The September Effect in the Stock Market: Real, Small, and Unexplained

The september effect in the stock market is one of the few seasonal claims that survives contact with the data. It is also one of the smallest. Both things are true at once, and the distance between them is where ordinary investors do themselves damage.

What the Record Actually Says

Schaeffer's Investment Research put the long-run figures on the page in a piece dated 7 September 2023, written by its senior quantitative analyst Rocky White. Reaching back as far as 1950, the index has averaged a 0.66 per cent September decline, with 44 per cent of those months finishing higher, and measured only from 2000 the average decline widens to 1.51 per cent while the share of positive months rises to 48 per cent. Those are 2023-vintage numbers, which matters: they predate three of the Septembers a reader has lived through since.

Other publishers land near but not on those figures, because the choice of start year and the choice between price returns and total returns both move the answer. The Stock Trader's Almanac has tracked calendar patterns for decades and arrives at the same ranking by a method of its own. The ranking is stable. The size of the gap is not.

A Tilt, Measured Against the Noise Around It

Put the monthly figure next to the annual ones and the scale problem becomes obvious. In the return series Aswath Damodaran of NYU Stern maintains, the S&P 500 returned 28.47 per cent in 2021, then -18.04 per cent in 2022, and -36.55 per cent in 2008, all including dividends. A seasonal average under one percentage point is a rounding error next to swings of that size.

The hit rate deserves the same treatment. If somewhat more than half of Septembers finish lower, then somewhat fewer than half finish higher, and an investor who stepped aside every August across the record would have missed every one of the good ones. That is not a pattern anyone can trade without also being right about when to come back.

Nobody Has Produced the Cause

The explanations on offer are stories, and they are presented as stories here rather than as findings. Managers tidy positions after the summer. Holidaymakers come back and sell. Losses get harvested early. Each is plausible, none has been demonstrated, and an anomaly with no established cause is exactly the kind that can thin out once enough people position for it.

The pattern is also unstable under conditioning, which is the real warning sign. White's own work shows the weakness concentrating in the back half of the month rather than spreading evenly through it, and shows the whole picture inverting after a strong start to the year: in years when the index was up by double digits through August, September instead averaged a 0.61 per cent gain with 54 per cent of those months positive. A statistic that changes sign when you add one filter is a clue about the sample, not a rule about the calendar.

There Is a Real Event Sitting in the Middle of the Month

Here is something checkable that the seasonal write-ups tend to skip. The Federal Open Market Committee does not meet evenly through the year. It holds eight scheduled meetings, and four of them carry a Summary of Economic Projections, the quarterly document in which officials publish where they expect rates and growth to go. One of those four falls in September every year. In 2026 the committee met on 15 and 16 September with projections attached, and the minutes of that meeting were released on 7 October.

That is not a cause of anything, and it is not offered as one. It is a reminder that the month is not empty. The same calendar window that seasonal charts mark as soft contains a scheduled projection round, a quarter end, and the start of the run-up to third-quarter results. Any of those can move prices in either direction, and none of them cares about the month's historical average.

It also gives you something better to do with the month than count. Instead of asking how September usually goes, read what the committee published and what companies said, which is the same discipline that applies when deciding whether a drop is ordinary turbulence or something structural.

Turning a Description Into an Instruction

The usual misreading converts a historical description into a trading rule. The sentence about the month being the weakest becomes a decision to step out beforehand. The first sentence reports what happened. The second is a strategy, and the figures above do not fund it.

Count what the strategy demands. You need the exit right, the re-entry right, and enough margin to absorb tax on anything you realise along the way. The arithmetic of being out of the market at the wrong moment is punishing in its own right, as the record on missing a handful of strong days shows. A sub-one-point average tilt does not cover those costs for an ordinary investor.

A second misreading takes the pattern to be universal and permanent. The numbers quoted here describe one index in one country over one stretch of history, and the post-2000 slice already differs from the full sample.

Chasing the Cause Is the Part Worth Skipping

Most coverage of this pattern spends its energy on the wrong question. Column after column hunts for the mechanism, as though naming it would make the month actionable. The judgement here is that the cause hunt is a dead end and the useful work lies elsewhere, in what the statistic does to your behaviour.

Two reasons. First, a seasonal tilt this small cannot clear the costs of acting on it even if the cause were proved tomorrow, so the answer would change nothing. Second, the conditional results undercut the premise: a pattern that reverses in strong years, as White's filtered figures show, is not one stable mechanism waiting to be identified. It is a thin average over a very wide distribution.

What the statistic is genuinely good for is expectation setting. Knowing the month has a soft record makes a red fortnight in September feel ordinary rather than alarming, and the single most expensive thing an investor does is react to ordinary movement as though it were information. A fixed schedule handles the rest: if you reset weights on a date, as our piece on how often to rebalance sets out, a weak September simply hands you slightly different weights on that date. Bolt a seasonal rule over the top of that and you are running two plans that will eventually contradict each other, which is where careful people make careless choices.

Show a cause that is both specific and still operating, with the effect holding up out of sample after publication, and the pattern would be worth more than calm. That evidence does not exist today.

September Is Weak. That Is All It Is.

The record supports the reputation and refuses to support the trade. Let the statistic do the one job it can do, which is to keep you in your seat during an unremarkable month, and keep your attention on earnings, rates and the economy, which is where the movement actually comes from.

This article is for educational and informational purposes only. It is not financial advice, and Investor Muse is not a registered investment advisor. Historical figures describe the past and do not predict future results. Speak with a qualified professional before you make decisions about your money.