What is zero gamma in trading?

Zero gamma is the price at which the estimated gamma of every outstanding option cancels out. Above it, the hedging that dealers are forced to do leans against price. Below it, the same hedging leans with price. It is one number, it moves every night, and it is the single most misread figure in options positioning data.

Data on this page: the newest snapshot stored for each ticker. Open interest — the number of contracts still outstanding — is published once per trading day, after the close, so every board here is a settled one rather than a live one. The most recent settlement date in this set is 2026-08-31. Each row in the table at the bottom of this page carries its own settlement date. The written explanation was last revised on 2026-07-29.

The short answer

Two things follow immediately, and both are worth fixing in your head before anything else.

• It is a boundary, not a destination. Nothing in the calculation says price will travel to it, stop at it, or bounce off it.
• It is not a place where gamma is absent. The gross amount of gamma at that price is usually enormous. The positive side and the negative side simply happen to cancel.

Zero gamma and the gamma flip are the same thing

You will see this number called zero gamma, the gamma flip, the flip level, the zero gamma level, and occasionally the gamma inversion point. They all mean the same price. Nobody has ever drawn a distinction between them that survived contact with the arithmetic, and this page uses the two main terms interchangeably for exactly that reason.

Where the number comes from is the subject of a longer guide: what is gamma exposure works through why a dealer who sold you an option has to keep trading stock, and how that becomes a dollar figure. The one-paragraph version: a market maker who is short a call hedges by holding stock, the amount of stock they need changes as price moves, and gamma is the rate of that change. Sum it across the board, count calls positive and puts negative, and you have net GEX. Zero gamma is the price at which that sum would be zero.

The double name causes trouble because the two words point at different halves of the same idea. Zero gamma describes the arithmetic: a sum that happens to equal zero at that price. Gamma flip describes the consequence: walk through that price and the sign of the sum changes, so the direction of the forced hedging changes with it. Readers who met one term first often assume the other must be a second, separate level and go looking for two lines on the chart. There is one line, and both names are pointing at it.

What is genuinely not standardised is the number itself. Two providers can publish different zero gamma levels for the same ticker on the same night without either being careless, because the calculation depends on choices nobody has agreed on: which expirations go in, how wide a price range the crossing is searched over, where implied volatility comes from, and whether that volatility is allowed to change as the search price moves. Every one of those choices shifts the answer. Ours are written out step by step further down this page, so you can see which version of the number you are reading rather than having to trust a single decimal.

How to read a gamma exposure chart

This is MSFT as of 2026-08-31, rebuilt from that evening’s open interest. Every bar is one strike.

• Horizontal axis: strike prices. Not time. A gamma exposure chart is a snapshot of the option board laid out by price, which is why it looks nothing like a candlestick chart and should not be read like one.
• Vertical axis: estimated gamma, with zero in the middle. Bars above the line are strikes where dealers are estimated to be long gamma; bars below are strikes where they are estimated to be short. On this snapshot 31 of the 54 strikes plotted sit above the line and 23 sit below.
• The dashed line is the last price ($503.39), and the solid line is zero gamma ($462.42) — here it sits -8.14% from the last price, below it.
• Zero gamma is not the smallest bar. This is the mistake almost everyone makes on their first chart. It is not read off any single strike; it is where the running total of all the bars crosses zero, which is why it gets its own vertical line. A strike can carry almost no gamma and be nowhere near the flip.
• Tall bars are concentrations, not walls that hold. They mark where hedging intensifies. That is a fact about positioning, not a fact about price.

Net gamma across the whole board here is $946M per 1% move, a positive reading. The chart is rebuilt nightly, so the shape you are looking at is the most recent one we hold rather than a fixed illustration.

Zero gamma does not mean there is no gamma

This is the misreading that does the most damage, and the name invites it. A total of zero is not the same as nothing being there. At the flip price the positive gamma sitting in one part of the board and the negative gamma sitting in another happen to be the same size, so they cancel in the sum. Both are still there, and both are still large. Nothing has switched off.

The snapshot above says it with real numbers. The MSFT board on 2026-08-31 carries $1.81bn of gross gamma — every strike added up without letting the signs offset each other — while the net figure, the one that has to cross zero for a flip level to exist, is $946M. The net is 52.4% of the gross. The other 47.6% is calls cancelling puts.

