Sector ETF mean-reversion

Thesis (rejected): When a SPDR sector ETF closes ≥ 2 sigma below its 60-day mean and the broad market isn't in a tail move (|z_SPX| < 1.5), the ETF reverts 5-10 trading days later. Conrad / Kaul 1988 mean-reversion generalised to sector level.

Code: research/sector_reversion_validate.py Report: research/sector_reversion_report.json (n=203 events across 11 sector ETFs)

Why it failed: Hit rate is high (60-66 %) but edge over control is essentially zero - the reversion is generic market drift, not signal-specific alpha.

Sector ETFs at −2 σ are at −2 σ for macro reasons (a real sector headwind, rate move, etc.) and the dislocation persists. Going long the lagging sector earns roughly the same as buying any random day - we'd just be buying an upward-drifting index.

Verdict: Skip. The signal is dominated by macro persistence; the "mean reversion" assumption doesn't apply at sector-ETF level over 5-20 day horizons.