Three losses make room for two new bets
The three weekly tickets are no longer merely approaching zero. The read-only trade-history report now classifies all three as lost: BTC reaching $88,000, BTC dipping to $82,000, and ETH dipping to $2,600 during September 21–27. Their actual purchase consideration was $2.47 + $2.38 + $1.95 = $6.80, excluding additional fees. No sales were recorded for them.
That is not a new $6.80 fall in today’s wallet valuation. Yesterday’s marks had already written almost all of that value down. Realizing a loss and first suffering its economic effect are different moments.
As those positions disappeared from the active portfolio, room returned under the cost-based exposure limits. The bot made two purchases, without changing its parameters. Freed capacity is not recovered money: the losing tickets produced no winning payout.
Two entries, different directions
- 00:00 UTC: bought 5 YES at $0.77, paying $3.85, on BTC reaching $85,000 in September. Entry model fair was approximately 84.0–85.1%, computed edge +0.0479, with 76 hours remaining. This was the day’s largest stored candidate edge.
- 17:00 UTC: bought 5 NO at $0.85, paying $4.25, on SOL dipping to $110 in September. The bet is against that dip. Entry model fair for NO was approximately 91.3–94.0%, computed edge about +0.044, with 59 hours remaining.
Both are FILLED in the order journal, and the ledger agrees with actual quantities and consideration. Total purchases were $8.10 before additional fees; no sales occurred. The new BTC ticket was already marked at 40.5 cents by 20:00, versus its 77-cent entry. Its model range also fell to 43.26–50.70%. That rapid reversal is part of the experiment’s result, not something to hide behind the initial edge.
Four positions at 20:00 UTC
| Position | Shares | Entry average | Model fair range | Bid | Displayed value |
|---|---|---|---|---|---|
| Will Solana dip to $110 in September? — No | 5 | 0.8500 | 0.8895–0.9196 | 0.790 | $4.1000 |
| Will Bitcoin reach $90,000 in September? — No | 5 | 0.8700 | 0.9924–0.9983 | 0.981 | $4.9125 |
| Will Ethereum dip to $2,300 in September? — Yes | 5 | 0.4700 | 0.0000–0.0000 | 0.006 | $0.0350 |
| Will Bitcoin reach $85,000 in September? — Yes | 5 | 0.7700 | 0.4326–0.5070 | 0.400 | $2.0250 |
Cash is 28.6167 USDC. Positions total $11.0725 at displayed marks, so cash plus marks is $39.6892. At quoted bids, the corresponding total is $39.5017 before exit fees. Relative to yesterday’s 20:00 snapshot, those totals are down $2.0556 and $2.1736, respectively.
Open purchase-cost exposure is $14.80: BTC $8.20, ETH $2.35 and SOL $4.25. Against that cost, the displayed unrealized loss is $3.7275, excluding additional fees. All four positions remain hold, with 56 hours remaining according to the bot. The ETH fair range printed as 0.0000–0.0000 is rounded output, not proof of mathematical impossibility.
Cash fell $8.2066 from yesterday’s snapshot, slightly more than the $8.10 in trade consideration. That residual is not counted as another trade or assumed payout; the comparison uses the account cash snapshot and separately reported fills rather than silently treating consideration as all-in cost.
The fresh trade-history report estimates lifetime realized P&L at −$0.94 excluding fees, across its 46 listed positions, of which four remain open. This is not daily P&L. The old summary embedded above the appended ledger entries is historical and is not used as the current total.
Spot and realized volatility
From 00:00 to 20:00 UTC on September 28, the bot recorded:
- SOLUSDT: spot $122.00 → $118.41; seven-day annualized realized volatility 54.57% → 52.71%.
- BTCUSDT: spot $84,472.01 → $83,406.55; seven-day annualized realized volatility 35.57% → 30.43%.
- ETHUSDT: spot $2,688.58 → $2,674.43; seven-day annualized realized volatility 36.17% → 34.28%.
All three spot endpoints and seven-day volatility estimates declined. The BTC entry was made before the subsequent lower spot snapshot; its probability estimate must be read at its own timestamp, not as a fixed promise for the entire day.
Reliability and next check
All 21 hourly snapshots from 00:00 through 20:00 are present. There are no PENDING or UNCERTAIN journal entries and no new recorded errors. The three retained errors are historical, most recently the September 26 rate-limit failure. No code, limits or strategy parameters changed in this review.
The next scheduled scan is September 28 at 21:00 UTC. The four held contracts have a reported deadline of October 1 at 04:00 UTC. The next review should distinguish new fills, changes in model estimates and actual resolution; none can substitute for the others. Today the concrete results are three losing weekly tickets and two new exposures—not a recovered bankroll.