BTC shocks lift ETH size, not same-sign lead

Verdict: REJECTED · Asset/TF: BTC→ETH/SOL 1h · Sample: aligned ~2019→2026, chronological 60/20/20

Hypothesis: after a BTC top-decile |log return| shock, ETH (and SOL) move more in the same sign as the BTC shock than (a) unconditional and (b) their own past-sign momentum without a BTC shock.

Q1 — magnitude

YES. After BTC shock events (consecutive top-decile bars collapsed), ETH forward |return| lifts about +44% to +81% vs unconditional across horizons 1/3/6h. Vol contagion is real.

Q2 — same-sign lead (KEY)

FAIL. ETH validation same-sign rates 0.41 / 0.43 / 0.44 vs own-sign-only 0.47 / 0.48 / 0.48 → lifts −5.6 / −4.6 / −4.1 pp (bootstrap p(diff>0) 0.002–0.025). That is slightly worse than own-momentum, and below 0.5 (mild fade), not a lead. Shuffled BTC-shock labels are ~flat. SOL with the frozen BTC threshold also fails.

Trading

Q3 skipped. Gate failed.

Look-ahead

Discovery-only BTC |r| 90th pct = 0.009927, causal features, event collapse — clean.

Final Verdict

REJECTED

Alts co-move and carry their own momentum. A BTC shock adds size, not incremental same-sign lead into ETH/SOL. Do not build “BTC dumps → short ETH for lead” from this. Consistent with the vol-shock paper: magnitude persists/spreads; sign does not.