Options expiration day has a reputation. Traders talk about "opex chaos", a "pin" into the close, triple-witching selloffs and a volatility "unclamp" the following Monday. We tested seven of these claims against 28 years of S&P 500 data. Most of them are false. Monthly opex is actually a little quieter than a normal Friday.
Data: SPY daily bars 1998–2026, continuous S&P 500 futures (/ES) daily bars 2010–2026, and /ES one-minute bars during regular hours (9:30–4:00 ET) from February 2021 to October 2026. Results marked (inference) are BullishAgent's own reading. Not investment advice.
The short version
| Claim | Verdict | What the data shows |
|---|---|---|
| Opex days have more range and volatility | False | Monthly opex range is 7–11% smaller than other Fridays. Quarterly opex is the same as other Fridays. |
| The open is wild on opex | False | The first 30 minutes are no bigger than on other Fridays. |
| Price "pins" into the close | Weakly true | Only on monthly opex, and only since 2022. The last hour is about 18% quieter. |
| Triple witching days are bearish | Leans true | Average −0.24% with only 38% up days, but the result is borderline. |
| Opex week is bullish | False | It returns less than an average week. |
| The week after opex is bearish | False | After monthly opex it is up 68% of the time, above the 61% average. |
| Volatility "unclamps" the Monday after | False | The range is normal and VIX rises less than on other Mondays. |
What opex is
Most US stock and index options expire on the third Friday of each month. If the market is closed that day (Good Friday, or Juneteenth in 2026), they expire on the Thursday before. In March, June, September and December, stock options, index options and index futures all expire on the same day. That is called quarterly opex or triple witching.
The theory behind the folklore: dealers who sold those options hedge them by buying and selling stock. When billions of dollars of options expire at once, that hedging has to unwind, so the day should be unusual.
How we tested it
- The comparison is other Fridays, not all days. Fridays already behave differently from Mondays, so comparing opex to "all days" would be unfair.
- Range is measured against its own recent normal. A 60-point day means something different when the market has been moving 40 points a day than when it has been moving 100. Each day's high-to-low range is divided by the average range of the 20 trading days before it.
- Returns use S&P 500 futures, not SPY. The reason is in the "dividend trap" section below.
- Significance is checked with a shuffle test. We drew thousands of random sets of ordinary Fridays the same size as the opex group and asked how often chance alone produced a gap as big as the one observed. That is the p value; below 0.05 is the usual bar.
Myth 1: opex means more range — false
| Day type | Days | Range vs. its 20-day normal |
|---|---|---|
| Monthly opex | 128 | 0.93× |
| Quarterly opex (triple witching) | 64 | 1.04× |
| Other Fridays | 627 | 1.07× |
| Day after opex | 192 | 0.97× |
| All non-opex days | 3,920 | 1.03× |
S&P 500 futures, full session, 2010–2026.
Monthly opex range is about 11% smaller than on other Fridays (p = 0.017). SPY's 28-year history gives the same answer (p = 0.012). Triple witching is no different from an ordinary Friday (p = 0.73).
So the claim gets it backwards. Opex days are not bigger. Monthly ones are slightly smaller.
Myth 2: the open is wild — false
Monthly SPX options settle on the opening prices of the stocks in the index, so you might expect a scramble at 9:30. There isn't one. The first 30 minutes of opex look the same as other Fridays (monthly p = 0.94, quarterly p = 0.20).
Myth 3: the pin into the close — weakly true
| Window (ET) | Monthly opex | Quarterly opex | Other Fridays |
|---|---|---|---|
| 9:30–10:00 | 0.41× | 0.46× | 0.41× |
| 10:00–12:00 | 0.58× | 0.70× | 0.68× |
| 12:00–14:00 | 0.45× | 0.47× | 0.47× |
| 14:00–15:00 | 0.32× | 0.32× | 0.32× |
| 15:00–16:00 | 0.32× | 0.41× | 0.40× |
Range of each window as a multiple of the average daily range. /ES regular hours, Feb 2021 – Oct 2026. 45 monthly opex days, 23 quarterly, 225 other Fridays.
The last hour of monthly opex is about 18% quieter than on other Fridays (p = 0.048). That is what a pin would look like: price held near a big strike while the options run out.
Two caveats:
- It only appears from June 2022, when daily-expiring SPX options arrived. From 2021 to May 2022 (16 opex days) there was no difference.
- Triple witching shows no pin. Its last hour is as active as any other Friday's.
Myth 4: triple witching is bearish — leans true, borderline
| Days | Avg. return | Median | Up days | |
|---|---|---|---|---|
| Quarterly opex | 65 | −0.24% | −0.18% | 38% |
| Monthly opex | 128 | +0.03% | +0.10% | 53% |
| Other Fridays | 627 | +0.05% | — | 55% |
S&P 500 futures, 2010–2026.
Triple witching is the only claim with some support. It is down more often than up, and the gap against other Fridays reaches p = 0.046. That only just clears the bar, and it is one of many tests in this study, so a result like this can show up by luck. The effect is also weaker since 2022 (−0.14%, 47% up, 17 days).
(inference) Treat it as a mild lean, not a rule to trade on.
The dividend trap
On SPY's raw price history, triple witching looks far worse: −0.56% on average and only 24% up days. Most of that is not selling.
SPY pays its quarterly dividend with an ex-dividend date on the third Friday of March, June, September and December, which is the same day as triple witching. On the ex-dividend date the price drops by the dividend, roughly 0.3%. Unless the chart is dividend-adjusted, that drop looks like a down day. In June 2024, SPY closed at 547.00 on Thursday and opened at 544.40 on triple-witching Friday, mostly because of a $1.76 dividend.
S&P 500 futures pay no dividend, so they show the real move: about −0.24%, not −0.56%. Any triple-witching statistic built on raw SPY prices is partly measuring the dividend.
Myth 5 and 6: opex week up, week after down — false
| Window | Avg. return | Median | Up % |
|---|---|---|---|
| Any 5-day window (baseline) | +0.26% | — | 61% |
| Opex week (5 days ending on opex) | +0.12% | +0.44% | 57% |
| Week after monthly opex | +0.45% | +0.74% | 68% |
| Week after quarterly opex | +0.10% | −0.08% | 44% |
S&P 500 futures, 2010–2026.
Opex week is not bullish; it returns less than an average week. The week after monthly opex does better than an average week, not worse (p = 0.35, so not a reliable edge either). The week after triple witching is softer (44% up), but that difference is not significant (p = 0.56).
Myth 7: the Monday "vol unclamp" — false
The idea is that once options expire, dealers stop hedging and the market is free to move. On the Monday after opex:
- Range is the same as a normal Monday (p = 0.31 intraday, 0.90 on daily bars).
- VIX rises less: +0.12 points on average, vs +0.39 on other Mondays.
Nothing unclamps.
What this means
- Don't plan for a bigger day because it's opex. If anything, monthly opex is a little quieter, and its last hour quieter still.
- Size for the market you are in, not the calendar. The range was measured against each day's own recent normal, and opex did not add to it.
- Triple witching has a mild downward lean. It is the only claim that holds up even partly, and it is borderline.
- Check the data source behind any opex statistic. Raw SPY prices add a fake 0.3% drop on every triple witching.
Method notes
- Opex date = third Friday of the month, or the prior trading day if the market was closed.
- Range = high minus low, divided by the average range of the prior 20 trading days (no look-ahead).
- Futures daily bars use the full trading session (6 PM to 5 PM ET) on the front-month contract. Intraday windows use regular hours only.
- Shuffle tests use 5,000–10,000 random draws from the comparison group.