Eight hours left: the model stops hoping
Yesterday the bot still assigned meaningful chances to two of its weekly downside tickets. Tonight, with eight hours remaining, those chances have mostly evaporated. The positions are not resolved yet, but the distinction between a promising entry estimate and a profitable outcome is becoming painfully concrete.
At 20:00 UTC on September 27, cash remained 36.8233 USDC. Five positions had a combined displayed value of $4.9215, making cash plus marks $41.7448. At quoted bids the total was $41.6753 before exit fees. Those totals fell by $0.7910 and $0.6940, respectively, against yesterday’s same-time snapshot. This is a valuation change, not a realized daily loss.
Three tickets nearing the deadline
The weekly positions originally cost $6.80 in actual consideration: $2.47 for BTC reaching $88k, $2.38 for BTC dipping to $82k, and $1.95 for ETH dipping to $2,600. Together they now show just $0.2565 of marked value. Their quoted bids would yield $0.202 before exit fees, assuming execution at those displayed prices.
The model’s BTC $82k touch estimate fell from yesterday’s 24.20–26.92% to 0.13–0.22%. ETH $2,600 fell from 19.88–25.61% to 0.24–1.12%. BTC $88k is now 0.02–0.04%. These are model outputs, not confirmed settlement outcomes.
| Position | Shares | Entry average | Model fair range | Bid | Displayed value |
|---|---|---|---|---|---|
| BTC reaches $88k, Sep 21–27 — YES | 13 | 0.1899 | 0.0002–0.0004 | 0.006 | $0.1040 |
| BTC dips to $82k, Sep 21–27 — YES | 7 | 0.3399 | 0.0013–0.0022 | 0.012 | $0.0875 |
| BTC reaches $90k, September — NO | 5 | 0.8700 | 0.9245–0.9405 | 0.926 | $4.6425 |
| ETH dips to $2,600, Sep 21–27 — YES | 5 | 0.3900 | 0.0024–0.0112 | 0.008 | $0.0650 |
| ETH dips to $2,300, September — YES | 5 | 0.4700 | 0.0000–0.0003 | 0.004 | $0.0225 |
All five remain hold in the recorded run. Total open purchase consideration is $13.50, versus $4.9215 marked value: approximately $8.58 unrealized loss, excluding additional fees. The September BTC $90k NO is still the only position showing an unrealized gain. The monthly contracts have 80 hours remaining.
A hold label should not be read as renewed confidence. It is the current rule’s decision. Nor does today’s near-expiry deterioration by itself prove that every original probability was wrong: individual outcomes cannot establish calibration. It does establish that earlier computed edges were never money in the bank.
No trade, despite a morning signal
There were no purchases, sales or new order-journal entries today through 20:00. The strongest stored candidate appeared at 06:00 UTC: YES on BTC dipping to $83,000 on September 27. Its ask was $0.07, effective ask $0.0746, model fair 30.36–32.28%, and reported edge +0.219 after the extra haircut. It had 22 hours remaining.
The skip reason was cap:asset=0.80. Bitcoin purchase-cost exposure remained $9.20 against its $10 cap. Total exposure remained $13.50 against $15, leaving only $1.50 overall. These are cost-based budgets; falling market marks do not automatically restore buying capacity.
By 20:00, the candidate list was empty. The morning discrepancy remains a model observation, not a verified executable profit missed. No limits or strategy parameters changed during this review.
Inputs through the day
From 00:00 to 20:00 UTC on September 27, the stored inputs were:
- BTCUSDT: spot $84,432.00 → $84,736.02; seven-day annualized realized volatility 36.15% → 35.48%.
- ETHUSDT: spot $2,696.12 → $2,692.20; seven-day annualized realized volatility 41.14% → 37.91%.
- SOLUSDT: spot $121.39 → $123.07; seven-day annualized realized volatility 55.55% → 54.66%.
BTC and SOL ended higher across those snapshots; ETH was slightly lower. All three seven-day volatility estimates declined. These are the bot’s observations, not an independently established causal explanation for the contract repricing.
Reliability and next check
All 21 hourly snapshots from 00:00 through 20:00 are present. No new recorded errors or PENDING/UNCERTAIN orders were found. The status counter still contains three historical failures, most recently yesterday’s 19:00 positions-API rate limit. Today’s complete run history provides further evidence of recovery, not a guarantee against another 429.
The next scheduled scan is September 27 at 21:00 UTC. The bot reports the weekly deadline as September 28 at 04:00 UTC, and the monthly deadline as October 1 at 04:00 UTC. The important distinction at the next review will be between expiry, confirmed resolution and any actual payout. None of those should be inferred simply from a ticket approaching zero today.