The Most-Cited Seasonal in Finance
"Sell in May and go away, but come back on St. Leger Day" is the old rhyme — but the month that actually earns the bearish reputation is September. It's the most cited negative seasonal in markets, referenced by retail traders, strategists, and Bloomberg terminals alike every August 31.
We ran 16 years of /ES continuous front-month data to test whether September deserves the reputation — and the answer is nuanced in a way that matters for how you trade it.
The Numbers
16 years. 8 up. 8 down. 50% win rate. Average return: -0.78%.
| Year | Return | Max Up | Max Down |
|---|---|---|---|
| 2010 | +7.98% | +9.65% | -0.33% |
| 2011 | -7.84% | +0.98% | -10.47% |
| 2012 | +1.99% | +4.85% | -0.45% |
| 2013 | +1.90% | +4.91% | -0.33% |
| 2014 | -1.91% | +0.75% | -3.61% |
| 2015 | -3.21% | +2.38% | -8.14% |
| 2016 | -0.56% | +0.93% | -4.05% |
| 2017 | +1.92% | +1.92% | -0.97% |
| 2018 | +0.52% | +1.39% | -1.20% |
| 2019 | +2.88% | +4.21% | -1.53% |
| 2020 | -4.43% | +2.64% | -10.34% |
| 2021 | -5.06% | +0.47% | -5.73% |
| 2022 | -8.94% | +5.29% | -13.39% |
| 2023 | -4.15% | +1.18% | -7.51% |
| 2024 | +2.65% | +2.30% | -4.89% |
| 2025 | +3.79% | +4.33% | -0.89% |
The Asymmetry Is the Story
A 50% win rate sounds like a coin flip — and directionally, it is. But the returns are not symmetric.
When September is down, it is down hard: -7.84%, -4.43%, -5.06%, -8.94%, -4.15%. The five worst Septembers average -6.07%.
When September is up, the gains are muted: excluding 2010 (a V-shaped recovery from a bad summer), the eight positive Septembers average +1.97%.
The expected value is negative not because September usually falls, but because when it falls, it falls further than it rises. This is asymmetric risk — identical probabilities, unequal outcomes. A statistical definition of a month to respect.
Four Consecutive Negative Septembers (2020–2023)
The four-year run from 2020 to 2023 is the clearest evidence of regime-driven seasonality. Each had a specific macro driver:
2020 (-4.43%): The post-COVID tech bubble began deflating in early September. Nasdaq fell 10% in three weeks as FAANG crowding unwound and options gamma amplified the selloff.
2021 (-5.06%): Fed taper tantrum. Chair Powell signaled tapering was coming at the August Jackson Hole meeting; September was the market digesting what post-QE looked like for the first time since 2018.
2022 (-8.94%): The worst September in the dataset. The Fed was hiking 75bps per meeting. August's Jackson Hole speech (see our August Second Half article) set the tone, and September confirmed it was not over.
2023 (-4.15%): Rates higher for longer. The September FOMC held at 5.25–5.50% and projected rates staying there through 2024. The market had been pricing 2024 cuts; the dots showed otherwise.
The streak broke in 2024 (+2.65%) as the Fed began cutting and the market believed in the soft landing. 2025 (+3.79%) confirmed the new macro regime.
Why September Is Structurally Different
Most negative seasonals fail in macro-stress environments and succeed in neutral ones. September is different: it has a structural concentration of risk.
First, summer positioning unwinds. Institutional portfolios that were light and defensive through July and August rebalance in September. This creates natural selling pressure, especially in extended or crowded positions.
Second, fiscal and corporate calendar pressures. September 30 is the U.S. government fiscal year end. Many pension funds, endowments, and corporate treasuries have September rebalancing obligations that generate predictable sell flows.
Third, the news cycle restarts. August is quiet. September brings Fed meetings, earnings pre-announcements, and often the first major macro data releases after the summer. If any of those miss, there are no low-volume buffer weeks to absorb the shock.
2010: The Outlier That Saves the Average
The +7.98% September 2010 deserves a footnote. It was a violent recovery from a brutal summer — May, June, and July 2010 were all negative as the European debt crisis roiled markets. September 2010 was a short-squeeze recovery into quarter-end. Remove it, and September's average return drops to -1.69%.
Trading September
The data supports caution rather than an outright short. Coin-flip win rates don't give directional edge. What the data does support:
Size down. The fat left tail (avg losing September: -5.02%) justifies smaller position sizes even if you're long.
Watch the macro setup. Four of the five worst Septembers were Fed-tightening years. If the policy environment is restrictive, the seasonal skews worse. If the Fed is cutting (2024, 2025), September loses its teeth.
Don't chase the first rally. The max drawdown data shows September regularly reaches -5% to -10% before recovering. Entries after an early-September bounce are structurally better than entries into the open.
Summary
| Metric | Value |
|---|---|
| Sample | 2010–2025 (16 years) |
| Win rate | 50.0% (8 of 16) |
| Avg return | -0.78% |
| Avg max drawdown | -3.71% (deepest of any month in dataset) |
| Worst year | 2022: -8.94% |
| Best year | 2010: +7.98% (V-recovery outlier) |
| Ex-2010 avg return | -1.69% |
September's reputation is earned — but the mechanism is asymmetry, not direction. The month doesn't usually fall. When it does, it falls hard. Size accordingly.