Dealer gamma level damps rest-of-day realized volatility beyond morning RV and IV

The canonical positioning mechanism, asked in the outcome space where this instrument has demonstrable power. When dealers hold positive gamma, hedging flows lean against price moves and should suppress how much the underlying actually moves. Does a morning read on dealer positioning predict lower rest-of-day realized volatility beyond both the morning's own realized volatility and the option market's own implied forecast?

Selected as HYP-010 by accumulated information value: the magnitude outcome space is where this instrument demonstrably has power (HYP-006 t=5.5, HYP-007 t=3.9), and gamma-damping is the canonical positioning mechanism in that space; NOT selected for HYP-004's gamma-level diagnostic p-value, which concerned direction. Standing limitations: then-unresolved OI-vintage label on all GEX quantities; options-surface-exposure multi-leg contamination uncorrectable. A null does not falsify strike-local gamma or inventory-signed variants.

Nine liquid names (two index ETFs and seven mega-caps) pooled with symbol intercepts over seven months of sessions. The signal is mean signed dealer gamma exposure over the first hour (09:30–10:30 ET), scaled by its own trailing 20-session mean absolute level using strictly prior sessions (at least 15 required, so warmup sessions drop). The controls are log morning realized volatility and log morning implied volatility over the same window, which is the strong incremental form: the signal must beat the market's own forecast, not just the morning's price action. The outcome is log realized volatility from 10:31 to the close. OLS with session-clustered errors; 999 permutation draws; preregistered negative direction.

A clear yes, in the registered direction. Beta −0.045 per unit of trailing-scaled positioning, t = −4.23, permutation p = 0.001, over 121 sessions across nine names, with a 95% interval of −0.068 to −0.025 that stays comfortably away from zero. The part that matters is what it beat: morning implied volatility is itself strongly predictive in the same regression (t = 5.34), and positioning still adds information on top of it. Dealer positioning is not a repackaging of the option market's own volatility forecast. This became the headline experiment of the dealer-positioning program and set up the two follow-ups that define its shape: whether the effect persists to the end of the session, and whether it lives at the name level or the market level. Both ran as their own preregistered experiments and publish in this ledger in turn.

The options surface behind positioning quantities covers only strikes that actually traded, not the full open-interest ladder. Statements about total or net positioning describe the traded subset of the book, not the whole book.

The open-interest snapshot behind positioning quantities was established as point-in-time valid by a prospective paired probe: open-settled, unchanged intraday and post-close, with no archive restatement. Results published before that verdict were produced under the then-unresolved vintage caveat. A weekly sentinel re-gates all positioning research automatically if the measured semantics ever change.

Formation windows close roughly a minute before outcome windows open, and measured feed publication latency consumes most of that slack. Live availability of each signal at the formation boundary is therefore tight rather than generous.

Sessions with incomplete implied-volatility inputs drop from the sample rather than appearing with imputed values. Reported session counts reflect those drops; nothing is filled in or interpolated.

This is a pooled effect and should be read as one. The cross-sectional version of the same claim, tested as its own preregistered follow-up, came back on the opposite side, which locates the damping at the market level rather than the name level. The positioning measure also aggregates every strike and expiration, and a material share of the underlying premium comes from multi-leg structures whose attribution cannot be corrected on this surface, so a null on any strike-local or inventory-signed variant would neither follow from nor contradict this result.

This is a volatility forecasting claim only; it says nothing about direction. It is a discovery-substrate result, preregistered but not yet confirmed on data the research process has never seen, and it is queued for prospective confirmation on naturally accruing unseen data before anything graduates.