Calendar effects in investing: January, sell-in-May, and what the evidence really shows

The January effect, the sell-in-May seasonal, the turn-of-the-month effect, and the holiday effect are all calendar-based patterns that have been documented in equity returns at various points in the last fifty years. Several of them were genuine in their time. Most have weakened or disappeared once the strategies designed to capture them became known, which makes calendar effects a useful case study in the difference between an empirical regularity and a tradable edge.

What calendar effects are

Calendar effects are systematic patterns in asset returns associated with specific points in the calendar—particular months, particular days of the week, particular dates relative to month-end or year-end. The patterns are statistical: an asset class is observed to deliver higher (or lower) average returns at certain calendar points than at others, and the gap is large enough to be statistically meaningful in the historical sample.

The January effect, first documented by Rozeff and Kinney (1976), describes an apparent excess return on small-cap US stocks in January relative to the rest of the year. Sell-in-May (the "Halloween effect") describes the pattern that equity returns from November to April have been materially higher than equity returns from May to October across many markets and decades. The turn-of-the-month effect describes higher returns on the trading days around month-end and the start of the next month relative to the middle of the month. Each pattern was first documented from a specific historical sample, and each has been re-tested many times since.

How they were proposed to work

The proposed mechanisms vary by effect. The January effect was attributed primarily to tax-loss selling: US investors sell losing positions in December for tax reasons, depressing prices, and the rebound when this selling pressure ends produces the January excess return. The effect was strongest among smaller, more thinly traded stocks, where end-of-year selling pressure was easier to dislocate prices and the rebound easier to capture.

Sell-in-May has been attributed to seasonal patterns in economic activity, in capital flows (year-end bonus payments and investor repositioning), and in vacation patterns that thin trading volumes during summer months. None of these mechanisms is uncontroversial, and the empirical persistence of the pattern across markets with very different economic structures suggests the underlying cause is not fully understood.

The turn-of-the-month effect has been linked to the timing of pension fund cash flows: monthly pension contributions are typically invested at month-end, producing a buying pressure that lifts returns around that point in the calendar. The mechanism is plausible but contested, and the size of the effect has shrunk substantially in recent decades.

What the evidence shows

The January effect has weakened materially since it was first documented. Mehdian and Perry (2002) and others have shown that the January excess return on US small caps has shrunk to a fraction of the original 1976 estimate, and in some recent samples it is statistically indistinguishable from zero. The pattern was first documented, it became known, capital deployed against it, and the apparent edge largely disappeared. This is a textbook example of the McLean-Pontiff (2016) finding: factor and anomaly returns decay materially after publication.

Sell-in-May has been more durable. Bouman and Jacobsen (2002) document the pattern across 37 markets over 1970–1998, and follow-up work has confirmed its persistence into more recent samples. The annualised gap between November–April and May–October returns has been approximately 10 percentage points in major equity markets over multi-decade windows, with the pattern present in nearly every market tested. The mechanism remains contested, but the empirical regularity has not faded as dramatically as the January effect.

The turn-of-the-month effect remains visible in some samples but is small enough relative to transaction costs that capturing it through active trading is unlikely to be profitable for retail investors. The pattern is more useful as evidence about the structure of trading flows than as a tradable strategy.

Limitations and trade-offs

Calendar effects illustrate a structural problem in empirical finance: any sample large enough to detect a small statistical pattern will also produce many false patterns by chance. With dozens of plausible calendar slices to test, the probability that at least one will produce a statistically significant result by chance is high, even if no genuine effect is present. The published literature systematically reports the patterns that pass statistical tests, which biases the published evidence toward overstating the persistence of calendar effects.

Even genuine calendar effects are often smaller than transaction costs and bid-ask spreads, which makes them statistically real but practically irrelevant. The January excess return on small-cap stocks, even in the original Rozeff-Kinney sample, was small enough that capturing it required low-cost trading that was not available to retail investors at the time. Modern trading costs are lower, but so is the magnitude of the effect.

The strongest practical conclusion is that calendar effects make poor standalone strategies. They can be useful as small overlays on a more robust underlying methodology—a slight tilt toward defensive positioning in a known weak month, for instance—but they cannot be relied upon as a primary source of return. Strategies that depend heavily on calendar patterns are the easiest to overfit and the most likely to disappoint out of sample.

Calendar effects in pfolio

pfolio's monthly rebalancing schedule is calendar-driven but signal-blind to seasonality: the platform does not adjust allocations based on calendar effects. Investors can backtest the impact of any seasonal pattern on the historical return series by comparing portfolios constructed with different rebalancing rules in the platform.

Related articles

• Systematic vs discretionary investing: why rules-based strategies outperform over time
• Backtesting investment strategies: methodology, limitations, and how to avoid overfitting
• Overfitting in quantitative investing: why backtested strategies fail in practice
• Factor investing explained: momentum, value, quality, and low volatility