The holiday effect in SPY is real in the averages but small in practice. From 1993 through 2026, the last session before a US market holiday averaged 0.11% and rose 57% of the time (n=297), about twice the 0.05% of an average day. The session after a holiday averaged 0.05%, the same as any other day. And holding SPY only on pre-holiday sessions would have compounded to about 35% over 33 years, against roughly 2,957% for holding through every session.
The old post at this address had a tongue-in-cheek title about making millions from SPY holiday effects. The original (January 2010) tested the S&P 500’s next-session return from the close before and after each exchange holiday over decades of index history. This version keeps the question and drops the promise: it measures what SPY has done around each named holiday since the fund started trading, with data through November 6, 2026.
How the holidays were picked
Every weekday with no SPY trading from January 29, 1993 to November 6, 2026 was found in the price file and matched to a named holiday by its calendar rule:
- New Year’s Day, Martin Luther King Jr. Day (closed since 1998, after the NYSE announced it in 1997), Presidents Day (Washington’s Birthday), Good Friday, Memorial Day, Juneteenth (closed since 2022), Independence Day, Labor Day, Thanksgiving and Christmas.
- When a holiday falls on a weekend, the observed weekday counts. One exception from the NYSE calendar: when January 1 is a Saturday, there is no New Year’s closure at all. That happened in 2000, 2005, 2011 and 2022, so New Year’s has fewer observations.
One-off closures were excluded. The ones in the data are the national days of mourning for Presidents Nixon (April 27, 1994), Reagan (June 11, 2004), Ford (January 2, 2007), George H.W. Bush (December 5, 2018) and Carter (January 9, 2025), the September 11-14, 2001 closure, and Hurricane Sandy (October 29-30, 2012). New Year’s 2007 was dropped too, because the Ford closure ran straight into it.
“Before” is the close-to-close return of the last session before the closure. “After” is the first session after it, measured from the pre-holiday close. All returns use SPY’s adjusted close, which includes dividends.
SPY before and after holidays
The pre-holiday session has been better than an average day; the post-holiday session has not.
| Day | n | Average % | Median % | % up |
|---|---|---|---|---|
| Session before a holiday | 297 | 0.11 | 0.13 | 57.24 |
| Session after a holiday | 297 | 0.05 | 0.04 | 51.85 |
| Every session | 8501 | 0.05 | 0.07 | 54.03 |
| Other sessions (not next to a holiday) | 7907 | 0.04 | 0.07 | 53.99 |
The pre-holiday average sits about 0.06 percentage points above a normal day. Its standard error is about 0.05 points, so the gap is roughly one standard error: consistent with a mild effect, but not strong evidence on its own. The 57% up rate against 54% tells the same modest story.
The session before each holiday
The pre-holiday lift isn’t spread evenly. It is concentrated before Good Friday, Independence Day, Thanksgiving, Christmas and MLK Day; the sessions before New Year’s and Presidents Day have been weak.
| Holiday | Years | n | Average % | Median % | % up |
|---|---|---|---|---|---|
| New Year’s Day | 1995-2026 | 27 | -0.30 | -0.36 | 33.33 |
| Martin Luther King Jr. Day | 1998-2026 | 29 | 0.20 | 0.23 | 62.07 |
| Presidents Day (Washington’s Birthday) | 1993-2026 | 34 | -0.15 | -0.10 | 38.24 |
| Good Friday | 1993-2026 | 34 | 0.41 | 0.28 | 67.65 |
| Memorial Day | 1993-2026 | 34 | 0.05 | 0.14 | 55.88 |
| Juneteenth | 2022-2026 | 5 | 0.23 | 0.22 | 60.00 |
| Independence Day | 1993-2026 | 34 | 0.24 | 0.25 | 70.59 |
| Labor Day | 1993-2026 | 34 | 0.01 | 0.04 | 52.94 |
| Thanksgiving | 1993-2025 | 33 | 0.21 | 0.26 | 66.67 |
| Christmas | 1993-2025 | 33 | 0.20 | 0.10 | 63.64 |
Each row has only 27 to 34 observations (Juneteenth has five), so single rows are noisy. The session before New Year’s is usually December 31, the last day of the year, which has its own year-end flows. The session before Christmas is often Christmas Eve, a 1:00 p.m. early close on the NYSE’s 2026 calendar.
The session after each holiday
After the break, results flip for several holidays. The sessions after Christmas and New Year’s have been strong; the Tuesdays after MLK Day and Presidents Day have averaged losses.
| Holiday | Years | n | Average % | Median % | % up |
|---|---|---|---|---|---|
| New Year’s Day | 1995-2026 | 27 | 0.44 | 0.25 | 59.26 |
| Martin Luther King Jr. Day | 1998-2026 | 29 | -0.27 | -0.18 | 44.83 |
| Presidents Day (Washington’s Birthday) | 1993-2026 | 34 | -0.28 | 0.01 | 50.00 |
| Good Friday | 1993-2026 | 34 | -0.12 | 0.04 | 55.88 |
| Memorial Day | 1993-2026 | 34 | 0.14 | -0.06 | 47.06 |
| Juneteenth | 2022-2026 | 5 | 0.24 | -0.27 | 20.00 |
| Independence Day | 1993-2026 | 34 | 0.02 | -0.09 | 41.18 |
| Labor Day | 1993-2026 | 34 | 0.03 | -0.26 | 41.18 |
| Thanksgiving | 1993-2025 | 33 | 0.14 | 0.08 | 60.61 |
| Christmas | 1993-2025 | 33 | 0.39 | 0.25 | 72.73 |
The post-Thanksgiving session is the Friday half day (1:00 p.m. close on the NYSE calendar). The post-Christmas session falls in the last week of December, which our December study covers on its own. In 12 of the 29 years since 1998, the Tuesday after MLK Day fell inside January option expiration week, another soft spot on the calendar.
Has the effect faded since 2010?
No. The pre-holiday session has done a little better since the original post, mostly in how often it rises. The post-holiday session has gone flat.
| Day | n | Average % | Median % | % up |
|---|---|---|---|---|
| 1993-2009, session before | 143 | 0.09 | 0.09 | 52.45 |
| 1993-2009, session after | 143 | 0.08 | 0.07 | 53.85 |
| 1993-2009, every session | 4263 | 0.04 | 0.07 | 52.85 |
| 2010-2026, session before | 154 | 0.12 | 0.17 | 61.69 |
| 2010-2026, session after | 154 | 0.02 | -0.00 | 50.00 |
| 2010-2026, every session | 4238 | 0.06 | 0.07 | 55.21 |
Since 2010 the pre-holiday session rose 62% of the time against 55% for all sessions. Before 2010 the two were almost equal (52% vs 53%).
Why “making millions” on it doesn’t add up
A small edge on a few days a year can’t compete with simply being invested. The script compounds SPY’s return over only the 297 pre-holiday sessions:
- Pre-holiday sessions only: about +35% over 1993-2026.
- Pre-holiday plus post-holiday sessions (594 sessions): about +54%.
- Every session (8,501): about +2,957%, with dividends reinvested.
Pre-holiday sessions are 3.5% of all sessions and account for 7.8% of the sum of SPY’s daily returns. That is better than their share, which is the holiday effect. It is also a thin slice of the total, before trading costs, taxes or the interest cash would earn on the other days. None of this says what the next holiday will do.
More calendar studies are on the Market Stats page, including September triple witching. Every number here was calculated from the daily data listed under Sources; past results only.