That is not a quirk of one ticker. Across the 43 tickers with a gamma estimate in the current snapshot set, the median net figure is 27.6% of gross. The cancelling is the normal state of an option board, not an unusual configuration.

Two practical consequences follow, and they are the reason this section is not just pedantry.

• A small net does not mean a quiet chain. A ticker can be sitting almost exactly on its flip level with an enormous amount of hedging still tied to the board. The net figure tells you which way the hedging leans, not how much of it there is. Gross gamma is the size question, and it is reported separately for that reason.
• A small net is a fragile net. When the surviving total is a low single-digit percentage of the gross, a modest change in open interest — one busy day at a couple of strikes, one expiration rolling off — is enough to move the crossing a long way, or to remove it. That is the mechanical reason the level is not stable from one snapshot to the next, which is what the next section counts.

Above it, below it, and sitting on it

The third row is the one worth dwelling on. When price is hovering at the flip, a single unremarkable day is enough to carry the whole board from the damping regime to the amplifying one. It is also the configuration in which a snapshot from last night is least reliable, because the thing it is describing is the thing most likely to have already changed.

Both readings are in the table at the bottom of this page: where the crossing sits, and how far the last price is from it. How long either regime lasts, and how large a change in open interest it takes to carry the crossing past the last price, are things we have not measured. What the section below does instead is count how far the level actually moved between the snapshots we have stored.

How far zero gamma moves between snapshots

Zero gamma is not a property of a company. It is a property of one option board on one evening, and every input to it resets overnight. Contracts expire and leave the sum. New open interest appears at new strikes. The share price moves, which changes which strikes are close enough to the money to carry meaningful gamma at all. Implied volatility changes, which changes how much gamma each contract contributes. A level built on four moving inputs is a level that moves.

We keep each night’s snapshot as it is produced, so this can be counted rather than asserted. Across the 575 pairs of consecutive snapshots currently stored — pairs no more than four calendar days apart, so weekends are included but longer gaps in the archive are excluded — the estimated zero gamma level moved a median of 1.0% from one snapshot to the next. 296 of those 575 pairs moved by more than 1%. The largest single move in the sample was MSFT, 20.9% between 2026-07-28 and 2026-07-29.

The pairs in this sample run from 2026-07-22 to 2026-08-31, so it is a small and recent one, and it covers only the 42 tickers on this site that had a zero gamma level in two nearby snapshots — not the whole market. 442 of the 575 pairs are one calendar day apart, and because the same tickers contribute on each date these are not 575 independent observations. It is a count of what the stored snapshots did. It is not an estimate of what the level will do tomorrow and should not be read as one.

Hold that figure next to the distances in the table at the bottom of this page. Across the 40 tickers there that have a level at all, the median sits 2.4% away from it, against a median change of 1.0% in the level itself between consecutive snapshots. Compare the two before treating “price is above the flip” as a durable description of anything. It is also why every figure here carries a date instead of being called live.

How we estimate the level, and what would make it wrong

A flip level is easy to publish and harder to explain. Here is ours in order, so you can judge how much weight the number deserves. None of it is proprietary; the whole calculation is arithmetic on public data plus one textbook pricing model.

1. Start from settled open interest, out to 45 days. For every listed contract on the ticker expiring within 45 days we take the open interest published after the close, together with that evening’s bid and ask. Where both sides of the quote exist we use the midpoint; where they do not, the closing price. Expirations further out are never fetched, so nothing about them enters any step below.
2. Back out an implied volatility for each contract. Implied volatility is the volatility figure that makes a pricing model agree with the price the contract is actually quoted at. We solve for it by bisection against a Black-Scholes model with the interest rate and the dividend yield both set to zero. Contracts quoted at or below their intrinsic value are dropped, because they carry no time value and therefore no usable gamma, and so are contracts whose quote the model cannot reach even at 500% volatility.
3. Refuse to publish on a thin chain. If fewer than 100 contracts survive that step, no gamma estimate is published for that ticker at all — not a zero, not a blank row, no page. A number built on a handful of stale quotes would look exactly like a number built on a liquid chain, and there would be no way for a reader to tell them apart.
4. Re-price the whole board at each candidate price. The crossing is searched for on a 121-point grid running from 0.90 to 1.10 times the last traded price. At each grid price every surviving contract’s gamma is recomputed for that price, multiplied by its open interest, by the 100 shares a contract controls, and by the dollar value of a 1% move, then calls are added and puts subtracted. This matters: the level is not read off the by-strike bars in the chart above, it is a fresh sum computed at each candidate price.
5. Take the crossing nearest the last price. Wherever two neighbouring grid points carry opposite signs, the sum passed through zero somewhere between them, and interpolating gives the price. If a board produces more than one crossing, the one closest to the last traded price is the one published.

