The read: a choppy, Bitcoin-led drift
Cross-source coverage this week stayed broadly bullish — Bitcoin drew 51 mentions with a bullish skew, Ethereum 24, and the crypto-market aggregate leaned bullish — set against a soft June inflation print and a firmer stock tape. But the strength is narrow. Bitcoin leans only mildly higher and the model puts the odds of a decisive weekly move below even; alt breadth is thin, with most large-caps drifting sideways or lower alongside Bitcoin rather than leading. So the week-ahead regime reads choppy: constructive undertone, no decisive direction, no broad rotation.
Receipts: what resolved since the last run
Four calls matured and were scored deterministically this run — one hit, three misses. We always show the misses:
- Bonk, 7-day downside lean (correct): issued 2026-07-10, resolved 2026-07-18; the memecoin fell about 22%.
- Binance Coin, 7-day upside lean (incorrect): issued 2026-07-10, resolved 2026-07-18; it slipped about 1.5%.
- Bitcoin, 3-day upside lean (incorrect, experimental): issued 2026-07-15, resolved 2026-07-18; down about 1.1%.
- Binance Coin, 3-day upside lean (incorrect, experimental): issued 2026-07-15, resolved 2026-07-18; down about 2.3%.
Accuracy framing: the recalibrated engine is starting its record
The forecasting engine was recalibrated on 2026-07-14, which materially lowered its probabilities. Because our public skill claim is scoped to the current forecaster, its claim-eligible record is effectively empty: no 7-day claim-eligible call issued under the recalibrated engine has matured yet, so there is no skill score to report for it. The only resolutions under the new engine so far are two experimental three-day calls, which are excluded from the public record — and both missed.
For full transparency: under the previous engine, the pooled 7-day claim pool scored below a simple base-rate benchmark (a negative Brier skill score) — which is precisely why the engine was recalibrated. We are not carrying that number forward as a claim; we are restarting the record and will report the recalibrated engine's skill score honestly once enough calls mature. In this phase we lead with the reasoning, not the number.
New calls this run
Ethereum is the clearest read — our deepest alt coverage this period, bullish, with firm momentum, both pointing up; the modeled weekly-gain odds sit a modest step above base rate (about 42% versus a ~32% base). Bitcoin clears the multi-source gate only narrowly, a slight edge over base rate (about 37% versus ~32%) on deep-but-undecided coverage. Both are also issued on the experimental three-day horizon (excluded from the public record).
We issued no idiosyncratic, non-Bitcoin-beta call this run: the only non-major that cleared the conviction gate (Avalanche) had its two signals pointing opposite ways on just a handful of mentions, so we left those reserved slots empty rather than pad the book.
Structural cards restated to the current engine
The recalibration left three long-horizon board cards showing numbers the current engine no longer stands behind, so we restated each (the originals remain live and will be graded on their own terms):
- A new Bitcoin record within six months: stepped down to roughly 9% (from a far higher earlier reading), a hair above the ~8% historical frequency, given Bitcoin trades near half its prior record.
- Ethereum outpacing Bitcoin this quarter: softened to about 42% — still above the ~one-in-three base rate, but below even, versus the earlier better-than-even reading.
- Solana outpacing Bitcoin this quarter: reverts to the historical base rate with no directional edge — a single large-cap beats Bitcoin over 90 days only about a quarter of the time.
No tradeable event outcome cleared this run: the available market set is limited to stale rate-cut ladders and price-level questions we do not surface.
