How fast do dealers hedge SPX 0DTE flow?

The hedging footprint, its half-life, and the placebo that makes it credible

Published 2026-08-28 · Sample SPX regular sessions, 2022-04-14 to 2026-08-26, 1,088 days at one-minute grain · Method customer delta imbalance from the tape-signed dealer book; day and minute-of-day fixed effects; day-clustered errors; last 40 sessions held out · Status measurement, not a signal

Every claim this terminal makes rests on one mechanical premise: when customers force option delta onto dealers, dealers hedge it in the underlying, and the hedge moves the index. The premise is stated everywhere and measured almost nowhere — the canonical footprint study (Hu 2014) is about equities, pre-0DTE, at daily grain. So the concrete question: a customer imbalance lands this minute — how many minutes until the index has absorbed the hedge?

The design

The dealer's option-delta inventory is a published series here (the DEX readout). Its minute-to-minute change, negated, is the delta the dealer must hedge that minute. Regress forward index returns at horizons of 1 to 30 minutes on that imbalance, with day and minute-of-day fixed effects, the last minute's and last five minutes' returns (reversal), and the imbalance's absolute size as controls — so the answer cannot be "big flow happens on trending days". Errors cluster by day; the last 40 sessions are held out and scored once.

The answer: two to three minutes

The footprint is front-loaded exactly as a hedging story requires: strongest in the first two minutes, statistically gone by the fifth. On the 40 held-out sessions the sign is right at all seven horizons tested (t between 1.1 and 1.7 — forty days is a small sample, and it is reported as such). The absorption time of SPX 0DTE dealer hedging is about two to three minutes. As far as we can find, that number has not been published anywhere.

The placebo is the point

A skeptic should say: any activity measure drifts with returns. So the same regression runs again with one change — every minute's imbalance keeps its exact magnitude and receives a coin-flip sign. Everything else is identical. The placebo predicts nothing, at any horizon, in either sample (t between −1.8 and +0.6, no front-loaded shape). The information is not in how much traded; it is in which way the signing engine says it traded — which makes this the sharpest external validation of the tape-signed book we have: random signs kill it, the volume-convention signs carry the wrong pattern, the measured signs carry the mechanism.

What this is not

0.04 basis points per standard deviation of minute imbalance is real money only to the hedging desks themselves. It is not a trading edge after any spread, and this study registers no claim that the imbalance predicts returns profitably. It measures the machine's latency and validates the signing; see the Terms — nothing here is investment advice.

Reproduce it

Everything needed is in the public session files at https://gex.live/snapshots/YYYY-MM-DD.json (dates at /sessions): the dex minute series is the dealer delta inventory and spot the index path. Difference the inventory, negate, and run the regressions above; the placebo is a seeded coin flip on the imbalance sign. Restart minutes (inventory jumps at engine restarts) are dropped at the 99.9th percentile of absolute imbalance.

Part of gex.live research. Measured on the free session archive; every session is free to replay.