The read
The market stays choppy. On the last closed daily bar Bitcoin sat about 3% above its 20-day average, but its 7-day change was about +0.5%, well inside the flat band, so the objective read has not moved since yesterday. The standing choppy call (to 9 October) stays in place and nothing new was issued for the regime. The older rotation call (to 6 October) is also still running.
Coverage volume dropped sharply in this window (41 analyzed items against 106 in the prior one). Bitcoin coverage skewed bearish, while Ethereum, Solana and the DeFi sector read constructive.
New calls (all short-horizon, all touch-graded)
The 7-day lane is full until 4 October, so today’s calls sit on the shorter experimental horizons. Each one is graded on whether price touches the stated percentage at any point before expiry, not where it closes.
- Pump.fun (PUMP), 3 days: leans to reach +5.90%, ~76% vs a ~67% base rate. Coverage is concentrated in one analyst.
- NEAR Protocol (NEAR), 3 days: leans to reach +6.46%, ~71% vs a ~63% base rate. Same concentration caveat.
- Solana (SOL), 3 days: leans to reach +2.71%, ~69% vs a ~61% base rate. Broadest coverage of the three 3-day calls.
- Bitcoin (BTC), 1 day: leans to reach +0.82%, ~62% vs a ~56% base rate. Mostly momentum; coverage is mixed.
- XRP (XRP), 1 day: leans to reach +1.80%, ~62% vs a ~57% base rate. Thin coverage.
Touch probabilities sit near 60-75% because touch base rates are that high. The gap over each base rate is the engine’s current estimate and has not been validated. It is not a measured edge. The 3-day calls are excluded from the public accuracy claim.
Standing structural reads were re-scored and all sit within the material band of their current rows, so none were re-issued. Ethereum was a conviction-passer but lost a 1-day slot on rank, and Monero was left off the slate because its last 1-day call has not been graded.
Receipts since the last run
Calls issued on or before 31 July were graded on a size condition as well as direction. Calls from 1 August to 1 October were graded on direction at the close. Calls issued since are graded on a touch of the stated level. Rows are marked by rule, and the three are never compared or pooled.
| Call | Issued | Matured | Resolved | Result |
|---|---|---|---|---|
| Bitcoin, reach +0.86% within a day | 2 Oct 03:35Z | 3 Oct 03:35Z | 3 Oct 03:40Z | Hit (touch; closed -0.98%) |
| Ethereum, reach +1.17% within a day | 2 Oct 03:35Z | 3 Oct 03:35Z | 3 Oct 03:40Z | Hit (touch; closed -1.50%) |
| NEAR, higher over 3 days, direction rule | 30 Sep 03:08Z | 3 Oct 03:08Z | 3 Oct 03:40Z | Miss (-5.43%) |
| Pump.fun, higher over 3 days, direction rule | 30 Sep 03:08Z | 3 Oct 03:08Z | 3 Oct 03:40Z | Miss (-5.09%) |
| Solana, higher over 3 days, direction rule | 30 Sep 03:08Z | 3 Oct 03:08Z | 3 Oct 03:40Z | Miss (-0.70%) |
| Regime: risk-on, direction rule | 26 Sep 02:05Z | 3 Oct 02:05Z | 3 Oct 03:40Z | Miss (BTC +0.23%, inside the flat band) |
| XRP outperforms BTC over the quarter, prior grading | 4 Jul 17:43Z | 2 Oct 17:43Z | 3 Oct 03:40Z | Miss (XRP +28.3% vs BTC +33.8%) |
Two hits and five misses. Both hits closed below where they started. They count as hits only because the touch rule grades the path, and that is exactly what the stated claim was. The whole 3-day experiment batch from 30 September missed.
Accuracy
The current engine has no headline Brier Skill Score yet. Zero 7-day calls issued under the touch rule have resolved, against a minimum of 25. The only touch-graded rows so far are two 1-day calls, which is far too few to read anything into. The record from earlier engines was negative: their probabilities scored worse than quoting the base rate. Nothing in this run changes that, and the record should be treated as unproven until the current engine builds a sample.
