January option expiration week, the Monday-to-Friday week ending on January’s third Friday, has been one of the weaker weeks of the year for the S&P 500. From 1974 through 2026 the index rose in 47% of January expiration weeks (n=53, average 0.09%), against 58% for every other month’s expiration week (n=581, average 0.23%). Most of that gap comes from 1994-2010, when January’s week rose in only 5 of 17 years; since 2011 it has been closer to a coin flip.
This page rebuilds a January 2011 post that listed every January expiration week since 1974 and pointed out the S&P 500 had fallen in each of the six previous ones (2005-2010, which the data here confirms). The tables below run the same question through January 2026, with comparisons the old post didn’t make.
How the weeks are measured
Each week runs Monday to Friday, and its return is the expiration Friday’s close against the previous week’s last close. Expiration day is the third Friday of the month, or the session before it if the market was closed that Friday (a Good Friday, for example).
A holiday-shortened week is still measured from the prior Friday’s close. That matters in January because of Martin Luther King Jr. Day, the third Monday of the month. It falls inside expiration week when January 1 is a Saturday, Sunday or Monday. The NYSE announced in June 1997 that it would close for the holiday, and the price data shows the first closure on January 19, 1998. So from 1998 on, some January expiration weeks have four sessions instead of five. They are kept in the sample, and Table 2 splits them out.
The study starts in 1974, the first January after Cboe opened on April 26, 1973 as the first exchange for listed options. The main series is the S&P 500 price index (no dividends). A SPY check with dividends follows at the end.
January expiration week against every other week
The short answer: January’s expiration week has trailed both other expiration weeks and ordinary weeks, by 0.08 to 0.14 percentage points on average and by 9 to 11 points on the share of up weeks.
| Weeks | n | Average % | Median % | % up |
|---|---|---|---|---|
| January expiration week | 53 | 0.09 | -0.04 | 47.17 |
| Other months’ expiration weeks | 581 | 0.23 | 0.39 | 57.83 |
| Non-expiration weeks | 2123 | 0.17 | 0.30 | 56.43 |
| All weeks | 2757 | 0.18 | 0.31 | 56.55 |
The median January week is slightly negative, and fewer than half of them closed higher. Still, 53 weeks is a small sample. The standard error on that 0.09% average is about 0.31 percentage points, larger than the gap to the other groups. On averages alone, this result is too weak to call January different from any other week.
The weak stretch was 1994-2010
The weakness is concentrated in one period. From 1974 to 1993 January’s expiration week rose 65% of the time; from 1994 to 2010 it rose 29% of the time.
| Weeks | n | Average % | Median % | % up |
|---|---|---|---|---|
| 1974-2010 (the original post’s window) | 37 | 0.13 | -0.02 | 48.65 |
| 2011-2026 (since the original post) | 16 | 0.00 | -0.17 | 43.75 |
| 1974-1993 | 20 | 0.84 | 0.97 | 65.00 |
| 1994-2010 | 17 | -0.71 | -0.78 | 29.41 |
| 1994-2026 (same years as the SPY check) | 33 | -0.36 | -0.26 | 36.36 |
| 1998-2026, 4 sessions (week includes MLK Day) | 12 | -0.54 | -0.40 | 33.33 |
| 1998-2026, 5 sessions | 17 | -0.53 | -0.78 | 35.29 |
When the old post was written, the recent record looked one-sided: down in every year from 2005 to 2010. Since then the week has been down in 9 of 16 years with an average of 0.00%, so the pattern has faded since the original post. The MLK split shows almost nothing: four-session and five-session weeks since 1998 averaged about the same (-0.54% vs -0.53%), each on a dozen or so observations. The holiday doesn’t explain the weakness.
