Study · Original data
Do GEX levels really pin Bitcoin? A test against control levels
Across 23 days of Bitcoin data, price reached GEX levels no more often than ordinary price levels at the same distance (32% vs 32% within 8 hours) and bounced no more often once there (52% vs 61%): we found no reliable sign that GEX levels pin price.
Key findings
- Bitcoin: price came within 0.15% of a GEX level within 8 hours in 347 of 1,077 level-hours (32.2%), versus 32.2% for ordinary levels at the same hour, side and distance (difference +0.0 points, 95% CI -2.3 to +2.5).
- When Bitcoin touched a GEX level it bounced 52% of the time (73 of 140 decided touches) versus 61% at round, midpoint and offset levels (168 of 276); difference -8.7 points (95% CI -18.9 to +0.6).
- Ethereum and Solana look the same: bounce rates 50% vs 51% (ETH, 103 vs 186 touches) and 49% vs 51% (SOL, 72 vs 118), and reach rates within 1.4 points of their controls.
- A bounce rate near 50% is what the same rule produces on a randomly shuffled price path (53% for Bitcoin), so GEX levels behaved like arbitrary lines rather than walls.
- On the options index price GEX levels were reached slightly more often than controls (+2.3 points across all three coins, 95% CI +0.9 to +4.1), but on the futures price the gap disappeared (+0.4, -1.5 to +2.4): at most a small, feed-dependent effect.
- In the 4 hours before the daily 3:00 AM ET (08:00 UTC) expiry, Bitcoin GEX levels held 64% (16 of 25) versus 58% for controls, but the 95% CI (-10 to +19 points) includes zero, the rate fell from 80% to 53% between the first and second half of the period, and Ethereum (47%) and Solana (20%) showed nothing.
About the data. GEX levels were recorded by Swiftedge every minute from 2026-10-07 to 2026-10-30. The study looks back at that period; it does not describe any live GEX level.
Does price get pulled to GEX levels?
| Market | Days | Level-hours | GEX reached within 8 h | Control reached | Difference (95% range) |
|---|---|---|---|---|---|
| Bitcoin | 23 | 1,077 | 32% | 32% | +0.0 pts (−2.3 to +2.5) |
| Ethereum | 10 | 355 | 50% | 49% | +1.4 pts (−2.5 to +6.4) |
| Solana | 6 | 187 | 47% | 46% | +0.5 pts (−3.7 to +4.1) |
Does price bounce off them?
| Market | GEX touches | Held | Control touches | Held | Difference (95% range) |
|---|---|---|---|---|---|
| Bitcoin | 140 | 52% | 276 | 61% | −8.7 pts (−18.9 to +0.6) |
| Ethereum | 103 | 50% | 186 | 51% | −1.6 pts (−14.1 to +10.1) |
| Solana | 72 | 49% | 118 | 51% | −2.2 pts (−13.5 to +7.6) |
By level type
| Market | Level | Level-hours | Reached | Control reached | Held (touches) |
|---|---|---|---|---|---|
| Bitcoin | flip | 217 | 20% | 22% | 40% (10) |
| Bitcoin | call wall | 76 | 26% | 22% | 44% (25) |
| Bitcoin | put wall | 148 | 37% | 36% | 42% (33) |
| Ethereum | flip | 61 | 74% | 75% | 47% (38) |
| Ethereum | call wall | 54 | 39% | 42% | 57% (21) |
| Ethereum | put wall | 97 | 35% | 35% | 63% (27) |
| Solana | flip | 50 | 38% | 36% | 30% (10) |
| Solana | call wall | 41 | 22% | 20% | 50% (6) |
| Solana | put wall | 43 | 28% | 26% | 32% (19) |
How it was measured
Data. Swiftedge's own once-a-minute GEX snapshots, computed from public options and futures data, for Bitcoin (2026-10-07 to 2026-10-30, 23 days), Ethereum (2026-10-20 to 2026-10-30, 10 days) and Solana (2026-10-24 to 2026-10-30, 6 days). Each snapshot holds the gamma flip, call wall and put wall, same-day (0DTE) and trade-flow versions of them, and the strike with the most gamma. Prices are 1-minute closes of the perpetual futures; every test was repeated on the options index price stored in the snapshots. Only information available at the time is used: a level counts from the minute it was published.
