The Halloween Effect
Ghosts, ghouls, goblins, and spooky math
The Halloween Effect might just be more of a trick than a treat — think less chocolate and more witch’s brew. The last day of October/beginning of November marks the start of a six-month period (November 1-April 30) during which, historical evidence shows, stocks worldwide have performed better when compared with the other six months of the year (May 1-October 31). Additionally, those returns have exhibited less volatility. Not only that, but the annualized returns from the aforementioned period tend to outperform those from a general buy-and-hold, equities-only strategy, often without an accompanying increase in risk.
Indeed, if we look at the last two years, the numbers seem to indicate something similar for the U.S. stock markets as per a mutual fund stand-in, the Vanguard Total Stock Market Index Fund (VTSMX). For the full year from November 1, 2016 to October 31, 2017, we see positive returns of almost 14% for the first six of those months as opposed to slightly less than 9% for the last six. This past year shaped up the same way: 3.81% with the Halloween Effect in play versus 2.60% from the first of May on through October.
(Of course, note that if you instead count May as the start of the year and the following April as the end, then 3.81% from November 2017–April 2018 doesn’t stack up well against that near 9% from the previous May 2017–October 2017 stretch. However, the pattern has held most of the time, and a comparison of mean returns from each half a year going back a ways demonstrates this.)
Such outperformance continues to lend credence to the concept of market timing and fuel the debate over active versus passive investment strategies. It also calls the practicality of Modern Portfolio Theory somewhat into question, at least with respect to the Efficient Market Hypothesis, reminding us that there seems to be something to behavioral finance.
And so, with another potential bumper period for stocks possibly about to get under way over the next half a year (or not, with all the volatility in the air of late), we owe it to ourselves to address the topic and speak to why we shouldn’t give the wolf in sheep’s clothing any candy; or, why it is not advisable to indulge in this active strategy in spite of evidence of a persistent seasonal pattern.
First, a quick review of the phenomenon in question. In 2002, Sven Bouman and Ben Jacobsen from the Netherlands published a paper in the American Economic Review entitled, “The Halloween Indicator, ‘Sell in May and Go Away’: Another Puzzle.” Therein they presented evidence that stock markets in many countries exhibit a strange seasonal anomaly and have been doing so for some time. To restate, six-month stock returns from the November-April period tend to outperform the six-month returns for stocks from May-October of any given year, and they tend to outperform the returns from passive buy-and-hold strategies when examined on an annualized basis.
In other words, without going too much into the mechanics of it all, you could apparently boost your investment returns by buying stocks sometime around now (as opposed to a diversified portfolio of equities and bonds and other asset classes) and holding them through April of next year, then selling off sometime at the end of April/beginning of May and moving your cash into short term, “risk-free” investments such as T-bills. After holding those through October, you’d then sell out of your T-bills at month’s end and move back into stocks. Rinse, wash, repeat.
This sequence explains the two nicknames the phenomenon goes by, suggested by the title of the Bouman-Jacobsen publication: The Halloween Effect, and Sell in May and Go Away. (It should be pointed out that while they were not the first to take note of the pattern, they were the first academics to thoroughly analyze it.)
Bouman and Jacobsen specifically examined the results from January 1970-August 1998, but three members of the Department of Finance at the University of Miami have since extended the study to 2012 and have found that, according to their methodology, the phenomenon persists. This is rather incredible given the low number of transactions you would undertake (meaning low transaction costs) and the simple fact that certain other calendar-based active trading strategies have failed to weather the test of time. The Halloween Effect, on the other hand, appears to withstand repeated scrutiny and the pattern has not dissipated upon being noticed by market participants.
The existence of such a trend is troublesome to some. The survival of such a distinct, easy-to-follow, repeatable, no-barrier-to-entry strategy flies in the face of a cornerstone of Modern Portfolio Theory, the Efficient Market Hypothesis (EMH), which dictates in part that there ought not be any discernible investment patterns; countless academic research pieces, literature, and theories; and entire bodies of thinking on lazy/passive/Boglehead/buy-and-hold investing approaches. It may even do a bit to invalidate the notion that one should have to educate themselves on investment matters at all or practice sound judgement (heaven forbid).
The persistence of the pattern notwithstanding, following such an active strategy could likely prove problematic one day, and here’s why (in no particular order):
Chasing ghosts: To engage in said strategy would be to disregard the maxim that past returns are not an indicator of future results, nor can they promise similar future returns. What has come to pass before does not guarantee that it will repeat itself. History helps to explain the way things are currently, it does not grant us the magical capacity to predict the future. Don’t fall prey to historical performance numbers and allow them to lull you into a false sense of security that things will always be this way, every year, like clockwork.
