Large wicks do not fade the next BTC bar

Verdict: NO_EDGE · Asset/TF: BTCUSDT 1h (primary), 15m (secondary) · Sample: chronological 60/20/20

Hypothesis: a large unused wick predicts the next bar fades — large upper wick → next down; large lower wick → next up. Distinct from close-location (where the close sits) and from streak length.

What happened

Frozen discovery P80 (1h ≈ 0.46 upper / 0.49 lower).

Q1 fade folklore is wrong-signed. On 1h validation, large upper wick → P(next up) = 0.538 vs unconditional 0.510 (ΔP +0.028, mean +3.9e−4, t=+3.7). Large lower → P = 0.495 (ΔP −0.015). That is a mild continue tilt, and it misses the prereg ±0.03 gate (upper close; lower far). 15m same pattern, also fails. OOS attenuates.

Q2: Wick↔loc correlation ~±0.41 (not the ≥0.90 “iff loc” identity). Some continue increment inside up-bars (ΔP≈+0.07), but moot once Q1 fails.

Trading

Q3 skipped for promotion. Curiosity continue is positive at 0 bps and dies at 5 bps.

Final Verdict

NO_EDGE

Do not build “rejection wick → fade” bots on BTC klines from this definition. Close-location fade was a thin real effect that died on fees; unused-range wicks do not even clear the directional gate.