October 19, 2005

The Halloween Effect - True or False?
by John Brasher, CallWriter Publisher

It has been a market bromide for decades: "Sell in May and go away." This piece of market wisdom is sometimes referred to as the Halloween Effect, a strategy for going long on stocks on October 31 and selling the positions on May 1 for a gain. It assumes that, with few exceptions, the November through April period will be more bullish than the May through October period. But is it true?

The Halloween Effect

Actually, it is true. One study done in 1999 extensively reviewed stock market prices in all the world's major markets, and even many minor markets (e.g., Denmark) going back many years. At least since 1970, the data shows that the Nov/Apr period significantly outperforms the May/Oct period in 36 of 37 markets studied. The Halloween Effect is much stronger in some markets than others, and is particularly strong in Europe. The effect is extremely pronounced in the U.K., with evidence for its existence going back to 1694! It is also clearly present in the United States markets, if not quite so strongly. Sven Bouman and Ben Jacobsen, The Halloween Indicator, ‘Sell in May and Go Away’: Another Puzzle (1999).

The figure below, taken from that study, shows the respective returns for the Nov/Apr and May/Oct periods from 1970 through 1998 for many country stock markets, including the United States:

That's pretty compelling evidence! While particular year represents exceptionss, such as 1987 in which a U.S. market crash occurred, the stock market almost always does better in the Nov/Apr half of the year. The study's authors looked at many possible causal factors, such as interest rate fluctuation, people taking summer vacations May/Oct, whether the effect is sector-specific (it isn't), variations in trading volume, seasonal news and the like. None of them remotely accounts for the effect.

While the data is unequivocal and clearly establishes the effect's existence worldwide, what we don't know is WHY it happens; just that it does. And it happens consistently even though it is well known and expected, meaning that arbitrage and trading patterns taking the Halloween Effect into account have not canceled or really even modified it, as one would expect.

Theoretically, the effect's existence is implausible, all the more since no one can ascribe a reason to it. Efficient market theory in particular would predict that the effect actually is impossible. Don't laugh; a scientist once "proved" mathematically that bumblebees cannot fly. Put differently, the chance of finding a Halloween Effect for the Nov/Apr six-month period compared to the May/Oct period is 50%; even odds. But a 50/50 coin-flipping outcome is not what happens in the real world - the effect's consistency and continuity is overwhelming. So much for the random walk theory...

Newsletter author Mark Hulbert recently did his own analysis of the Halloween Effect since 1896 and found that it has been pronounced in the U.S. stock market during the last 50 years, but was not really found in the 1896-1951 period (the above study did not address the effect's existence in the U.S. prior to 1970). Hulbert suggests that no one can explain why it was present since 1951 but not before that. And right he is. But this is hardly surprising, since no one can explain the effect at all. Yet it occurs in virtually every major and minor market; even in the Southern Hemisphere, where the seasons are reversed and Christmas occurs during their summer time.

But a Related October Myth Falls

Many Wall Street myths just don't stand up to analysis of the price data. A good example is the election-year effect, which posits that markets do better in election years; but so many election-year markets or humdrum - or down - that it's hard to make that case. And heaven knows, incumbent Presidents try to make the effect appear. The January Effect and many others just can't be validated on the numbers.

Another Wall Street canard has long been that October is the stock market's worst month of the year. This has always seemed to make sense to the market, at least to that segment that believed in the Halloween Effect. After all, a traditionally weak October would set the stage for and appear to magnify the Halloween Effect. And this October hasn't been any joyride. But after extensive data analysis, Mark Hulbert demonstrates that it just ain't so. Measured several different ways, October isn't remotely the worst month - August is. He did his analysis comparing the low price each month to the price at the end of the following month - for example, the low in October would be compared to the ending price for November.

Using this analysis, he found that October is in fact among the six strongest months of the year, year in and year out, which are: March, June July, October, November and December. This conclusion validates the Nov/Apr Halloween Effect, since four of the six strongest months occur in the Nov/Apr cycle! He doesn't know why this occurs, either. Well, neither does anyone else, but it's no coincidence.

Thinking that perhaps comparing a month's low to the next month's last price (which at most could involve two months of price data), Hulbert looked at longer-term comparisons. He concluded that the only time period over which October could be considered the low point of the year (compared to subsequent months) is the six-month period - in other words, the Nov/Apr period... once again validating the existence of the Halloween Effect.

A corollary to the "Sell in May" saying has been: "But remember to come back in September." The weight of evidence, however, argues in favor of coming back in October, not September.

What Does the Halloween Effect Mean for Traders?

I've noticed a certain amount of what seems almost to be panicky thinking in recent weeks as the market has sold off. Last week in particular set off alarm bells for some. A few writers speculate that the DJIA is looking to test the 10,000 support level. And of course that could happen. We could even be in for a major correction. The fact that the Halloween Effect has been so reliable for so many decades provides a comfort level that the market isn't about to go over the falls in a barrel; but no guarantee. While exception years occur, it's reliable enough to trade on and a hell of a lot of people and institutions do it.

So what does the Halloween Effect mean for traders? For long stock players (specifically, buy-and-hold investors with an intermediate-term horizon), the end of October is the time to be placing bets. For covered call writers, the end of October inaugurates a period in which the market generally rises - the November expiration month is where the ride begins. The Nov/Apr period is a great place for covered writers to be making money and can even be pretty good for writing out-of-the-money calls. Any long or bullish strategy usually works well in the Nov/Apr time corridor.

But while putting our faith in the Halloween Effect, let's keep our powder dry. If the market goes into a correction, keep your covered call money in your pocket while looking for great bear call spreads - option trades in which the trader sells a lower-strike call and buys a higher-strike call, generating a credit.

Traders should not be panicked by recent October pullbacks. Absent much stronger evidence of a major correction, there is no reason to doubt that the Halloween fairy is coming this season. Don't fear October; use it to your advantage.

Question:  When you place a limit net debit order on a covered call trade, will that be considered one or two commissions?

Answer:  A net debit covered call order is a type of limit order. Instead of entering a traditional limit order on both the stock buy and sale of the calls, it is possible to enter the order as a net debit, in which the limit specifies the maximum amount (debit) the trader is willing to be debited for the trade. Thus if the stock is $20 and the 20 Call is bid at $1.00 (the price at which market makers are willing to buy it), the trader could simply enter the order with a net debit limit of $19. The trader might get a better fill than that - perhaps $18.90 or better - but $19 is the most being risked to put the trade on. The trader in this example doesn't necessarily care what the stock and option prices are - the net debit is the limit. I always use net debit orders. Most brokers (not all) allow limit orders to be entered as net debits. Note that a covered call can be closed just the opposite way, by entering the order as a net credit.

Two trades still are being run, however, to effect the covered call write: the stock buy and the sale of the calls. So a covered call trade will involve two trades (legs) to open and the same number to close. So you've got two commissions to pay on open and close.

Good luck and good trading!