Tricked by ghouls: Hindsight is 20/20. We only know this pattern to exist in retrospect after each six month period is complete, which is to say that we always observe it after-the-fact. What’s more, nobody predicted that this particular pattern would emerge and persist. Herein lies an important distinction. If someone had predicted that the Halloween Effect would emerge, we could assume that they utilized some existing evidence and reasoning to arrive at the prediction, that they had actionable information or otherwise read the tea leaves correctly and could offer up a concrete explanation, but instead, we are aware of the pattern only because it has come to pass; the pattern (still) exists ultimately by chance alone and remains somewhat of a mystery. Needless to say, chance is not a sound basis for investment decisions.
To expand on this latter point, consider the following “casual observation” made by Edwin D. Maberly and Raylene M. Pierce in their response to the Bouman-Jacobsen paper, “Stock Market Efficiency Withstands another Challenge: Solving the ‘Sell in May/Buy after Halloween’ Puzzle”. Intriguingly,
…a preponderance of major economic and/or political events that negatively impacted world equity prices have occurred during the May-October periods.
Specifically, they call attention to the stock market crash of ’87 (October, 1987) and the catastrophic collapse of hedge fund Long Term Capital Management (August, 1998). After adjusting for these “outlier” events in their regression analysis, they find the pattern to be much less apparent for their stand-in for the markets in the United States, the S&P 500. (When they adjust for another anomaly known as the January Effect as well, the pattern dubbed the Halloween Effect becomes that much less statistically significant).
Let’s fast forward from 1998 to 2008. As you can see from the above graph of SPY (the SPDR S&P 500 ETF Trust, years 2004–2013 shown here), you would have avoided the most sickening of plunges of the US markets (early October, 2008) during that global financial crisis by paying heed to the Halloween Effect. More importantly, though, the root causes of each of the three financial fallouts mentioned above were long baking in the oven.
While some market participants had a sense that these crises were brewing, you would have been hard pressed to find anyone who could have predicted the exact month in which these buildups would have culminated in such dire results. Not to mention that some months out of the off-period may have been otherwise positive for stocks.
Yet, when looking at entire six month periods punctuated by such dramatic events as these, those other benign months and their results are overshadowed. Once again, this all smacks of chance. Are you willing to bet that a preponderance of such events will continue to occur sometime in the May-October timeframe, well into the future?
Feeding your money to the goblins: The reality that this pattern has held many times does not reduce the risk involved in subscribing to it. If it so happens that you employ this strategy on an off year for this phenomenon, you may feel more pain than anything else in holding a non-diversified, 100% stocks portfolio. The result could be lower returns or — worst-case scenario — loss of capital and increasing distance between you and your financial goals. And yes, there have been off years. Who knows with all the recent volatility, this latest go around could shape up to be yet another. Essentially, there’s no room for error and depending on your willingness, ability, and need to take financial risk, you could actually end up doing yourself a disservice.
The spooky math: The mean standard deviation of returns from such a strategy has been — and will continue to be — greater than that from, say, a well-balanced and diversified lazy three-fund portfolio comprised of domestic and international equities and bonds (and perhaps other poorly intercorrelated asset classes). Put another way, the variance of returns will be greater, meaning that when all is said and done, the real return can differ from the expected return by a larger amount as compared with the safer passive and diversified alternative.
Translation: the volatility of returns on the all-stock portfolio from November 1-April 30 will be higher than that of a diversified one, even if it is consistent with that of a buy-and-hold, equities-only approach or an improvement over the volatility of returns from the other half of the year. That volatility increases your exposure to the risk of not realizing the returns you expect and can rattle a person more than a fright at the local haunted house.
If you remain lured by the promise of higher returns from a mechanical equity strategy and the costumed wolf continues to have your ear, simply recall from our little conversation here that not every year in historical record has panned out in favor of the Halloween Effect, and chances are that moving forward, not every year will.
Otherwise, if you still find yourself captivated, some final words of wisdom from Richard Wyckoff dating back to 1930 might help:
At the time many thought that the market could be beaten by mechanical methods; that is, by some means other than human judgment. [Charles] Dow suggested a few of these. [Roger] Babson had one or more. All kinds of individuals came forward with ways of beating the stock market; each was certain his method would make a fortune. Not long afterward, however, after further study, I decided that methods of this kind, which substitute mechanical plays for judgment, must fail. For the calculations on which they are based omit one fundamental fact, i.e., that the only unchangeable thing about the stock market is its tendency to change. The rigid method sooner or later will break the operator who blindly follows it.