Test 1: are prices pulled to the levels? Every full hour we took each GEX level between 0.4% and 2.0% away from price and checked whether a 1-minute close came within 0.15% of it, or crossed it, in the next 8 hours (4 and 24 hours as checks). Each GEX level was compared with ordinary price levels (round numbers, midpoints between them and offset levels) at the same hour, on the same side of price and in the same 0.25%-wide distance band.
Test 2: do prices bounce at the levels? We used the same rule as Swiftedge's public track record. A touch is a 1-minute close within 0.15% of a level after price had been at least 0.4% away within the previous 6 hours. The touch held if price then moved 0.5% back the way it came before a 1-minute close 0.3% through the level, and broke if the opposite happened first (2-hour window, level frozen at the touch). Control levels (the nearest round number, the midpoint and an offset level) went through exactly the same rule. Touches within 30 minutes of each other count once per group, so several GEX levels sitting at the same price are not counted several times.
Statistics. Single bounce rates carry 95% Wilson intervals; differences and the hourly reach rates carry bootstrap intervals that resample whole UTC days, because hours within a day are not independent. A logistic regression with approach speed, side and recent volatility checked whether 'is a GEX level' adds anything (coefficient -0.13, 95% CI -0.38 to +0.12), and a shuffled price path shows what the bounce rule gives when no real levels can exist.
Limits of this study
- Short window: 23 days for Bitcoin, 10 for Ethereum and 6 for Solana. The snapshot recording ended on 30 October 2026, so the sample cannot grow.
- One market regime: Bitcoin traded above its gamma flip almost the whole time, so a negative-gamma market is untested.
- Dealer positioning is estimated (dealers assumed long calls and short puts, plus a trade-flow variant); other assumptions give other levels.
- Absolute bounce rates depend on the price feed (Bitcoin control levels 61% on futures prices vs 55% on the index price); the GEX-versus-control comparison points the same way on both.
- The gamma flip's own track record is too thin to quote: 8 of 11 touches held on the index price but 4 of 10 on the futures price.
- Many sub-results were checked (three coins, eleven level types, three horizons, two price feeds). Isolated 'significant' cells, such as Bitcoin put walls on the index price, are expected by chance and did not repeat on the other feed or coins.
- Reach rates are counted per level and hour, so one level appears in many hours; the intervals resample whole days.
- Descriptive research on past data; not a trading signal and not advice.
Cite this
Swiftedge Research (2026). Do GEX levels really pin Bitcoin? A test against control levels. https://swiftedge.uk/research/gex-pinning-test
<a href="https://swiftedge.uk/research/gex-pinning-test">Do GEX levels really pin Bitcoin? A test against control levels - Swiftedge Research</a>
Common questions
What is GEX?
Gamma exposure: an estimate of how much options dealers would have to buy or sell as price moves, built from options open interest. Popular GEX levels are the zero-gamma flip and the largest call and put strikes.
Do GEX levels pin price?
Across 23 days of Bitcoin data, price reached GEX levels no more often than ordinary price levels at the same distance (32% vs 32% within 8 hours) and bounced no more often once there (52% vs 61%): we found no reliable sign that GEX levels pin price. The tables show each market.
Why compare with control levels?
Price touches many levels by chance, and round numbers attract orders. A level only means something if it is reached or defended more often than an ordinary level at the same distance from price.
What about the hours before options expiry?
In the four hours before the daily 3:00 AM ET (08:00 UTC) expiry, Bitcoin's GEX levels held a little more often than controls, but the difference was within the noise and did not appear in the other markets.