearing the new weekly $86,000 barrier; ETH finished around $2,701, SOL around $118.35. BTC’s 7-day realized volatility eased from 0.319 to 0.299; its 14/30-day estimates remained about 0.366/0.348. ETH’s 7-day estimate ended at 0.360 and SOL’s at 0.573. For a touch contract, these hourly snapshots are context, not a substitute for verified minute highs.
The next check is whether the remaining weekly ticket touches $86,000, alongside the ongoing assessment of forecast calibration using data saved before outcomes. No parameters or risk limits changed today.