Seasonality Investing · Stock Seasonality, 8 lessons · Lesson 1
Why is September so hated?
September fell in 43 of 77 years, but rose in 34. We use numbers to check what seasonality is, and what it is not.
QToday's question
Should you sell stocks when September arrives?
You've probably heard that "September is the cruelest month for stocks." This lesson starts with one number, looks at the story and data behind it, and ends with a reversal and one thing you can try today.
MAPLearning map for this lesson
We build it up in three steps. At the end of each step, a 'Quick check' lets you test your understanding before moving on.
If you're new, start here. Being able to calculate an average is enough. If you're curious about what moves stock prices, you may want to read Learn (Economics) Lesson 1 first: Understand the economy and stock prices come into view. If you don't know a term, tap the dotted underline for an explanation.
- 1 · One number43 of 77
You can remember the S&P 500's September record as a single number, and understand why that number needs a period and a sample size attached.
- 2 · Insight · 3 · DataA tilt in probabilities
You understand that is statistics, not prophecy, and can read the average, , and by comparing them with the .
- 4 · Reversal · 5 · PracticeLower your expectations
You can read "sell in September" as "lower your expectations for September," and build a 10-year monthly table for a stock you follow yourself.
1 · One number43 of 77
Since 1950 there have been 77 Septembers for the S&P 500, and 43 of them ended negative. The average September return is −0.62%, and September is practically the only month of the twelve with a clearly negative average (February is close to zero at −0.04%). That's why September has long been called "the cruelest month for stocks."
Whenever this course uses a number, it always states the period and sample size too. We write "since 1950, September fell in 43 of 77 years" instead of "September is weak" because the same number carries a completely different weight depending on how many observations it was collected from.
Quick check"September is a bear market" and "Since 1950, September has fallen in 43 of 77 years." How do these two sentences differ?
2 · InsightSeasonality is a tilt in probabilities
is the phenomenon where stock returns have historically skewed in one direction in the same stretch of the calendar, such as a particular month, a particular weekday, or the turn of the year. The key is this one sentence: Seasonality is not a prophecy that "it rises in that month" but a statistic that "the historical average and have been tilted in one direction." Understanding this difference reduces the chance of making a big mistake because of seasonality.
Easy analogy · An umbrella in the rainy season
There are many rainy days in July. That doesn't mean it rains all of July. Still, it raises the odds that you'll bring an umbrella. Seasonality is the same. It's not "September must fall" but information like "September had a slightly higher chance of rain." Just as bringing an umbrella doesn't mean cancelling your outing, seasonality doesn't mean overturning your whole plan.
So why does the calendar leave a mark on stock prices? Many explanations say it's because the schedule of money flowing in and out is tied to the calendar. Salaries and pension contributions, tax filings and year-end closes, earnings releases and dividend record dates come around at similar times every year. As for why September was weak, there are only hypotheses, such as repositioning after vacations and fund fiscal-year-end loss harvesting, and no confirmed cause.
The 'schedule of money' on a one-year calendar
This is a simplified summary of schedules commonly cited in the US and Korean markets. The size of the actual effect differs each year, and the evidence for each cause is covered in Lesson 4 (money flows) and Lesson 6 (taxes).
Three sentences this course repeats
1 Seasonality is not a prediction but a tilt in probabilities. 2 Only patterns with a cause survive. 3 The calendar is a filter, not a signal.
Quick checkHow does "The S&P 500 rises in November" differ from "November had a 70% win rate since 1950"?
3 · DataA 12-month report card read with four numbers
Seasonality statistics are easy to be fooled by if you look at the average alone. Read a month's record with four numbers. The average is the sum divided by the count, the is the middle value when sorted by size, the win rate is the share of years that ended positive, and the is how many times that month has occurred. If the average is high but the median is low, it means one or two large numbers pulled the average up.
And always compare against the . Pooling every month from February 1950 to September 2026 (920 months) gives an average of +0.75% and a win rate of 60.4%. That means 6 out of 10 randomly chosen months were positive, so a month with a 60% win rate is not a 'strong month' but an 'ordinary month.'
S&P 500 monthly report card
Calculated directly from month-end values of Yahoo Finance ^GSPC daily closes (price index, excluding dividends and costs). Feb 1950–Sep 2026 (76–77 observations per month), and 20–21 per month since 2006. Retrieved 2026-10-09. The dotted line is the baseline of all months.
Since 1950, the months with the highest averages were November (+1.86%, 76 periods), April (+1.54%, 77 periods) and December (+1.41%, 76 periods). For September, both the average (−0.62%) and the median (−0.35%) are negative, and its win rate (44%) was also the lowest of the twelve months. With three numbers pointing the same way, 'September weakness' is not an illusion created by one or two crashes.
Only September remained, while December and January weakened
S&P 500 average monthly return (%). Calculated directly from Yahoo Finance ^GSPC month-end closes, price index basis, through September 2026.
Redraw the same table for the last 20 years and a different face appears. September was still the weakest since 2006 at an average of −0.50%, but its win rate was 52% (11 of 21). December's average fell sharply from +1.41% to +0.59%, and January's from +1.08% to +0.29%. By contrast, April (+2.29%), July (+2.32%) and November (+2.09%) had higher averages over the last 20 years.
| S&P 500 | Average | Median | Win rate | Sample |
|---|---|---|---|---|
| September · since 1950 | −0.62% | −0.35% | 44% | 77 |
| September · since 2006 | −0.50% | +0.43% | 52% | 21 |
| November · since 1950 | +1.86% | +2.03% | 70% | 76 |
| All months (baseline) | +0.75% | - | 60.4% | 920 months |
Since 2006, September has a sample of only 21, so the numbers can change a lot with just one or two years' results.
