Dow Jones Seasonality
What three decades of daily closes actually say about calendar patterns — average month, best month, worst month, weekday effects — and which of those patterns deserve your attention.
Seasonality is the most over-sold idea in market folklore, so this page plays it straight: the numbers below are computed directly from every completed month in our dataset, the current month is excluded until it finishes, and the caveats get equal billing with the patterns. Use it to calibrate headlines ("stocks enter their historically weak season") against what the data really shows.
Average return by calendar month
This month right now
Historically, August has been one of the weaker months for the Dow: across 34 completed Augusts in our dataset the index averaged −0.51% and finished higher 56% of the time. The best August was 2020 (+7.57%); the worst was 1998 (−15.13%). See the full month-by-month breakdown.
The full month-by-month table
The average alone hides a lot, so we include the median (which resists distortion by one extreme year), the share of months that finished positive, and the single best and worst instance of each month. A month with a positive average but a sub-60% win rate is a coin flip with good outliers, not a reliable pattern.
| Month | Average | Median | Positive | Best | Worst |
|---|---|---|---|---|---|
| January | +0.28% | +0.42% | 59% | +7.17% (2019) | −8.84% (2009) |
| February | −0.01% | +1.18% | 63% | +8.08% (1998) | −11.72% (2009) |
| March | +0.54% | +1.05% | 63% | +7.84% (2000) | −13.74% (2020) |
| April | +2.17% | +1.79% | 74% | +11.08% (2020) | −5.00% (2024) |
| May | +0.49% | +1.05% | 66% | +4.59% (1997) | −7.92% (2010) |
| June | −0.11% | −0.08% | 49% | +7.19% (2019) | −10.19% (2008) |
| July | +1.61% | +1.00% | 77% | +8.58% (2009) | −5.48% (2002) |
| August | −0.51% | +0.48% | 56% | +7.57% (2020) | −15.13% (1998) |
| September | −0.70% | +0.06% | 50% | +7.72% (2010) | −12.37% (2002) |
| October | +1.75% | +2.27% | 65% | +13.95% (2022) | −14.06% (2008) |
| November | +2.69% | +3.00% | 76% | +11.84% (2020) | −5.32% (2008) |
| December | +0.97% | +1.26% | 68% | +6.86% (2003) | −8.66% (2018) |
Which patterns are real?
Two calendar effects show up persistently enough in the Dow's history to take semi-seriously. The first is the September effect: September is reliably the weakest month in the table above, a result that has held across most multi-decade samples of U.S. stocks and for which no fully convincing explanation exists — candidates include mutual-fund fiscal year-ends, post-summer repositioning, and simple self-fulfilling reputation. The second is the November–April vs May–October split behind the "sell in May" adage: the winter half of the year has historically delivered the bulk of the index's gains, though the summer half is still positive on average — which is precisely why acting on the adage has been a losing proposition. Sitting out a mildly positive half-year to avoid mild underperformance costs real money and triggers real taxes.
Everything else deserves heavy discounting. With roughly three decades of data, each calendar month has only about 34 samples — small enough that one 2008-sized October or one 2020-sized April visibly bends its average (that's why the table shows each month's best and worst instance: check how much lifting a single year does). Weekday effects are even weaker. The famous "Monday effect" — Mondays being the market's worst day — was a robust finding in academic studies of mid-20th-century data, but it has largely evaporated in modern, electronically-traded markets, as the table below shows.
| Weekday | Avg daily change | Positive days | Sessions |
|---|---|---|---|
| Monday | +0.071% | 55.3% | 1,643 |
| Tuesday | +0.064% | 51.9% | 1,792 |
| Wednesday | +0.034% | 52.6% | 1,790 |
| Thursday | +0.008% | 52.4% | 1,756 |
| Friday | +0.017% | 55.0% | 1,745 |
Average daily change and share of positive sessions by weekday, across all trading days in our dataset. Differences of a few hundredths of a percent are well within noise.
The honest summary: seasonality is context, not strategy. It is useful for setting expectations — knowing that a weak September is historically ordinary can stop a panicked reaction to one — and nearly useless for timing. No pattern in these tables comes close to overcoming trading costs, taxes, and the risk of simply being out of the market on its best days, which cluster unpredictably. We look at what those best days have in common on the records page.