SPX Dealer Positioning Explained: Why the Standard OI-Based Model Misleads

The question “what are dealers positioned in?” is not a philosophical one - it decides whether a rally is going to pin or accelerate, whether a put wall will hold or crack, and whether the last hour will compress or expand. Every major GEX tool answers that question the same way: they assume. This guide is about why the assumption fails on SPX and what to do instead.

The mechanics of dealer hedging are not the topic here - that is covered in the GEX guide. This guide is specifically about the sign problem: how do you actually know which side of the book dealers are on?

The standard assumption model

Almost every retail-facing GEX product runs the same algorithm:

• For each strike on the SPX chain, pull open interest for calls and puts.
• Assume calls are net long by dealers and puts are net short by dealers.
• Multiply OI by per-contract gamma and the sign fixed by the assumption.
• Sum across strikes to get aggregate GEX.

The assumption comes from a plausible story about who trades what. Retail buys calls (lottery tickets, upside expressions), so dealers sold them and are short calls. Yield-seekers sell puts (cash-secured-put strategies), so dealers bought them and are long puts. Under that story, the dealer inventory has a fixed sign per strike, and open interest tells you the size.

It is a story about long-run averages that is right often enough to look useful and wrong often enough to burn traders who lean on it.

Where the assumption breaks

0DTE flow is two-sided

The assumption bakes in a directional bias from retail behaviour. 0DTE breaks that. Same-day flow on SPX is approximately balanced between calls and puts, and much of it is sold to dealers rather than bought - closing trades, spreads, and strategies that leave dealers long inventory they did not choose. In a 0DTE-heavy session, the assumption’s sign is wrong by lunch.

Systematic vol strategies write options

Covered-call ETFs, put-write funds, dispersion books, and volatility-target strategies all write options in size. Every written contract flips the assumed dealer sign at that strike. As these strategies have grown into meaningful open interest, the aggregate dealer position has shifted from the long-call / short-put assumption toward something much more mixed.

Intraday sign changes

Even in sessions where the aggregate OI-based sign is roughly right, dealer inventory moves through the day. A large institutional put block at 11:00 can flip dealer positioning at that strike from short to long. An overnight sweep before the open can invert the entire near-the-money profile. The OI-based model is a snapshot; positioning is a movie.

Regime-dependent bias

In quiet vol regimes, systematic strategies dominate and dealer positioning skews away from the assumption. In stress regimes, hedging flow dominates and it swings back. The assumption model has no way to track this - it is regime-blind by construction.

The assumption model is not stupid. It is a reasonable prior. It is just a prior - not an observation. When flow overwhelms the prior, an observer with the tape sees it. A model built on the prior does not.

A flow-based approach

A flow-based approach ignores the assumption. Instead of inferring dealer positioning from open interest, it derives it from actual options flow - the trades that print during the session. Over the course of the day, that flow reveals which strikes dealers absorbed inventory on and which they gave inventory on. The resulting position estimate reflects what happened, not what usually happens.

A flow-based estimate agrees with the assumption model on slow days and diverges in every session where positioning actually matters - high-flow sessions like FOMC, CPI, quad witching, and any 0DTE-heavy Friday.

When the two models disagree

On any given day, plot both. Most of the time they roughly track. The interesting sessions are the disagreements. A few worth watching for:

• Post-CPI drift. On sessions with a bullish CPI print, the standard model shows positive gamma and predicts pinning. A flow-based estimate often shows that dealer positioning flipped short in the first ten minutes as retail bought calls aggressively. The tape trends up instead of pinning; the assumption model is wrong for the whole session.
• Systematic-write ETF expiry. Big covered-call ETFs write monthly calls just above spot. The standard model treats those calls as dealer-long, producing a call wall. A flow-based estimate sees the writing flow, flips the sign, and shows that the “wall” is actually a magnet - spot pushes through and accelerates.
• Late-day 0DTE reversal. Standard model shows positive gamma into the close. A flow-based estimate shows that the last hour’s flow flipped dealers short. Instead of pinning, the tape breaks the closest wall and cascades.
• Quad-witching gappy open. Standard model shows an ambiguous read because the OI is being rolled. A flow-based estimate picks up the sign change intraday, while the OI model will not have it until the next daily snapshot.

What a flow-based estimate lets you do

• Read walls correctly. Walls in a flow-based estimate are structural. Walls in the assumption model are sometimes just OI concentrations with the wrong sign attached.
• Trust the zero-gamma flip. Zero-gamma is only meaningful if the underlying positioning estimate is right. A flow-based estimate is where you actually see the sign change.
• Interpret sudden character shifts. When the tape’s character changes mid-session without an obvious catalyst, a flow-based estimate usually has a story - a large systematic block, a dispersion unwind, a fund closing a hedge. The assumption model has nothing to say.
• Size 0DTE trades with the right regime. On 0DTE-heavy sessions, the flow-based sign is the one that predicts pin vs cascade. Sizing pin trades in a mis-signed regime is how accounts get wiped out on Fridays.

What a flow-based estimate does not do

It does not tell you where SPX will go. It is a regime layer, not a directional signal. It does not fix bad execution or replace a thesis. It is not a shortcut to profits - it is a correct input to a discretionary or systematic process that was previously using a wrong input.

It also does not cover positioning outside of the SPX chain. Futures positioning, cross-asset hedges, and dispersion books are not captured. What it does capture is what dealers hold on the SPX chain right now, and that is enough to reshape most intraday trading decisions.

See live SPX dealer positioning and gamma exposure on the SPX GEX page. Includes both the OI-based and flow-based views side by side.

Frequently asked questions