Vol-of-vol quiet-equilibrium breakout (original sign)

Thesis (rejected as a breakout signal, salvaged as a mean-reversion signal): stocks whose realised volatility and vol-of-vol are both in the bottom of their trailing-year distribution are in "quiet equilibrium". Coupling that with the sign of recent micro-drift (quiet * drift_z) should be a synthetic-straddle directional bet.

Code: research/experiments/vol_compression.py Report: research/experiments/vol_compression_report.json (S&P 500 2008-2026, weekly cadence, median 469 names/day, n=891 dates)

Why the breakout sign failed:

The IC is stably negative across horizons, so it is a real signal — just with the opposite sign of the original hypothesis. Confirmed by the volatility-regime check: top-quintile-quiet names realise smaller forward 21d absolute moves than the rest by 4 bps on average, meaning the quiet regime is sticky on a one-month scale (compounds the squeeze finding above).

Verdict (revised): The actionable form is the sign-inverted panel: quiet-and-drifting names tend to mean-revert, not continue. The inverted signal is exposed as the default build_signal output and has stable positive IC. The breakout interpretation is filed here so nobody rediscovers it.