Six things in that recipe can put the level in the wrong place. Listing them is fairer than letting two decimal points imply a precision that is not there.

• The sum stops at 45 days. Expirations beyond that are not fetched, so they cannot enter the total or move the crossing. This is a near-dated flip level rather than a whole-chain one, and a provider who sums every listed expiration will get a different number for the same board on the same night.
• The dealer convention is an assumption, not a measurement. Counting calls positive and puts negative encodes a belief that dealers are long every call and short every put. Open interest never reveals who holds which side. Where that belief is wrong for a ticker, the sign of the sum is wrong, and a level derived from a wrong sign is not a level.
• Volatility is frozen while the search price moves. Each contract’s implied volatility is solved once, at the last traded price, then held fixed as the search walks up and down the grid. Real option markets do not behave that way — implied volatility moves when price moves — and a volatility surface that shifted along with the price would put the crossing somewhere else. This is the largest shortcut in the whole calculation.
• Zero interest rate and zero dividend. A convention that keeps the estimate reproducible without a rates feed. Both terms enter Black-Scholes through the time to expiration, so the gap between this convention and a rate-aware model widens the longer a contract has left to run. That is a property of the formula, not something we have measured against real quotes. On the current snapshot set, 40.5% of the gross gamma in the sum sits in contracts expiring within two weeks and the remaining 59.5% in the 15-to-45-day contracts at the far edge of what we fetch.
• The search window is ±10% of the last price. A board whose only crossing sits outside that window is reported as having none. That is a limit of the search, not evidence that no such price exists, and we report it as an absence rather than extrapolating past the edge of the window.
• Quote quality feeds straight through. A wide or stale bid-ask on an illiquid strike produces a wrong implied volatility, which produces a wrong gamma for that strike, which shifts the sum. How much one bad quote moves the total depends on where the gamma sits. On the current snapshot set the five strikes carrying the largest net gamma add up to a median of 30.7% of a board's gross gamma. We do not score quote quality strike by strike, so that is a statement about concentration, not a bound on the error.

When there is no zero gamma level at all

Two completely different situations leave a ticker with no flip level to quote, and they mean opposite things about the data. Anywhere that reports both as an empty cell is discarding the more interesting of the two.

• The chain is too thin to estimate gamma at all. Fewer than 100 contracts survive the implied volatility step, so there is no gamma estimate for a crossing to live in. In the current snapshot set that applies to 0 of the 43 tickers computed here. Those tickers get no gamma page rather than a page of blanks.
• The chain is fine, but net gamma never changes sign. The board is liquid, the sum is computed at all 121 grid prices, and it stays positive — or stays negative — the whole way across. There is no crossing because the sign never changes anywhere inside the search window. What arranges a board that way is a property of how its open interest is distributed, which we do not decompose here. In the current snapshot set 3 of the 43 tickers with a gamma estimate are in this state (GME, NIO, SNAP). That is a real description of the board, not a failure: the entire searched range reads as a single regime.

The paid GET /v1/gamma/{symbol} endpoint reports the second case as an explicit flip_status rather than a bare null, precisely so that an AI agent reading the response does not collapse it into the first case and conclude the data is missing.

The count is a property of the boards on the night rather than of the ticker list. On the current snapshot set the calculation resolves a zero gamma level for 40 of the 43 tickers computed here. If you see a site that always has a flip level for every ticker every day, that is worth a question, not a compliment.

What zero gamma does not tell you

The hedging mechanism is real. Dealers do hedge, the flows are large, and they are forced rather than discretionary. But several things people routinely read into this number are not supported by the calculation that produces it.

