Seasonal Investing Patterns: What 'Sell in May' Really Means

Every spring the same line makes the rounds of trading desks, newsletters and family dinner tables: sell in May and go away. It is a rhyme, not a research finding, yet it has outlived several market regimes because it compresses a genuine observation — returns are not spread evenly across the calendar — into something you can remember. That is also the problem. A rhyme is easy to repeat and hard to falsify, and the distance between "there is a mild seasonal tilt in some markets" and "move your portfolio twice a year" is where most investors do themselves damage. What follows is where the pattern comes from, what the long-run data can and cannot support, and how to hold seasonality at a sensible distance.

Where the rhyme comes from

The phrase is usually traced to London, and to a social habit rather than a statistical discovery. The City's senior figures left for the summer, trading desks thinned out, and the market's centre of gravity shifted to the country. The full version — sell in May and go away, come back on St Leger's Day — points to a September horse race. It is a diary entry dressed up as an investment rule.

Its American cousin is the Halloween indicator: hold equities from November through April, step aside from May through October. Notice what the origin implies. The saying was built for a particular market, in a particular city, at a particular time. It was not a law of finance waiting to be discovered; it was institutional memory that later got tested. Some of that testing found a tilt. None of it found a rule.

What the historical record can actually support

Long-run studies of major equity indices, particularly in the US and UK, have generally found that the November-to-April stretch delivered higher average returns than May-to-October. That much is widely reported and broadly holds up over many decades. What gets lost in the retelling is everything that decides whether the finding means anything to you:

• The gap is modest. The winter months have tended to be better, not spectacular. A modest edge is easily swallowed by a single bad month, a currency move or a badly timed re-entry.
• The window you choose decides the answer. Start the "summer" in April and the pattern weakens; end it in September and it may flip. Serious seasonal work pre-commits to a definition rather than choosing one that fits.
• Dividends and costs change the picture. Price-only indices flatter a strategy that sits out six months. Total return, net of dealing costs and taxes, is the only number that matters.
• Averages are not years. The pattern is a property of a long series, not of the twelve months you will actually live through. Plenty of summers have beaten the winters around them.

There is a second, subtler issue: statistical significance. A seasonal difference that looks clear across ninety years can look like a coin toss across ten — and ten years is roughly the horizon over which an investor has to live with a decision.

Why an effect might exist, and why that is not enough

There are plausible mechanisms. Summer liquidity falls as desks empty, and thinner markets can behave differently. Tax years matter: December selling for tax losses in the US, an April year-end in the UK, both nudging flows. Earnings seasons cluster. Fund flows follow the calendar, and so does risk appetite — holidays, heat, holidays again. Older economies were agricultural, and credit tightened at harvest.

None of that is crazy. But public, documented, easy-to-trade patterns attract capital, and capital narrows them. The Monday effect is the cautionary tale: heavily documented in the 1970s and 1980s, it faded markedly after everyone knew about it. Publication is a solvent. A pattern that survives in the data may already be too small, after costs, to trade.

The costs the slogan leaves out

Acting on seasonality means getting two decisions right, not one. You have to leave at the right time and come back at the right time, and the second call is harder. The years that hurt are the ones where May-to-October rallies on thin volume while you wait on the sidelines, watching the thing you sold climb without you.

Add the practical drag. Every round trip costs commission, spread and, in many jurisdictions, tax on realised gains — a cost you may only defer, not avoid. Add the dividends you forgo while out of the market, and the compounding you interrupt. Add the behavioural cost: the discipline required to buy back in November after a strong summer is precisely the discipline most people lack. A seasonal rule that works on paper often fails in a human portfolio because the human is the weak link.

The rest of the calendar's folklore

Sell in May is only the loudest voice in a crowded room. You will meet these too:

• The January effect: small caps outperforming in early January, traditionally tied to tax-loss selling in December. It has weakened as it has become known.
• September as the weakest month: historically true in average terms for several indices, and useless as a forecast for any given September.
• The Santa Claus rally: the last days of December and first days of January. Small, erratic, endlessly written about.
• "As January goes, so goes the year": a genuine historical regularity that has become much less reliable, and which has no convincing causal story.
• The Monday effect: once a robust negative average for Mondays, now largely history — a useful reminder of what happens when a pattern gets famous.

How to pressure-test a seasonal rule

If a calendar idea crosses your path, run it through a few filters before it touches a portfolio.

1. Ask for the full series, not a chart that starts in a convenient year. Request results by decade, and by market.
2. Fix the rule in advance. Write down the entry and exit dates, the index, and whether returns are total. Then test. Do not adjust afterwards.
3. Count the losing years. A rule that is right on average but wrong four years in ten will feel very different to hold.
4. Charge it for costs and tax. Use your own dealing costs and your own tax position, not a frictionless academic assumption.
5. Look for the mechanism. If nobody can explain why the effect should exist and persist, treat it as a description of the past.
6. Size it as a tilt, not a switch. Shifting a slice of exposure is survivable if the pattern fails. Going to cash is not.

Keeping seasonality in its place

The useful version of sell in May is not a trading instruction. It is a reminder that markets have rhythms, that long-run averages hide a lot of variation, and that the calendar is one input among many — valuation, diversification, costs, tax and your own time horizon all weigh more.

For most investors, the practical response is to know the pattern, note it in a written plan, and otherwise leave the portfolio alone. If you are genuinely tempted to trade it, do it with a modest tilt and a pre-agreed rule for getting back in, so the decision is made by the plan rather than by the headline. And because the tax and account consequences of any calendar-driven switch depend on your jurisdiction and circumstances, take professional advice before you act on it.

Photo: RDNE Stock project / Pexels