Quick checkA certain month has a 62% win rate. Can we call it a 'strong month'?
4 · ReversalSeptember rose in 34 years too
The record of 77 Septembers: did it fall every year?
S&P 500 September monthly returns, 1950–2026 (77 periods). Calculated directly from Yahoo Finance ^GSPC month-end closes.
Scanning the cells, red (up) and blue (down) cells are mixed together. Years that ended negative were only slightly more numerous, and years that ended positive are nearly half. There are recent years with a positive September return too, such as 2024 (+2.02%) and 2025 (+3.53%). Selling stocks to avoid September and buying them back costs trading fees and taxes, and you would miss the 34 Septembers that were positive.
Common misconceptions
"Should you sell all your stocks in September?" No. Since 1950, 44% of Septembers were positive. It's merely information that expected returns were a bit lower.
"Will buying in November make a profit?" No. A 70% win rate (76 periods) also means that three out of ten times it was negative.
"Just find the months with the highest averages?" If you dig through twelve months and several indexes, you are bound to find plausible-looking numbers by chance. This is called , and you filter it out with a cause and with other periods and other markets (Lesson 7).
Quick checkWhat would "lower your expectations in September" look like as an actual action?
5 · One thing to practiceA 10-year monthly table for a stock you follow
Seasonality is much easier to see when you fill it in yourself than when you memorize someone else's table. Pick one stock or ETF you're interested in and enter its last 10 years of monthly returns in the table below. Just copy the 'change vs previous month' from the monthly chart in your brokerage app. As you type, the monthly average, median, win rate and sample size are calculated at the bottom. Today's assignment is just one: look at how much the average differs from the median.
Practice table · Monthly returns over the last 10 years (%)
Your inputs are saved only in the browser on this device. The example values are monthly returns (%) calculated from Yahoo Finance ^GSPC month-end closes.
Fill in the example and you'll see that the S&P 500 over the last 10 years behaved quite differently from the 76-year average. A sample of 10 is easily shaken by one or two big moves, and a month where the average and median are far apart is the sign of that. These figures are educational examples and do not recommend any specific trade.
The figures in this course are on a price-index basis, so dividends, trading costs and taxes are left out. The fact that returns tilted a certain way in the past is no guarantee that they will tilt the same way in the future.
Quick checkIn my table, a certain month has an average of +3% but a median of +0.5%. What should I check first?
✓What we learned today
- Since 1950, the S&P 500 fell in September in 43 of 77 years, with an average of −0.62%. September is practically the only month of the twelve with a clearly negative average.
- Seasonality is a tilt in probabilities. It is not a prophecy that 'it rises in that month' but a statistic that 'the historical average and win rate have been tilted in one direction.'
- Read a month's record with four numbers, average, median, win rate and sample size, and compare them with the baseline of all months (average +0.75%, win rate 60.4%, 920 months).
- Over the last 20 years (since 2006), September was still the weakest with an average of −0.50%, but its win rate was 52%. December and January weakened, and April, July and November had higher averages.
- The reversal: September rose in 34 years too. What the data says is not 'sell in September' but 'lower your expectations for September.'
Today we used September alone to confirm what seasonality means. In Lesson 2, we look at the 6-month cycle that split $10,000 into $1.17 million versus $34,000, and the reversal in which that gap disappeared over the last 10 years.
Quiz5-question quiz
Pick an answer and it's graded right away with an explanation.
Q1. Which best describes seasonality?
Seasonality is not 'repeats every year' but 'a skew in the average.' Even September rose in 34 of 77 years.
Q2. The S&P 500's September rose in 34 of 77 years since 1950. About what is the probability of rising?
34 ÷ 77 ≈ 44%. It's only slightly worse than a coin flip (50%), so read it as 'lower your expectations for September.'
Q3. A certain month has a high average return but a low median. What should you suspect first?
The median is less shaken by extreme values. If only the average is high and the median is low, a few big numbers may have pulled the average up.
Q4. The probability of rising for all months of the S&P 500 is about 60%. How should you read a month with a 61% probability of rising?
A probability of rising only means something when compared with the baseline. Around 60% is similar to picking any month at random.
Q5. Which best describes how this lesson says to use seasonality?
Seasonality is a filter, not a standalone signal. Use only patterns that can be explained, together with other evidence.
NextTerms · Next lesson · Related tools
A historical tendency for stock returns to lean one way in the same stretch of the calendar, such as a particular month, weekday, or the turn of the year. It does not mean the pattern repeats every year; it is a statistical tendency in which the average return and the share of up periods have been tilted.
Probability of a rise (win rate)Win rate, % upThe share of the whole sample in which the return was positive. Look at it together with the average to check whether the average is being pulled by a few large numbers.
MedianMedianThe value that falls in the middle when values are lined up by size. It is less affected by one or two outliers, so it is used to check whether the average has been inflated.
Sample sizeSample size (n)The number of observations used to calculate a statistic. In monthly seasonality it means how many times that month has occurred; the smaller the number, the more likely the result is due to chance.
BaselineBaselineThe performance of "any random time," against which you compare a particular period. For example, only by comparing with the average of all months for the S&P 500 (+0.75%) and its share of up months (about 60%) can you tell whether a particular month was really strong.
Data mining biasData-mining biasThe phenomenon where, if you search through many periods and indices, a plausible-looking pattern appears purely by chance. It is filtered out with a causal explanation, out-of-sample periods, and data from other countries.
For educational purposes only; not investment advice. Past statistics do not guarantee future returns. S&P 500 figures were calculated directly from month-end values of Yahoo Finance ^GSPC daily closes (price index, excluding dividends and trading costs, Feb 1950–Sep 2026, retrieved 2026-10-09).