If you spend any time around markets, you'll eventually hear someone repeat a calendar rule with the confidence of a law of physics: sell in May, avoid September, expect a rally around Christmas. These sayings stick around because they're catchy, easy to remember, and — unlike most market lore — they actually have some statistical backing. The real question isn't whether seasonal patterns exist in the historical record. It's whether they're strong enough, and reliable enough, to do anything about.
This guide walks through the patterns that show up most often in U.S. and global stock data, what serious research has actually found, and — just as important — where these effects quietly fall apart once you account for the way real money is taxed and traded. Treat what follows as context for understanding the calendar, not a system for beating it.
The 'Sell in May' / Halloween effect
The most studied seasonal pattern is the so-called Halloween effect, captured in the old adage "sell in May and go away" — the idea being that you sit out the summer and return around Halloween. The pattern claims that returns from roughly November through April have historically been higher, on average, than returns from May through October.
This isn't just folklore. Economists Sven Bouman and Ben Jacobsen documented the effect across a large number of countries in a 2002 paper, finding that the winter half of the year tended to outperform the summer half in most of the markets they examined, often by a noticeable margin. The persistence across so many different markets is what made the finding hard to dismiss as a fluke in one dataset.
The Halloween effect is one of the few calendar anomalies that has survived being documented — but surviving in the data is not the same as being tradeable after costs.
What the research does not say is that summer is a losing period. On average, the summer half still tends to deliver positive returns in many markets — just smaller ones than winter. "Sell in May" is really a statement about the difference between two halves of the year, not a signal that stocks crash in June. That distinction matters enormously, because a strategy built on sitting in cash for six months gives up real expected return for a pattern that is, at best, a modest tilt.
Why September has a bad reputation
September is the month investors love to dread. Across long stretches of U.S. market history, September has been one of the weaker months for average returns — often the only month with a negative average over very long windows. That reputation is real in the aggregate.
But "weak on average over a century" hides a messy reality. Plenty of individual Septembers have been strongly positive. The negative average is driven partly by a handful of brutal months scattered across decades, not by a steady, dependable drift lower you can count on this year. Nobody has a convincing, settled explanation for why September underperforms — theories range from tax-related selling to post-summer portfolio repositioning to pure coincidence dressed up after the fact. When the mechanism is unclear, you should hold the pattern loosely.
The year-end and Santa Claus rally
At the other end of the calendar sits the cheerier folklore. The Santa Claus rally traditionally refers to a tendency for stocks to rise in the last few trading days of December and the first couple of January, and more broadly there's a perception that the fourth quarter and the turn of the year are friendly to equities.
Several non-exclusive stories get offered for this: lighter holiday trading volume, year-end bonus money flowing into the market, tax-loss selling finishing up and reversing, and fund managers tidying up positions. Some of these are plausible. None of them guarantees the pattern repeats in any given year. The window is short, the historical edge is small, and like every calendar effect, it's an average wrapped around a wide spread of outcomes — including years where late December was flat or fell.
Why these patterns might exist at all
It's worth pausing on the "why," because a pattern with no plausible cause is more likely to be noise that will vanish. A few honest possibilities:
- Real seasonal flows. Tax deadlines, bonus cycles, and institutional rebalancing do happen on a calendar, and they could nudge demand at predictable times.
- Behavioral and mood effects. Some researchers have tied seasonal mood shifts (less daylight, summer risk-taking) to risk appetite, though this is contested.
- Data mining. If you slice a century of returns by month, week, and holiday, something will look significant by chance. Patterns discovered after the fact deserve heavy skepticism until they hold up out of sample.
Seasonality also doesn't operate in a vacuum. Calendar tendencies can be completely swamped by the bigger forces driving markets in a given year — a rate-cutting or rate-hiking cycle from the central bank, or a sharp deterioration in market breadth. If you want to understand why a "weak" month ripped higher anyway, the answer is far more often found in Fed policy and the macro backdrop than in the calendar.
See it on your own assumptions
Averages are slippery. A pattern that looks compelling over 1950–2020 can look very different over the last 20 years, or in a different index, or once you change the start month by a few days. The most honest way to relate to seasonal claims is to test them against your own time window and your own market rather than taking a snippet of folklore on faith.
That's exactly what our Seasonality Heatmap is for. You can pick a market, choose a date range, and see month-by-month historical returns laid out as a grid — including how wide the spread is around each average. Run the numbers yourself and you'll quickly notice how a single ugly year can drag a whole month's average down, and how thin many of these "edges" really are.
Where it falls short
This is the part the calendar sayings always skip. Even when a seasonal pattern is genuinely present in the data, several things stand between it and your account:
- The effects are small. The historical seasonal "edge" is a modest tilt in average returns, not a large, dependable gap. Small edges are easy to lose to ordinary market noise.
- They're noisy year to year. An average drawn from many decades tells you almost nothing about what a single September or December will do. The year you act on the pattern is the year it can simply fail to show up — or reverse.
- Trading costs eat the edge. A strategy like "sell in May, buy back in November" means moving in and out of the market on a schedule. Spreads, commissions where they apply, and slippage all chip away at a thin advantage.
- Taxes are brutal on calendar trading. Selling each spring in a taxable account can trigger realized gains taxed at higher short-term rates, turning paper outperformance into a worse after-tax result than simply holding.
- Being out of the market is its own risk. A handful of the market's best days drive a large share of long-run returns, and they don't politely wait for "in-season" months. Sit in cash to dodge a weak stretch and you risk missing the rebound that more than makes up for it.
- Anomalies can fade once known. A widely publicized pattern invites people to trade ahead of it, which can shrink or erase the very edge that made it famous.
Put those together and you get the consistent finding behind almost all market-timing research: most active timers underperform a simple, low-cost buy-and-hold approach over long periods, once costs and taxes are counted. Seasonality is interesting context. It is not a license to jump in and out of the market, and the historical record does not support treating it as one.
Frequently Asked Questions
Is "sell in May and go away" actually real?
There's genuine evidence that the November-to-April half-year has historically outperformed the May-to-October half in many markets — the Halloween effect documented by Bouman and Jacobsen (2002). But "real in the data" and "worth trading" are different things. The summer half is usually still positive on average, the gap is modest, and the costs and taxes of switching in and out of the market tend to erode whatever edge exists. It's better understood as a curiosity than a plan.
Should I really avoid the market in September?
September has a poor long-run average, but that's an average across many decades, and plenty of individual Septembers have risen strongly. There's no reliable mechanism that tells you this September will be weak. Sitting out a month because of its historical reputation also risks missing strong days that disproportionately drive returns. Use the pattern as mild context, not a signal to go to cash.
How can I check these patterns for myself?
Don't take a one-line saying at face value — look at the actual month-by-month history for the market and time period you care about. Our Seasonality Heatmap lets you do exactly that, including showing how much each month's returns vary year to year. You'll often find the "edge" is far thinner and noisier than the folklore suggests.
Do seasonal patterns override what the Fed or the broader market is doing?
Almost never. Calendar tendencies are small compared with the major forces driving markets in any given year. A monetary policy shift or a collapse in participation across stocks will overwhelm a seasonal tilt easily. If you're trying to read the environment, start with Fed policy and market breadth before you reach for the calendar.