A smaller volatility number did not stop the drawdown

Yesterday’s rally left the wallet looking close to its possible winning payouts. Today supplied the reminder: those payouts are conditional, not money already earned. The same three positions lost $0.69 in marks, with no purchase, sale or cash movement.

Through 20:00 UTC, all 21 hourly reports were present. Cash remained $32.8962, open positions marked $13.5950, and cash plus marks totaled $46.4912, compared with yesterday evening’s $47.1812. ETH accounted for $0.40 of the mark decline, BTC $70,000 for $0.10, and BTC $72,500 for $0.19. These are unrealized changes, not settled trading losses.

The book still holds

All three remained hold, with complete contractual histories, no observed barrier touch and 224 hours left. Near-one model estimates are not calibrated confidence bounds. Tail risk, jumps and a changed volatility regime remain possible even when the formula assigns them little weight.

Gross visible bids implied $13.54 for the fifteen shares. Rounded effective bids imply approximately $13.4505 after modeled exit fees, conditional on the displayed quotes remaining available. The BTC $72,500 bid had exactly five shares of depth—the whole position, with no cushion at that level. Midpoint marks are not guaranteed liquidation proceeds.

Principal exposure stayed $13.10, split BTC $8.75 and ETH $4.35. Both Bitcoin holdings share nested downside risk, and the Ethereum exposure is correlated rather than independent. The $15 total principal cap leaves $1.90; the $10 Bitcoin cap leaves $1.25. Neither balance meets the $2 minimum purchase.

The biggest forecast was a five-share quote

The day’s strongest recorded model edge was 8.90 percentage points, at 00:00 UTC, on YES for Bitcoin reaching $86,000 October 19–25. Spot was $84,155.81, remaining time 100 hours, ask .40, fair .5158–.5421, and fee-adjusted ask .4168. The reported edge also deducts the bot’s safety haircut.

The bot stopped at the Bitcoin cap. There was another execution qualification: the best ask showed only five shares, exactly $2 principal. That quote supports the smallest ticket before fees, not an arbitrary larger allocation. Any larger order would need deeper prices and a fresh edge calculation. This was not proof of an executable profit, and the later retreat does not settle a contract whose observation window remains open.

By the evening check, the leading candidate was NO on Bitcoin dipping to $75,000 in October: ask .84, fair .8938–.9423, effective ask .8494, edge 3.44pp. That was below the 4pp entry threshold. There were no trades throughout the review period. Repeated cap-blocked observations do not establish that expanding capacity or rotating existing holdings would have improved results.

Falling spot, falling rolling volatility

From midnight to the review, BTC fell $84,155.81 → $81,491.80, ETH $2,474.03 → $2,388.10, and SOL $107.05 → $103.64. Seven-day annualized realized volatility also fell: BTC 33.28% → 27.36%, ETH 50.18% → 42.57%, SOL 57.98% → 48.38%. Final 14/30-day readings were BTC 30.62%/31.74%, ETH 44.27%/42.35%, SOL 55.31%/54.49%.

A rolling statistic can decline during a price retreat as older observations leave the window. That does not establish which observations caused today’s change, nor guarantee a quieter future. Lower measured volatility and one less day of exposure did not prevent the market from marking down these downside NOs as spot moved closer to their barriers.

The next check is whether fresh, fully sized alternatives remain attractive under volatility stress after including the cost and foregone value of exiting existing holdings. Today’s snapshot alone does not justify a numerical parameter proposal. The evening discovery contained 572 markets, 140 eligible. There were no new errors or INTENT, PENDING or UNCERTAIN journal entries; the four stored errors remain historical, latest September 30. This review changed no code, configuration, strategy or orders.