What Is the September Effect? Why Stocks Tend to Drop in September

Search Fintrev

Search our guides, reviews and glossary.

The September effect is the observed tendency for stock market returns to be weaker in September than in other months. It is a calendar anomaly. It describes a historical average, not a rule that stocks fall every September, and it has no proven cause that would let anyone predict a given year’s result. Over long U.S. records, September has averaged a loss for the S&P 500. Yet stocks have still risen in September in a large share of years, and the pattern has weakened in recent decades.

How the September effect is measured

The effect is a statement about averages across many years. To test it, you calculate each year’s September return for an index:

September return = (index close on the last trading day of September ÷ index close on the last trading day of August) − 1

Repeat this for every year in the sample, then summarize the results three ways:

1. Average (arithmetic mean): the sum of all September returns divided by the number of years. A few severe losses can pull this figure down sharply.
2. Median: the middle result when all Septembers are ranked. It shows what a typical year looked like and is less affected by extreme months.
3. Frequency of declines: the share of years in which September ended lower than August.

These three measures can disagree. A long sample can show a negative average alongside a positive median, or a negative arithmetic mean alongside a positive geometric mean (the compounded growth rate). The calculation is usually done on price returns; including dividends shifts every month’s figure slightly upward but does not change the ranking logic.

Worked example (hypothetical)

Assume an index posts these September returns over five years: -6%, +2%, +1%, -3% and +1.5%.

• Average: (-6 + 2 + 1 - 3 + 1.5) ÷ 5 = -0.9%
• Median: ranked, the results are -6, -3, +1, +1.5, +2, so the middle value is +1%
• Down years: 2 of 5, or 40%

In this example, September “averages a loss,” but three of the five years were positive and the typical September was a gain. One bad year created most of the negative average, the same shape found in recent real data.

What the historical record shows

From 1928 through 2025, the S&P 500 averaged a decline in September. One analysis covering 1928 through 2021 put the average loss at about 1%. September is not the worst month every year, though, and in some years it has been among the best.

The pattern has faded. Large September declines have occurred less frequently since around 1990 than before. One explanation is pre-positioning: investors who expect a weak September sell earlier, in August. That moves the weakness to a different month rather than leaving it in September.

September has hosted some memorable market disruptions:

• the original Black Friday gold panic of 1869
• large single-day drops in the Dow Jones Industrial Average after the September 11 attacks in 2001
• sharp falls in 2008 as the subprime crisis escalated

Events like these are part of the long-run average, but they were driven by specific shocks, not by the calendar.

Academic studies have also reported September weakness outside the U.S., including Germany’s DAX, the Malaysian market and the Baltic stock markets. The strength of the effect varies by market and period.

Why September might be weaker

No cause has been established, but several explanations are commonly offered:

• End of summer repositioning. Investors returning from lighter summer trading may act on sales they postponed, adding selling pressure.
• Tax-loss harvesting. Investors and funds may start selling losing positions in September to realize losses before year-end.
• Mutual fund fiscal years. Some mutual funds end their fiscal year around this time, and managers may sell losing holdings before reporting.
• Quarter-end institutional selling. As the third quarter closes, institutions may lock in profits or rebalance going into the final quarter.
• Household cash needs. Some individual investors may sell stocks to cover back-to-school costs.
• Self-fulfilling sentiment. If enough traders expect a weak September, selling in advance can make it so, at least briefly.

September effect vs. October effect

The October effect is the belief that October is a dangerous month for stocks. It rests mostly on dramatic events: the Panic of 1907, the 1929 crash and Black Monday in October 1987. Despite those episodes, October’s 100-year record is positive overall. The two ideas therefore differ in substance:

• The September effect is about a weak average return.
• The October effect is about the memory of crashes, not about weak average performance.

Both depend heavily on the period studied. Economists doubt both as forecasting tools.

The September effect belongs to a wider family of calendar anomalies, such as the January effect and the “sell in May and go away” adage: seasonal patterns in historical returns without a demonstrated causal mechanism.

Limitations: why it is a weak trading signal

It conflicts with market efficiency, and that tends to erode it. A predictable monthly decline is an anomaly under the efficient market hypothesis. Once a pattern is widely known, traders act on it early, and the pattern weakens or shifts. The move toward August selling fits this explanation.

Results depend on the sample period. Start the calculation in 1928, 1945, 1950 or 2000 and you get different answers about how weak September is, and even whether it is the worst month. A result that changes with the starting year is a fragile basis for decisions.

The odds are close to even. A roughly 45% chance of an up September remains, and an average loss of well under 1% is small compared with ordinary monthly swings.

Leaving the market has costs. Consider a hypothetical investor who sells everything at the end of August and buys back at the end of September every year. In years when September rises, that investor misses the gain. In taxable accounts, the investor may also realize capital gains and pay trading costs. Those costs apply every year, while any benefit appears only in some years.

Correlation is not causation. The September average includes crises that would have hurt markets in any month. Removing a few such years can change the result substantially.

FAQs

The Bottom Line

Investors tempted to time the market around September should treat the effect as a fragile historical average. Its strength depends on the sample period, the odds of a down September are close to even, and the pattern has faded as traders sell earlier in August. Selling before September and buying back after risks missing gains and adds trading costs and possible taxes every year.