Gamma exposure (GEX) explained, and what we could and could not test on Nifty options
"Positive gamma, sell the range; negative gamma, buy the breakout." The idea has spread from US markets to Indian option-trading groups. Before using it, it helps to know what gamma exposure really measures, what an Indian trader can and cannot observe, and what a careful back-test shows.
Gamma in one paragraph
An option's delta says how much its price changes when the underlying moves one point. Gamma says how fast delta itself changes. Market makers who sell options to clients hold the opposite position and keep their total delta near zero by trading the underlying (futures or stock). If the market makers are net long gamma, a rise in price makes their delta positive, so they sell the underlying to rebalance, and a fall makes them buy. That trading leans against moves and tends to dampen volatility. If they are net short gamma, they must buy into rises and sell into falls, which amplifies moves.
What "GEX" aggregates
Gamma exposure is the sum, over all strikes and expiries, of open interest × gamma × contract size, with a sign. Plotting it against the underlying level gives a curve; the level where it crosses zero is the flip line. Above it dealers are supposed to be long gamma (calm, mean-reverting); below it, short gamma (jumpy). The largest single-strike contributions are called gamma walls.
The assumption hidden in the sign
The open interest data published by exchanges shows how many contracts are open at each strike, not who holds them. The standard GEX recipe therefore assumes dealers are long the calls and short the puts (customers buy puts for protection and sell calls for income). In the US, with a large body of research and some vendor data on dealer flow, this is an approximation. In India, we have no published dealer-positioning data, and a lot of volume is from proprietary firms, retail sellers of options and hedgers whose side cannot be inferred. A different sign assumption changes the flip line and the regime. That is the first reason to treat any Indian GEX number as an assumption-driven estimate.
Weekly options make it harder
Most Nifty option activity is in weekly expiries, whose gamma is very large close to expiry and very time-sensitive. A profile calculated from the previous close can be out of date by 11 a.m., because open interest, implied volatility and time to expiry have all changed. A dealer-based picture needs intraday open interest by strike, which our provider does not supply historically.
What we could actually test
A trading brief we received described a full system built on live GEX from a validated data vendor, with tick-level option history for back-testing. We have neither. What we could do, from public exchange data:
- Download the end-of-day Nifty option files published by NSE for December 2020 to October 2026 (2.6 million option rows).
- For each session, rebuild a gamma curve from the previous close's open interest, with each strike's own implied volatility, using the call-positive / put-negative convention. Use the curve through the next day to classify the regime (positive or negative gamma at the current price) and find the flip line and walls.
- Trade the brief's structure around that classification: credit spreads at the prior-day or opening-range extremes in positive gamma, and breakout debit spreads in negative gamma.
We called this version "v1D" (daily). A result on v1D says nothing about a true intraday system built on vendor data; it tests only what an independent trader can build from free data.
What we found
- The regime was mostly positive. Under this convention, about 87% of sessions closed with positive net gamma, and the flip line sat a median 0.24% below the close. Negative-gamma "breakout" conditions were rare: about four debit-spread trades in 1,000 sessions.
- The credit-spread side produced about +₹14 a trade before charges of about ₹98 in development (137 trades), and a negative gross result in validation (28 trades, about −₹182). Net results were small losses: −₹17 and −₹25 a day on ₹10 lakh.
- A version without gamma did no worse. Replacing the gamma regime with a simple rule (price and implied-volatility direction) produced −₹23 a day in development and −₹3 in validation. So in this data the gamma classification added no measurable value.
- The brief's constraints conflicted: a 0.20-delta credit spread collects about 14% of its width, so the 25% rule rejected almost every signal: see why.
What this does not show
- It does not show that dealer gamma does not matter. It shows that a daily, open-interest-based approximation under a sign assumption had no detectable edge in these option spreads after costs.
- All spreads were partly model-priced, because the strikes the rules wanted lie outside the observed data window.
- Only Nifty was tested; Bank Nifty, Sensex and MCX crude need their own option history.
- The sample is 165 trades. A small real effect could be missed.
How to use GEX responsibly
- Treat any GEX figure as an estimate under stated assumptions; find out the sign convention, the expiries included and the update time.
- Never size a trade on the flip line alone; combine it with defined-risk structures and a stop that does not depend on the model.
- Ask for out-of-sample evidence that includes costs: a plausible mechanism is not a proof.
- Remember that on expiry day, gamma dominates everything, and prices can move well beyond the "walls".
The full audit of the study, every trade and the ablation without gamma, is on the Strategy Review. See also how we test trading ideas. This is educational research; it is not a recommendation to trade options.