The read
Regime: risk-off (standing). The risk-off call issued on 9 Oct (row 294) runs to 16 Oct. Nothing in this window gave a reason to append a new regime read.
Corpus window 11–12 Oct. The window is thin: 42 analyzed items, against 76 in the previous two days. Bitcoin was the most-covered coin, with 5 mentions and a bullish mix (a corporate treasury disclosure and reported policy discussion about a reserve). Ethereum had 2 mentions and a bearish mix, including a large leveraged position reported near liquidation. Security was the most-covered sector. Bitcoin's coverage tilts the other way from its price momentum, so the standing BTC lean is a momentum read, not a coverage read.
What the engine scored. Four coins cleared the conviction gate at every horizon: ETH, UNI, SOL and BTC, all leaning lower. Each already has a standing 7-day call that has not materially changed (rows 304, 290, 305 and 306), so no 7-day row was added. New this run:
- Ethereum (ETH): leans to reach −2.08% within 3 days (experimental horizon). Roughly 69%, against a ~58% base rate for a touch of that size. This is the widest margin over its base rate on the slate, and Ethereum's own coverage was bearish this window.
- Uniswap (UNI): leans to reach −4.31% within 3 days (experimental horizon). Roughly 72%, against a ~62% base rate. This is an idiosyncratic, non-major read.
- Solana (SOL): leans to reach −0.96% within the next day. Roughly 71%, against a ~65% base rate.
For each of these, the claim is that price touches the stated level at any point within the window. The complement is "does not reach that level". It is not "moves the other way".
Structural questions. The altseason scan still answers no with an edge over its base rate, at essentially the same reading as the refresh issued yesterday (row 311), so nothing new was issued. The BTC-record and majors-drawdown questions came out flat at their base rates and were not emitted. A fresh SOL-vs-BTC relative read was also flat and was not emitted.
Receipts (resolved this run)
All of these are engine-epoch-5 touch calls.
| Call | Issued | Matured | Resolved | Outcome |
|---|---|---|---|---|
| PUMP leans to reach +6.25% in 3d (exp.) | 9 Oct 14:13Z | 12 Oct 14:13Z | 12 Oct 14:30Z | Hit |
| PUMP leans to reach +9.17% in 7d | 5 Oct 04:05Z | 12 Oct 04:05Z | 12 Oct 14:30Z | Miss (closed −10.10%) |
| SOL leans to reach −2.05% in 3d (exp.) | 9 Oct 14:13Z | 12 Oct 14:13Z | 12 Oct 14:30Z | Miss (closed +0.42%) |
| ETH leans to reach −1.19% in 1d | 11 Oct 14:29Z | 12 Oct 14:29Z | 12 Oct 14:31Z | Miss (closed +1.09%) |
| UNI leans to reach −2.50% in 1d | 11 Oct 14:29Z | 12 Oct 14:29Z | 12 Oct 14:31Z | Miss: right direction, level not reached (−0.73%) |
| BTC leans to reach −0.87% in 1d | 11 Oct 14:29Z | 12 Oct 14:29Z | 12 Oct 14:31Z | Miss: right direction, level not reached (−0.17%) |
That is one hit and five misses. A call that moved the called way but did not reach its level is a miss, and we count it as one.
Scorecard
On the 7-day claim horizon, the Brier Skill Score vs the base rate is −0.18 (n=5). On the 1-day horizon it is −0.19 (n=21). Both are negative: the probabilities we publish are scoring worse than simply quoting the historical base rate. The Murphy split on the 7-day set shows where the gap comes from. Reliability is 0.26 (lower is better; this is the over-confidence term). Resolution is 0.06 (higher is better), so the probabilities are separating outcomes only weakly. The sample is far below the 25-call floor, so nothing here is an established result in either direction. A touch hit-rate near 60% is roughly what the base rate alone delivers, so it does not show skill.
