The read
Regime: choppy (week ahead). Bitcoin's own trend lean is weak. Momentum across the large-cap majors (XRP, SOL, DOGE, BNB, ADA) still points lower, but most of those reads rest on momentum alone, so they do not clear the engine's multi-source gate. Read together, that favours a range-bound week for Bitcoin, inside the flat band the regime grader uses, over a clean risk-off leg. This read is graded on Bitcoin's realized trend at expiry.
New relative call: Ethereum vs Bitcoin, 90 days. Ethereum leans to finish at or below Bitcoin's return over the next 90 days, at roughly 65% against a ~56% historical base rate. The corpus signal and momentum point the same way. This differs materially from the standing read issued on 2 October, which was close to a coin flip. That read stays open and is graded independently. Nothing has been edited or superseded.
No new asset calls. The engine's gate passed only Bitcoin and Ethereum this run. Both coins already carry their full two concurrent calls (a 7-day and a 3-day, all leaning lower), so nothing new was added. The higher-scoring momentum-only leans failed the gate because they lack a second signal, and they stay uncalled.
Structural questions. Altseason and the Bitcoin-record question each already carry their two standing reads, so no refresh was added. The drawdown question produced no emittable read this run.
Corpus note. This window's corpus snapshot is empty because news ingestion paused after 13 October. It resumed today, so the corpus signal currently draws on the trailing week.
Receipts (resolved since the last run)
- Zcash (ZEC), 7-day, leaned to reach -6.15%: issued 9 Oct 14:13Z, matured 16 Oct 14:13Z, resolved 16 Oct 15:13Z. Hit. The level was touched along the path, even though the close at expiry was +0.85%.
- Week-ahead regime: risk-off: issued 9 Oct, matured 16 Oct 14:13Z, resolved 16 Oct 15:13Z. Miss. Bitcoin moved -1.06%, which is inside the flat band, so the realized label was not risk-off.
- Ethereum vs Bitcoin this quarter, lean softening below even (stated yes at ~42%): issued 18 Jul, matured 16 Oct 08:10Z, resolved 16 Oct 15:13Z. Hit. ETH +31.5% vs BTC +26.1%.
- Solana vs Bitcoin this quarter, reverting to the base rate (stated no at ~75%): issued 18 Jul, matured 16 Oct 08:10Z, resolved 16 Oct 15:13Z. Miss. SOL +39.0% vs BTC +26.1%.
Track record
Our probabilities have not yet beaten the base rate. Over the trailing 30 days to this run (16 October), the Brier Skill Score against the base rate on the 7-day claim pool is -0.074 (n=7). On the 1-day lane it is -0.171 (n=23). Both samples are well below the 25-call floor, so the verdict is still insufficient data. The Murphy split shows reliability at 0.099, which is where over-confidence shows up, and resolution at 0.054, which measures how well our probabilities discriminate. Over-confidence is still the larger term.
As a separate directional-accuracy note: on the same 7-day pool the hit rate trails the base rate by about 18 points. This is not a skill measure, because it ignores the stated probability.
Touch calls ask whether a level is reached at any point within the horizon. Their base rate is close to 60%, so a probability near 0.6 reflects the question type rather than skill.