• It is not a direction. Negative gamma means wider ranges, not lower prices. A market can spend a week below zero gamma and rally the whole time.
• It is not a support or resistance level, and we have not measured whether price respects it. That is a separate empirical question, and answering it honestly needs a null — a comparison against randomly placed levels tested the same way — because price bounces around near any price you nominate. Until that test is run and published, a hit rate for the flip level is a number nobody has earned the right to quote, including us.
• It rests on an assumption that is a convention, not a measurement. Every public GEX figure, ours included, assumes dealers are long every call and short every put. Open interest never reveals who holds which side. Where that assumption is wrong for a ticker, the sign of the estimate is wrong too.
• It cannot update intraday. Open interest settles overnight. Any figure you see during the session — here or anywhere else — describes the previous close’s positioning, however live the dashboard looks.
• Sometimes there is no zero gamma level at all. If gamma keeps the same sign right across the searched strikes, there is no crossing to quote — a real state of the board rather than a broken calculation, and a different situation from a chain too thin to estimate at all, as the section above sets out.

The question that follows is whether anyone has evidence either way. There is academic work on mechanical hedging flows — Baltussen, Da, Lammers and Martens, “Hedging demand and market intraday momentum”, Journal of Financial Economics 142(1), 2021, pp. 377–403, studies how hedging that has to happen relates to intraday price behaviour. It is worth reading. It is also not a test of the thing on this page: it does not measure whether a flip level estimated from public open interest, under the assumptions listed above, gets respected as a level. A literature on hedging flows existing is not permission to quote a hit rate for our number, and we do not quote one.

Where zero gamma sits right now

Every ticker with a chain deep enough to estimate gamma, computed nightly from end-of-day open interest and closing quotes. The last column carries each ticker’s own settlement date. Of the 43 estimated, 40 have a zero gamma level inside their traded strikes — 20 above the last price and the rest below — and 5 are within 1% of it, the unstable case. The other 3 have no crossing at all.

The flip is also the core field of the paid GET /v1/gamma/{symbol} endpoint — it returns flip plus a flip_status that names the no-crossing state, so an AI agent can read the latest nightly level directly (API reference).

Common questions

What is zero gamma in trading?

Zero gamma is the price at which the estimated gamma of every outstanding option nets out to nothing, so the hedging dealers are forced to do neither damps nor amplifies the move. It is the same number people call the gamma flip level. Above it, hedging is estimated to lean against price and ranges tend to compress; below it, hedging leans with price and ranges tend to widen. It marks a boundary between two behaviours, not a level price is expected to reach.

Is zero gamma the same as the gamma flip level?

Yes. The two terms describe the same price and are used interchangeably, sometimes in the same sentence. You will also see "gamma flip", "flip level", "zero gamma level" and "gamma inversion point" for it. None of them means anything different.

How do you read a gamma exposure chart?

Strikes run along the horizontal axis and estimated gamma runs vertically, with zero in the middle. Bars above the centre line are strikes where dealers are estimated to be long gamma, bars below are strikes where they are short. Zero gamma is not where a bar is small; it is the price where the running total of all those bars crosses zero, which is why it is drawn as a separate vertical line rather than read off a single bar.

What does it mean when price is below zero gamma?

The estimate says dealers are net short gamma, so their hedging runs with price rather than against it: they buy into strength and sell into weakness. Mechanically that tends to widen ranges and let moves extend. It describes the volatility environment the estimate places the market in. It is not a directional signal and says nothing about whether price will rise or fall.

Does zero gamma actually work as a support or resistance level?

We have not measured that, so we do not claim it. The hedging mechanism behind the number is real and well documented, but "price respects the flip level" is a separate empirical claim that needs testing against a null of randomly placed levels before anyone should believe a hit rate. Treat zero gamma as a description of the current regime, not as a level to trade against.

Read next

What is gamma exposure (GEX)? is the longer version of the mechanism behind this number — delta, then gamma, then the arithmetic, then the assumption underneath every free GEX figure on the internet. What is max pain? covers the other calculation built on the same open interest, and max pain vs gamma exposure shows what a disagreement between the two is actually telling you.

Zero gamma here is a derived estimate computed nightly from end-of-day open interest and closing quotes using a Black-Scholes approximation with zero rate and zero dividend (how we compute it). Not raw market data, not a live quote. Educational only: not investment advice, not a recommendation, not a price forecast.