U.S. Stock Market Seasonality: All 12 Months Ranked, 2000–2026
Every month's mean, median, hit rate and standard deviation — and the reason the fourth column beats the first three at telling you what to do.
Every month, ranked and qualified
27 years of S&P 500 monthly total returns, 2000–2026. The mean is what people quote; the standard deviation beside it is why the mean is a weak instrument.
| Month | Mean | Median | Positive | Std dev | Best | Worst |
|---|---|---|---|---|---|---|
| January | −0.0% | +1.4% | 52% | 4.2 | 2019 +7.9% | 2009 −8.6% |
| February | −0.5% | −0.4% | 48% | 4.3 | 2015 +5.5% | 2009 −11.0% |
| March | +0.8% | +1.0% | 59% | 4.8 | 2000 +9.7% | 2020 −12.5% |
| April | +1.9% | +1.2% | 70% | 5.0 | 2020 +12.7% | 2022 −8.8% |
| May | +0.8% | +1.2% | 74% | 3.7 | 2025 +6.2% | 2010 −8.2% |
| June | −0.1% | +0.1% | 59% | 4.0 | 2019 +6.9% | 2008 −8.6% |
| July | +1.5% | +1.6% | 67% | 3.7 | 2022 +9.1% | 2002 −7.9% |
| August | +0.3% | +1.2% | 63% | 3.6 | 2020 +7.0% | 2001 −6.4% |
| September | −1.2% | +0.4% | 52% | 4.9 | 2010 +8.8% | 2002 −11.0% |
| October | +1.3% | +1.9% | 62% | 5.6 | 2011 +10.8% | 2008 −16.9% |
| November | +2.1% | +2.6% | 77% | 4.4 | 2020 +10.8% | 2000 −8.0% |
| December | +0.6% | +0.8% | 65% | 3.5 | 2010 +6.5% | 2018 −9.2% |
The ranking by mean runs November, April, July at the top and June, February, September at the bottom. Every one of those means is smaller than its own month-to-month standard deviation, and the closest call is a factor of 2.1. That is the single most important line in this article: the seasonal signal is real in the average and invisible in any individual year.
The worst month is not what it looks like
September has the lowest mean at −1.2% — the origin of every "worst month for stocks" headline. But its median is +0.4% and it finished positive in 52% of these years. A handful of severe Septembers (2002 at −11.0%, for one) drag the average below zero while the typical September is unremarkable.
The best month, November, is the cleaner effect: mean +2.1%, median +2.6%, positive in 77% of years. Mean and median agree, which is what a real tendency looks like as opposed to a tail artefact.
The interactive monthly heatmap, year by year →
"Sell in May" — the arithmetic
Compounding the six-month blocks rather than averaging months separately:
- November–April: +4.9% on average, positive in 19 of 26 runs. Worst: 2000–2001 at −12.6%.
- May–October: +2.4% on average, positive in 18 of 26 runs. Worst: 2008 at −30.1%.
So the seasonal gap exists — about 2.5 percentage points per half-year in this sample. And it does not support the instruction. The summer half is positive, on average and in 69% of runs; selling it means paying spreads and taxes to step out of a stretch that has usually made money, on a signal derived from 27 observations. "Sell in May" is a description of a mild tilt that has been marketed as a rule.
What 27 years can and cannot tell you
- It cannot separate seasonality from events. This window contains the dot-com unwind, 2008 and 2020. One month — October 2008, at −16.9% — moves October's mean by 0.7 points all by itself.
- The sample is 27 per month. Not 6,750 trading days — 27 independent observations. A 3.4-point spread in means on 27 draws with a 4-point standard deviation is not a finding you would act on anywhere else.
- A published calendar effect is the easiest kind to trade away. The reasonable prior for any pattern that has been in print for decades is that it is weaker now than in the sample that found it.
- What survives is boring. Stay invested; the average year is up +8.1% on price despite an average intra-year drawdown of −16.2%.
Intra-year drawdown vs full-year outcome → · The annual ledger → · The same heatmap for the Nasdaq-100 →
FAQ
What is the best month for the stock market?
November: mean +2.1%, median +2.6%, positive in 77% of the 2000–2026 sample. April and July are next.
What is the worst month for stocks?
September by mean (−1.2%), but its median is +0.4% and it was positive in 52% of years — the negative average comes from a few severe Septembers, not from a typical one.
Does 'sell in May and go away' work?
The gap is real and small. November–April averaged +4.9% against +2.4% for May–October across 26 runs. The summer half is still positive and positive in 69% of runs, so selling it trades a usually-profitable stretch for costs and taxes.
Computed from /api/sp500/monthly.json, the same file behind the monthly heatmap. The positive-share column is the JSON's published "probability" field, so this page and the chart cannot disagree. 2000–2026, published 2026-09-05; the current month is included through the data date, so its column moves until the month closes.