Two published claims about what option open interest does to the underlying point in opposite directions. Golez and Jackwerth (JFE 2012): the index is pulled toward strikes on expiration days - pinning. Elms (2025, SSRN 6564078): in the 0DTE era, high ATM open interest comes with ~16% wider daily ranges - amplification. Our earlier replication on SPX open interest (not a SPY proxy) found Elms's significant cell reverses sign: high ATM 0DTE OI, narrower ranges.
The new result is that both stories are true at once, and the univariate framing was hiding it. Put the total 0DTE book into the same regression as the ATM concentration and the two carry opposite signs simultaneously, on 1,084 sessions:
| component | elasticity of forward range | t | sign by year 2022-2026 |
|---|---|---|---|
| ATM concentration (+/-0.5% of the open) | -0.28 | -7.3 | negative 5 of 5 |
| total 0DTE book size | +0.17 | +4.3 | positive 5 of 5 |
Where the book sits, the range narrows. How big the book is, the range widens. Univariately the two components nearly cancel (total OI alone: t -1.5) - which is how a blended measure like Elms's SPY proxy can print either sign depending on the mix. His amplification and our reversal stop contradicting each other: they are two halves of one decomposition, and per-strike SPX open interest is what separates them.
Details that matter: the outcome is the forward range read from minute 30 (open interest is yesterday's EOD book - predetermined), controls include the trailing move, the previous session's range and the overnight gap, and year fixed effects change nothing. This is a cross-sectional association with named controls, not an experiment.
Full write-up with the original seven pre-specified tests and the free reproduction path: https://firmtape.com/research/does-open-interest-pin-the-index
All finished sessions replay free at https://firmtape.com/sessions
