Today's read
Over the 7–8 October window, coverage of the majors cooled. Bitcoin drew 11 mentions with a bearish tilt, and Ethereum drew 7 with a neutral one. Total analyzed items fell to 59, from 111 in the prior window. The market-regime read is still the standing choppy call issued on 2 October, which matures on 9 October. We did not re-issue it today.
Our engine combines our analyzed coverage with price momentum. Four reads cleared its conviction gate and were issued today, and all four lean lower:
- Uniswap (UNI), 7 days — leans to reach −6.87%. The engine puts this at about 74%, against a base rate of about 65% for a move that size. Momentum carries most of the lean. Coverage was constructive earlier in the week (a fee-switch value-capture piece and a pullback plan), but the latest item reports that pullback setup stopping out, which tilts coverage slightly lower.
- Uniswap (UNI), 1 day — leans to reach −2.60%. About 73%, against a base rate of about 64%. Same drivers, shorter window.
- Ethereum (ETH), 1 day — leans to reach −1.19%. About 62%, against a base rate of about 54%. Ethereum coverage is close to balanced: analysts flag failed acceptance near resistance and a fresh local low, while others still see a recovery. The lean comes mainly from momentum. It sits beside the standing 7-day Ethereum upside call. That call is a separate horizon graded on its own terms, and nothing here offsets it.
- XRP, 1 day — leans to reach −1.44%. About 69%, against a base rate of about 66%. This read is momentum-led. XRP coverage is thin and leans slightly the other way, so this is not a coverage call.
PUMP also cleared the gate on the upside, but it already has two standing calls, so nothing new was issued for it. The 3-day experimental lane is full until 9 October. The structural scan produced nothing new: the altseason re-score is not materially different from the standing call, and the record-high question already has its two standing calls.
These probabilities are touch probabilities: the chance that price reaches the stated percentage at any point within the horizon. The complement is "does not reach that level", not a move the other way. Base rates for touch calls sit near 55–65%, so a probability around 60% is not by itself a sign of skill.
Receipts
The two seven-day calls marked prior grading were issued under the previous rule, which scored direction alone at expiry. Current-epoch calls are graded on whether price reached the stated level at any point within the horizon. The older rows are marked, and the two rules are not compared.
| Call | Issued | Matured | Resolved | Outcome |
|---|---|---|---|---|
| Ethereum — leans to reach −1.13% within the next day | 7 Oct 13:55Z | 8 Oct 13:55Z | 8 Oct 13:59Z | Hit — level reached; closed −2.10% |
| Solana — leans to reach +1.37% within the next day | 6 Oct 13:40Z | 7 Oct 13:40Z | 7 Oct 13:51Z | Miss — level not reached; closed −4.11% |
| Ethereum — leans to reach +1.13% within the next day | 6 Oct 13:40Z | 7 Oct 13:40Z | 7 Oct 13:51Z | Miss — level not reached; closed −5.77% |
| Bitcoin — leans higher over the next 7 days (prior grading) | 1 Oct 03:18Z | 8 Oct 03:18Z | 8 Oct 13:56Z | Miss — closed −0.67% |
| Avalanche — leans higher over the next 7 days (prior grading) | 1 Oct 03:18Z | 8 Oct 03:18Z | 8 Oct 13:56Z | Miss — closed −0.24% |
Accuracy
The headline measure is the Brier Skill Score against the base rate. It scores the probability we publish, not just the direction. It is not available yet. The current epoch's 7-day headline pool has no resolved calls (0 of the 25 needed), so there is no headline score, and no reliability/resolution split either.
The one-day horizon is the only epoch-5 claim horizon with resolved calls. Across 9 calls to this run (8 October, 14:01Z), including the Ethereum hit above, it scores −0.104 against the base rate. That means its probabilities have so far done worse than quoting the historical frequency. The sample is far too small to settle anything. The previous engine version, graded on a different rule, closed its 7-day record at −0.018 on 36 calls, slightly worse than the base rate. Our probabilities have not yet beaten the base rate, and we report that plainly.