Where January ranks among the twelve expiration weeks
January’s week has the lowest share of up weeks and the lowest median of all twelve, though its average sits mid-table, above May, June, July, August and November. March, April, October and December have been the strongest.
| Weeks | n | Average % | Median % | % up |
|---|---|---|---|---|
| Jan | 53 | 0.09 | -0.04 | 47.17 |
| Feb | 53 | 0.25 | 0.12 | 52.83 |
| Mar | 53 | 0.47 | 0.86 | 64.15 |
| Apr | 53 | 0.80 | 0.85 | 64.15 |
| May | 53 | 0.01 | -0.03 | 49.06 |
| Jun | 53 | -0.06 | 0.15 | 56.60 |
| Jul | 53 | 0.04 | 0.24 | 54.72 |
| Aug | 53 | 0.03 | 0.37 | 52.83 |
| Sep | 53 | 0.21 | 0.64 | 58.49 |
| Oct | 53 | 0.39 | 0.85 | 67.92 |
| Nov | 52 | -0.01 | 0.09 | 53.85 |
| Dec | 52 | 0.36 | 0.52 | 61.54 |
With twelve groups of about 53 weeks each, some spread between months is expected by chance alone. A month at the bottom of this table is not proof of a cause. September’s expiration week is a quarterly “triple witching” week, and the week right after it has a much worse record than any of these; see September triple witching.
January expiration week, year by year since 2011
Year by year since the old post, the week has been mixed: seven up years, nine down, and one large loss in 2022 that pulls the average down.
| Year | Dates | Sessions | MLK Day in week | Week % |
|---|---|---|---|---|
| 2011 | Jan 18-Jan 21 | 4 | yes | -0.76 |
| 2012 | Jan 17-Jan 20 | 4 | yes | 2.04 |
| 2013 | Jan 14-Jan 18 | 5 | no | 0.95 |
| 2014 | Jan 13-Jan 17 | 5 | no | -0.20 |
| 2015 | Jan 12-Jan 16 | 5 | no | -1.24 |
| 2016 | Jan 11-Jan 15 | 5 | no | -2.17 |
| 2017 | Jan 17-Jan 20 | 4 | yes | -0.15 |
| 2018 | Jan 16-Jan 19 | 4 | yes | 0.86 |
| 2019 | Jan 14-Jan 18 | 5 | no | 2.87 |
| 2020 | Jan 13-Jan 17 | 5 | no | 1.97 |
| 2021 | Jan 11-Jan 15 | 5 | no | -1.48 |
| 2022 | Jan 18-Jan 21 | 4 | yes | -5.68 |
| 2023 | Jan 17-Jan 20 | 4 | yes | -0.66 |
| 2024 | Jan 16-Jan 19 | 4 | yes | 1.17 |
| 2025 | Jan 13-Jan 17 | 5 | no | 2.91 |
| 2026 | Jan 12-Jan 16 | 5 | no | -0.38 |
The last ten years split five up and five down. Take out 2022 and the post-2011 average turns positive. Results like that come from a short sample, not a stable effect.
The same test with SPY and dividends
SPY tells the same story for the years it covers. Including dividends, SPY’s January expiration week averaged -0.37% from 1994 to 2026 and rose 36% of the time, against about 57% to 59% for other weeks.
| Weeks | n | Average % | Median % | % up |
|---|---|---|---|---|
| SPY January expiration week | 33 | -0.37 | -0.35 | 36.36 |
| SPY other expiration weeks | 361 | 0.26 | 0.48 | 59.28 |
| SPY non-expiration weeks | 1320 | 0.23 | 0.33 | 56.67 |
This matches the S&P 500 index over the same years (-0.36%, 36% up), so dividends don’t change the picture. The SPY window starts in 1994, which leaves out the strong 1974-1993 years. That is why it looks worse than the full-history table above.
What the numbers do and don’t show
The record supports a modest statement: since 1994, the S&P 500 has done worse in January’s expiration week than in a typical week, and most of that came before 2011. It doesn’t show a reliable edge. The sample is small, the result depends on the start year, and the years since the original post look close to random. The holiday calendar has its own patterns, covered in SPY holiday effects, and the rest of the seasonal studies are on the Market Stats page, including the market in December. All numbers here are past results, before trading costs and taxes.
