Every year around this time, the same stat starts making the rounds.

September is the worst month for stocks.

And if all you do is look at the average return, nod your head, and get bearish, the research is pretty easy.

It’s also incomplete.

I spent this week digging through S&P 500 data going back to 1928 because I wanted to understand what’s actually behind September’s reputation.

Not just what September does on average, but when it tends to matter, what conditions make it dangerous, and what the current market actually looks like entering the month.

That’s the bigger lesson here.

Seasonality isn’t a signal by itself. It’s context.

And if you’re going to use statistics in markets, you’ve got to do the work underneath the headline number.

Let’s get into it.

The first mistake with seasonality is treating an average like a forecast.

September has a lousy reputation because some of its worst outcomes have been really bad.

Those tail events pull the average lower and make the typical September look worse than it actually is.

But rather than arbitrarily removing bad years, there’s a better question:

What did the market look like before those bad Septembers happened?

That’s where things get interesting.

Going back to 1928, when the S&P 500 entered September with a negative year to date return, September averaged negative 2.99%.

Only 38.7% of those Septembers were positive.

And when those Septembers were down, the average loss was a nasty negative 6.78%.

That’s where years like 2001, 2002 and 2008 live.

Now flip the conditions.

When the S&P 500 entered September with a positive year to date return, the average September return improved to basically flat at negative 0.16%, the median was +0.59%, and 52.2% of Septembers were positive.

Go one step further.

When the market entered September already up more than 10% for the year, September averaged +0.14%, the median was +0.97%, and 57.9% finished higher.

Same September. Completely different distribution.

That’s the point.

September doesn’t create weakness. It exposes it.

If there’s one filter from this study that really stands out, it’s the simplest one.

Is the S&P 500 above or below its 200 day moving average entering September?

Going back to 1929, there have been 70 Septembers where the S&P entered the month above its 200 day.

The results:

Average return: +0.02%
Median return: +0.50%
Positive: 52.9%
Stayed better than negative 2%: 74.3%
Lost more than 8%: just 1 of 70

Now look below the 200 day.

There have been 27 Septembers in that regime.

Average return: negative 3.82%
Median return: negative 3.67%
Positive: just 33.3%
Lost more than 8%: 7 of 27

The probability of an 8% plus September loss has been roughly 18 times higher below the 200 day than above it.

That isn’t a small difference.

That’s a different market regime.

And it makes intuitive sense. Weak markets are more vulnerable to volatility. Selling feeds on itself. Support breaks. Fear creates more fear.

Strong markets tend to absorb those shocks much better.

The calendar stays the same. The market underneath it doesn’t.

This is where statistics actually become useful.

We don’t want to know what September does in some imaginary average market.

We want to know which historical conditions most closely resemble the market we have today.

Entering September 2026:

S&P 500 year to date: +12.9%

That puts us in the stronger historical bucket where September has averaged roughly flat and the median return has been positive.

S&P 500 vs. 200 day: +10.4% above it

That puts us firmly in the regime where major September drawdowns have historically been much less common.

VIX: roughly 14

The VIX data only goes back to 1990, so this is where that shorter dataset belongs. Volatility isn’t elevated and there isn’t significant fear already priced into the market.

Presidential cycle: Midterm

That’s the caution flag. It argues for expecting more volatility, but the historical record doesn’t argue for automatically expecting downside.

None of this means September has to go up.

That’s not what we’re doing here.

Breadth has softened somewhat in the short term. There are areas of vulnerability, including parts of semiconductors and Consumer Discretionary. That matters because if volatility does show up, weak areas are usually where the damage shows up first.

But the broader trend remains intact.

And until the evidence changes, I’d rather respect the trend than fear the calendar.

The biggest takeaway from this study really isn’t about September.

It’s about how we use data.

Anyone can pull up a seasonal chart and tell you September is historically weak.

That’s the easy part.

The real work starts afterward.

What’s driving the average?

What happens in bull markets versus bear markets?

What happens above the 200 day versus below it?

What happens when the year is already strong?

What does the current market actually look like?

That’s how seasonality becomes useful instead of just another scary headline.

Knowing the statistic isn’t the edge.

Understanding the conditions behind the statistic is.

Or maybe even simpler:

The calendar tells us what to pay attention to. Price tells us what to do.

September can absolutely get messy.

But right now, the evidence says respect the possibility of volatility without automatically betting against the trend.

That’s the difference between using seasonality and being used by it.

Anyway, that’s my two cents.

Check it out.

🚀 Throw it on 1.5x speed and let it rip.

👍 Give it a like. It’s the easiest way to show me some love.  

The Sunday Stalk List

Tomorrow, I’ll break down more of what I’m buying in the Sunday Stalk List.

If you want clean charts, clear setups, and tactical insights, this one’s for you.

It hits your inbox every Sunday so you know exactly what to stalk for the week ahead.

Cheers,

Larry Thompson, CMT CPA

Follow Me On Twitter (X)

Follow Me on StockTwits