Market folklore: can you rely on the seasonal strategists?

Seasonal trends such as the Halloween indicator are part of market folklore – but are they reliable?

Wall Street traders wear 2013 new years party glasses: the four-day trading period between Christmas and the new year has seen average gains of 1.06 per cent since 1896 – 10 times that recorded by other four-day trading periods. photograph: allison joyce/getty images

November may be dreary, but seasonal market strategists always welcome its arrival, noting the bulk of market gains have accrued during the November-April period. Seasonal trends – the Santa Claus rally, the "sell in May" effect, the January barometer – have long been part of market folklore, but are they supported by evidence? And if so, why?

Halloween indicator
Also known as "sell in May and go away", a $10,000 November-April investment in the Dow Jones Industrial Average in 1950 compounded to $694,073, compared to a small loss for May-October.

One study found the effect in 81 of 108 countries analysed, with annual global returns of 6.9 per cent between November-April compared to 2.4 per cent from May through October (the gap has widened further over the last 50 years). It’s especially strong in Ireland, and dates to 1694 in the UK.

Why? Seasonal affective disorder, investors’ summer holidays, pension flows and all kinds of unsatisfactory explanations have been offered. One paper suggested an “optimism cycle”, arguing investors are optimistic as they look towards the new year, but this later becomes more difficult to maintain, creating a summer lull. The theory is supported by earnings evidence, analysts initially predicting optimistic profits growth before revising downwards as the year progresses.


Caveats? The strategy didn’t work in the US between 1900-1950, denting claims of its robustness. Secondly, questions regarding a fundamental catalyst mean it can’t be assumed the divergence will continue indefinitely.

Santa Claus rally
The Santa Claus rally, according to the annual Stock Trader's Almanac, refers to the last five trading days of December and the first two days of January. Since 1950, the S&P 500 has risen by an average of 1.5 per cent during this period.

Market commentator Mark Hulbert notes the four-day trading period between Christmas and the new year has seen average gains of 1.06 per cent since 1896 – 10 times that recorded by other four-day trading periods. Santa has cheered investors on 78 per cent of occasions, compared to just 55 per cent for other four-day periods.

Why? Christmas bonuses being invested and the Christmas feel-good factor are the most common explanations. Markets’ general tendency to be strong at the beginning and end of months is another obvious factor.

January barometer
As goes January, according to the Stock Trader's Almanac, so goes the year. The Almanac, which says a number of market-moving events being crammed into January make it a predictive month, says the barometer has been badly wrong on just seven occasions since 1950. A positive January is particularly predictive, it says.

However, while one study found January has been predictive in the US dating back to 1940, it underperforms a buy-and-hold strategy. Furthermore, of 18 countries studied, results were statistically significant in just four cases.

Additionally, the excellent CXO Advisory website, which analyses academic papers, backtests strategies and frequently debunks the claims of supposed market gurus, notes that January has no more predictive power than April, May, August or December if one goes back as far as 1871. This inconsistency undermines the indicator’s credibility, it concludes.

Presidential cycle
Politicians' manipulation of the economy, theorists say, explains why, since 1833, market returns during the third and fourth years of US presidential terms are more than triple that recorded in the first and second years.

Year three is strongest. Ned Davis Research found markets averaged gains of 11 per cent in year three compared to 4 per cent in year one between 1900 and 2010. The Dow hasn't had a losing third year since 1939. Additionally, US election games explain a "sizable fraction" of returns in other markets, Goldman Sachs said last year. Sceptics say the small sample size means it may be a fluke, but the theory is certainly believable.

Turn of the month
Markets tend to be very strong at the beginning and end of months. One study examined returns beginning the last trading day of the month and ending three days later. Between 1897 and 2005, all excess market returns occurred during this period. That is, "during the other 16 trading days of the month, on average, investors received no reward for bearing market risk". The effect was found in 30 of 34 countries.

CXO Advisory recently examined market performance during the first and last four days of all months since 1950, and found returns were almost triple ordinary eight-day periods. It was also evident in the small-cap Russell 2000 and all nine S&P 500 sectors examined and was “robust to both bull and bear states”.

Additionally, MarketSci blogger and quantitative strategist Michael Stokes notes that, despite being in the market for just one-third of the time, the strategy outperformed buy and hold between 1970 and 2008. The average drawdown was just 4.1 per cent, compared to 11.5 per cent for the S&P 500, so volatility was also lower.

Why? The most common suggestion is that pay cheques are being put to work in the market but the initial study referred to could find “no good explanation”. This “persistent peculiarity in equity returns poses a challenge to both ‘rational’ and ‘behavioural’ models of asset pricing”, it concluded.

Other patterns
Countless other patterns have been documented – September's ugly reputation, pre-holiday rallies, market declines on Mondays among them.

Many are inherently unreliable or likely to vanish in time. For example, the Monday blues identified by Prof Jeremy Siegel are no more, the day-of-the-week effect disappearing from international markets.

Similarly, the Nasdaq rose on the last trading day of the year on every occasion between 1971 and 1999 before falling on 11 of the next 12 occasions. And only a fool would deduce anything significant about the fact US markets have not fallen on a year ending in a “5” since 1875.

Chance dictates that apparent “patterns” will emerge, patterns that salesmen, spoofers and the statistically illiterate are only too happy to trumpet.

However, it would be premature to dismiss all seasonal trends as fluke or data mining. Michael Stokes admits he is “not a big fan of seasonality plays” but the turn-of-the-month phenomenon is “strong enough and consistent enough” to demand consideration.

Money manager Mebane Faber says such strategies "must have a fundamental reason" behind them, which is why he can "get on board with the presidential cycle". Similarly, psychologist and trader Brett Steenbarger has written that while he does not trade patterns mechanically, he does "focus on the most robust patterns and utilise them as context for trade ideas".

Ultimately, it seems, markets are not so inefficient that they can be easily exploited, but nor are they so efficient as to be devoid of quirks.