Dealer gamma exposure links signed option inventory to local hedge adjustments. Work through shares, dollar GEX and gamma P&L, then examine ownership assumptions and gamma flips.
Dealer gamma exposure
Dealer gamma exposure measures the signed sensitivity of dealers’ aggregate option delta to a change in the underlying price. It translates an option book’s gamma into a local estimate of the stock or futures adjustment needed to maintain a delta hedge. When dealer inventory is inferred from public data, the result is an estimate conditional on that inference.
The practical question is concrete: if spot rises, would this assumed book need to buy or sell more of the underlying? The answer can help assess a potential source of market feedback. It does not establish how much trading will occur or where prices will go.
Start with the dealer’s signed position
For options on one stock, let ni be signed dealer contracts in series i: positive for long, negative for short. Let mi be shares per contract and Γi=∂Δi/∂S the long option’s gamma per underlying unit. Aggregate share gamma is
Gshares=i∑nimiΓi.
Its units are shares of delta change per USD 1 spot move. For a small change dS, with time, volatility and other pricing inputs fixed,
dΔoptions, shares≈GsharesdS,dh≈−GsharesdS.
Here h is the stock hedge in shares, so dh>0 means buying additional shares. The second equation assumes the desk restores delta neutrality with stock and leaves the option book unchanged. It gives the incremental trade, not the hedge’s total size. Initial option delta would be needed to calculate that total.
For ordinary vanilla options, a long call and a long put both have positive gamma; selling either reverses the sign. The OIC’s gamma explanation supports this distinction. A call-versus-put label alone does not determine dealer gamma.
One short book, three different numbers
Consider a stipulated dealer position, not observed inventory: short 100 option contracts, each covering 100 shares, with long-option gamma 0.02 per dollar and stock at USD 100. Freeze all other inputs and use initial gamma throughout the local calculation.
Gshares=(−100)(100)(0.02)=−200shares per USD.
A 1% rise is a USD 1 rise. Option delta changes by approximately −200 shares, so restoring neutrality requires buying 200 shares. A separate 1% fall reverses both signs:
Separate spot scenario
Spot change
Option delta change
Incremental neutralising stock trade
USD 99
−USD 1
+200 shares
Sell 200 shares
USD 101
+USD 1
−200 shares
Buy 200 shares
These are local approximations. Actual gamma can change between 100 and either endpoint.
For a stipulated short book of 100 options, a 1% rise from spot USD 100 requires buying 200 hedge shares and a 1% fall requires selling 200. The matched long book requires the opposite trades.
Original frozen-input scenarios, not market observations. Multiplier 100 shares; long-option gamma 0.02 per dollar. Lines use constant initial gamma and show incremental trades from the same starting hedge.
State the GEX convention before reading the number
One dollar-scaled convention reports the gamma-driven change in option dollar delta for a 1% spot rise, valuing the delta change at the reference spot:
GEX1%=0.01S2i∑nimiΓi=0.01S2Gshares.
For this book,
GEX1%=0.01(100)2(−200)=−20,000USD per 1% move.
One factor of S converts 1% into a dollar spot move; the other values the resulting share-delta change. The corresponding neutralising purchase has reference-spot notional +USD 20,000. This is not an execution bill: buying 200 shares at 101 would cost USD 20,200 before fees. Nor is it the full change in SΔ: that quantity also changes because the existing delta is valued at a new spot. The stated GEX isolates the gamma contribution.
It is also not gamma profit or loss. The short option book’s quadratic Taylor contribution for a USD 1 move is
21Gshares(dS)2=21(−200)(1)2=−100USD.
An initially delta-neutral book with its stock hedge held fixed has this approximate instantaneous curvature residual, before other effects. The option book alone also has a first-order delta term. Neither calculation is a realised trading-period profit estimate. “Gamma exposure” and dollar gamma labels vary across reports, so compare signed positions, multipliers, units and bump sizes before comparing their magnitudes.
Why the feedback can change sign
If dealers are net long gamma, a rise makes their option delta more positive: the neutralising trade sells underlying. A fall leads to buying. Those trades lean against the price move and can dampen it.
If dealers are net short gamma, the same logic gives buying after a rise and selling after a fall. Those trades follow the move and can amplify it. The figure reverses only the stipulated position sign to show this mechanism.
The qualification is economic, not just statistical. Dealers may hedge in bands, net exposures across products, change option positions, use futures or accept temporary delta. Other market participants may supply or demand much more liquidity. Converting an estimated hedge target into a price-impact forecast requires assumptions about execution and market depth. A negative GEX number does not by itself predict a falling market: the short-gamma adjustment follows either direction of spot movement.
Cboe’s May 2025 report summary provides a useful empirical counterweight: for its study of SPX same-day options, balanced customer activity left estimated net market-maker gamma hedging small relative to market liquidity. That dated, product-specific finding is not a universal result for every option market or period.
A public chain is not a dealer inventory ledger
Open interest counts outstanding contracts. Each contract has a long side and a short side; the count alone cannot identify the dealer’s side. The OIC’s open-interest FAQ illustrates one outstanding contract with a buyer and seller, and explains why opening and closing trades affect the count differently.
Thus a calculation using positive call open interest and negative put open interest is imposing an ownership heuristic. It is not measuring positive call gamma and negative put gamma, and it is not observing dealer ownership. Two books with identical option-chain open interest can give opposite dealer GEX if the assumed dealer side is reversed. Even trade-based inference needs an explicit method for identifying counterparties and whether risk was opened or closed.
A reproducible estimate should identify the inventory inference, data timestamp, included expiries and strikes, contract specifications, and Greek model. End-of-day open interest cannot describe every subsequent intraday position change. Expiring positions and offsetting risk outside the selected chain can materially change the aggregate.
A gamma flip is a spot level where a recalculated aggregate changes sign under those assumptions. Computing it requires a whole curve Gshares(S), not just today’s scalar. The modeller must specify what happens to implied volatility as spot changes—for example, a sticky-strike surface keeps each strike’s implied volatility fixed. Different surface rules, expiry coverage or inventory signs can move the zero, remove it or produce several zeros. It is a model output rather than a price barrier.
Finally, spot is not the only reason delta changes. Volatility changes and time passage can alter it through vanna and charm. For those sensitivity definitions and their calculation, Greeks and Numerical Sensitivities has a public preview; the full lesson requires membership. It supplies the Greek mechanics rather than a method for recovering dealer inventory.
