Gold Miners vs Gold Beta-Hedged Residual Reversion: fade a >=2-sigma 10-session GDX-vs-GLD idiosyncratic move, dollar/beta-neutral long-short, 15-session time stop (USEQ 1-DAY)

Hypothesis

Two-instrument long-short pair on USEQ daily bars, OHLCV only. GDX starts in 2006 and GLD in 2004, so the overlap is about 19 years. GDX (gold miners) behaves like a levered claim on gold (GLD) because of the miners' operating leverage. Each session the strategy computes: (1) a rolling OLS beta of GDX daily log returns on GLD daily log returns, using data up to the previous session; (2) the daily residual; (3) the residual summed over the last resid_window sessions; and (4) a z-score of that sum against the prior 250 sessions. Large miner moves that gold does not explain are expected to partly reverse within weeks, because miner cash flows stay tied to the gold price. Examples are equity risk-off selling, ETF creation/redemption pressure and sector rotation. This design avoids the dead families in the lessons: the trigger is not a calendar date (L179), there is no crypto supplementary-feed gate (L111), and it uses no options (L102). Leverage is not a research parameter. Gross exposure comes from equity through the fixed gross_exposure_pct.