September triple witching is the third Friday of September, when stock options, stock index options and stock index futures expire on the same day. The witching week and the day itself have looked ordinary for the S&P 500. The week after has not: from 1990 through 2026 the index fell in the week after September triple witching in 28 of 37 years, with an average of -0.93% against 0.18% for a typical week.
This page rebuilds a September 2010 post at the same address. The original, written the day before that year’s expiration, tracked SPY over the five sessions after every September witching day since 1990. This version keeps the 1990 start, uses the S&P 500 index, adds the witching week and the day, compares all four quarterly expirations, and checks what has happened since 2010.
What triple witching means
Triple witching is the quarterly expiration. Every month, standard stock and index options expire on the third Friday. In March, June, September and December, stock index futures expire the same day. That makes three contract types at once:
- Stock options on individual shares and ETFs.
- Stock index options, such as standard monthly S&P 500 (SPX) options. Per Cboe’s SPX specifications, standard SPX options stop trading on the business day before the settlement value is calculated, usually a Thursday.
- Stock index futures. CME’s E-mini S&P 500 contract specs list the quarterly March, June, September and December cycle, with trading in the expiring contract ending at 9:30 a.m. ET on the third Friday.
Quadruple witching was the name for the same day once single-stock futures joined in; Reuters reported the first one in December 2002. Those contracts traded on OneChicago, which ceased trading on September 18, 2020, per an SEC release. That last session was itself a September witching day. Without single-stock futures, “triple witching” is the accurate name again, though you’ll still see both.
When the third Friday is a market holiday, expiration moves to the session before. In this sample that happened twice: March 2008 (Good Friday, so Thursday March 20) and June 2026 (Juneteenth, so Thursday June 18).
September triple witching by the numbers
The week and the day are close to average. The week after is the outlier, with fewer than one in four up.
| Period | n | Average % | Median % | % up |
|---|---|---|---|---|
| Witching week (Mon-Fri) | 37 | 0.10 | 0.64 | 59.46 |
| Witching day (Friday) | 37 | -0.01 | -0.01 | 48.65 |
| Week after | 37 | -0.93 | -1.01 | 24.32 |
| Any week, same years (baseline) | 1923 | 0.18 | 0.28 | 56.53 |
| Any day, same years (baseline) | 9281 | 0.04 | 0.06 | 53.65 |
How each row is measured, on the S&P 500 price index: the witching week is the expiration day’s close against the previous week’s last close. The witching day is the close-to-close change on that session. The week after runs from the expiration close to the last close of the next calendar week.
A -0.93% average over 37 weeks has a standard error of about 0.39 percentage points, so the gap to a normal week is wider than noise would usually produce. It is still one statistic from one sample, and a few big weeks carry it (more on that below).
How September compares with the other quarterly expirations
September has the weakest week after of the four quarterly months. June is also negative. December’s week after has been strong, the only quarterly month where it rose more often than not.
| Period | n | Average % | Median % | % up |
|---|---|---|---|---|
| March | 37 | 0.04 | -0.53 | 32.43 |
| June | 37 | -0.41 | -0.40 | 29.73 |
| September | 37 | -0.93 | -1.01 | 24.32 |
| December | 36 | 0.71 | 0.62 | 66.67 |
| March, June, December combined | 110 | 0.11 | -0.24 | 42.73 |
| Non-quarterly monthly expirations | 295 | 0.16 | 0.32 | 58.31 |
The table shows the week after each expiration. March, June and September all have up-week shares around 24% to 32%, well under the 58% after the eight non-quarterly expirations. So part of September’s weakness looks like a quarterly-expiration effect, and part looks like September itself. December is the exception, and the market in December has its own page.
The witching day itself shows no quarterly pattern worth noting. September’s day averaged -0.01% and rose 49% of the time; across March, June and December combined the day averaged -0.03%, close to the -0.05% for non-quarterly expiration days. The witching week was up 59% to 69% of the time in every quarterly month, a little above the 55% for non-quarterly expiration weeks.
Has the week after held up since 2010?
On average, yes; on consistency, less so. Since the original post the week after has averaged about the same, but it has been down in 11 of 17 years rather than 17 of 20.
| Period | n | Average % | Median % | % up |
|---|---|---|---|---|
| 1990-2009 (before the original post) | 20 | -0.92 | -1.40 | 15.00 |
| 2010-2026 (since the original post) | 17 | -0.95 | -0.63 | 35.29 |
| 1990-2026 | 37 | -0.93 | -1.01 | 24.32 |
The first session after the witching day shows the change more clearly. From 1990 to 2009 the S&P 500 rose on that Monday in only 5 of 20 years (average -0.45%); from 2010 to 2026 it rose in 8 of 17 (average -0.04%), close to a normal day.
The five worst weeks after were 2011 (-6.54%), 2022 (-4.65%), 1999 (-4.35%), 2003 (-3.81%) and 2008 (-3.33%). Leave them out and the other 32 still average -0.37% with 28% up, so the pattern is not only those five. The best week after was 2001 (+7.78%), the rebound week after the post-9/11 reopening.
September triple witching, year by year since 2010
Recent years have been mixed. From 2016 to 2026 the week after was down in 6 of 11 years, close to even.
| Year | Witching day | Week % | Day % | Week after % |
|---|---|---|---|---|
| 2010 | Sep 17 | 1.45 | 0.08 | 2.05 |
| 2011 | Sep 16 | 5.35 | 0.57 | -6.54 |
| 2012 | Sep 21 | -0.38 | -0.01 | -1.33 |
| 2013 | Sep 20 | 1.30 | -0.72 | -1.06 |
| 2014 | Sep 19 | 1.25 | -0.05 | -1.37 |
| 2015 | Sep 18 | -0.15 | -1.62 | -1.36 |
| 2016 | Sep 16 | 0.53 | -0.38 | 1.19 |
| 2017 | Sep 15 | 1.58 | 0.18 | 0.08 |
| 2018 | Sep 21 | 0.85 | -0.04 | -0.54 |
| 2019 | Sep 20 | -0.51 | -0.49 | -1.01 |
| 2020 | Sep 18 | -0.64 | -1.12 | -0.63 |
| 2021 | Sep 17 | -0.57 | -0.91 | 0.51 |
| 2022 | Sep 16 | -4.77 | -0.72 | -4.65 |
| 2023 | Sep 15 | -0.16 | -1.22 | -2.93 |
| 2024 | Sep 20 | 1.36 | -0.19 | 0.62 |
| 2025 | Sep 19 | 1.22 | 0.49 | -0.31 |
| 2026 | Sep 18 | -0.08 | 0.17 | 1.21 |
2011 shows how lumpy this is: a 5.35% witching week followed by a 6.54% drop. Most years are much smaller moves either way.
What the numbers do and don’t show
The week after September triple witching has been one of the weaker weeks on the calendar since 1990, for the S&P 500 and on a sample of 37. That is a description of the past, not a forecast. Late September also brings the end of the quarter and whatever news the month carries, so expiration may not be the cause. The pattern is also well known, and widely known patterns can fade. For the January version of this question see January option expiration week; for closures and short weeks see SPY holiday effects; the full list is on Market Stats. Every number here was calculated from the daily data listed under Sources, before costs and taxes.
