by Martin Zweig
As seen in the top summary line of table 27, covering a twenty-three year period which I have tabulated, there were a total of 223 pre-holiday trading days. Of these, the market rose in 193 cases, or 83% of the time. Only 28 times, or in 12% of the cases, did the market fall; the market was even 12 times, which is 5% of the cases. If you ignore those dozen cases of a flat market, the market rose 193 times out of 221 cases, a phenomenal 87% success rate. Again, ignoring the unchanged cases, it means that the odds of the market rising on the day before a holiday are about seven out of eight. Although these calculations were concluded in June 1985, a basically similar market pattern has persisted since then.
Had you invested an initial $10,000 on the pre-holiday trading strategy back in 1952, you would have had $27,188 in June 1985, despite having been in the market a total of only 233 days, which is actually less than the equivalent of one full year (there are about 255 trading days in a normal calendar year). The result shows an average gain by the broad market of .43% on the day prior to all these holidays. That works out to an extraordinary annualized gain of 189.7%. Indeed, the major drawback with holiday trading is that there aren’t enough holidays! If you could somehow invent a holiday for each trading day of the year, you could roughly triple your money in a year, and after a few years could retire. Of course, if every day were a holiday, the market would never open!
TABLE 27
PRE- AND POST-HOLIDAY PRICE ACTION ZWEIG UNWEIGHTED PRICE INDEX: January 1952 to June 1985
Direction of Market
Holiday
Up
Down
Unchanged
$10,000 Investment
Return per day
Annualized Return
Pre-holiday:
Easter 26 5
3 $10,906 +.26% + 68.0%
Memorial Day 25 4
5 $11,402 +.39% +111.7%
July 4 28 5
0 $11,644 +.46% +127.0%
Labor Day 31 2
0 $12,335 +.64% +180.4%
Thanksgiving 27 4
2 $11,325 +.38% +102.4%
Christmas 25 7
1 $11,302 +.37% +100.6%
New Year’s 31 1
1 $11,893 +.53% +146.3%
Post-holiday:
Thanksgiving 30 2
1 $12,286 +63% +176.6%
Christmas
22
10
1
$10,967
+.28%
+ 74.7%
Total 7 pre-holiday days only: 193 (83%) 28 (12%) 12 (5%) $27,188 +.43% +189.7%
Total of all 9 days: 245 (82%) 40 (13%) 14 (5%) $36,633 +.44% +179.6%
I unearthed a second fact concerning two of the holidays, namely that the day after both Thanksgiving and Christmas showed a very bullish bias as well. As you’ll note in table 27, the market rose 30 times after Thanksgiving, fell twice, and was unchanged once. The annualized gain of 176.6% is the second highest for all of the holiday cases. On the day after Christmas the market was up 22 times, down 10, and unchanged once. That’s the worst of the cases, but it still produces a nifty 74.7% annualized gain.
Over a total of 299 holiday trades, the market rose 245 times, declined 40 times, and was unchanged on 14 occasions. Ignoring the unchanged instances, this means that the market was up 245 out of 285 days, a success rate of 86%. A $10,000 investment would have grown to $36,633, a gain of .44% per day, or 179.6% per annum. Had you traded around these holidays over the years, you would have been in the market 9 days per year, and in those 9 days would have made 3.99%. In other words, by being in the stock market only about 4% of all trading days, you would have made roughly 4% on your capital. The other 96% of the time you could have earned the normal interest rate. Thus, by taking on only 4% of the market’s overall risk, you could have increased your return over that of a normal T-bill or a money market portfolio by nearly 4 percentage points a year. Of course, that assumes no transaction costs, which is not necessarily the case.
The question, then, is how to trade around the holidays without seeing most of your profits eroded by transaction costs. One way is to trade no-load mutual funds, especially since it’s easy to tell the funds in advance exactly which day you want to buy or sell. The drawback is that very few funds will be amenable to such in-and-out activity.
The second way is to trade stock index futures. Since the Value Line Index is nearly identical to my Zweig Unweighted Price Index, you could opt to trade Value Line futures on the Kansas City Exchange. Transaction costs are nominal. The drawback, however, is that the futures can trade at various premiums over the actual market index … or occasionally even at discounts. If the premium stays at the same relative level during the day that you trade, it would not be a factor. However, you might run into the disheartening situation where the premium shrinks and essentially wipes out your profit.
For example, suppose that on the day before a holiday the Value Line Index itself is at 200 and the future is trading at 204, a four-point premium. Suppose on the strong holiday the Value Line Index advances one point, equal to one-half of a percent, and closes at 201. It’s possible that the future may simply stagnate and stay at 204 on that day, not giving you any profit at all, as the premium shrinks from four points down to three points. Of course, if the future goes up in line with the actual index, it would rise a point to 205 and you would earn your profit.
Over the long haul, it’s probable that the futures will produce roughly the same amount of change as the actuals, but there’s no guarantee. Moreover, in some of the cases, you’re bound to feel the negative effects of premium shifts, but at other times the shifts in premiums may work for you. You could also trade futures on the New York Composite Index, the Standard & Poor’s 500 Index, or the Major Market Index, a composite of twenty blue chip stocks, traded on the AMEX.
There are other ways to benefit from the holiday tendency. If you are going to buy stock anyhow, it would not be a bad idea to buy it a day before the holiday in order to increase your odds of getting off to a good start. If you are thinking of selling a stock, don’t sell the day before the holiday. Rather, wait until the close just prior to the holiday, or even hold over the holiday and sell on the opening the day after the holiday … except, of course, for Thanksgiving and Christmas, when you should hold on for at least one more day.
In addition, if you are willing to risk the transaction costs and the lack of diversification, you might want to trade a few large market-type stocks around the holiday period. Such active market leaders as IBM, Digital Equipment, Merrill Lynch, or Texas Instruments might make good trading vehicles at that time. However, the price tendencies of these stocks might not show the equivalent performance of my broadly based Zweig Unweighted Price Index.
The obvious question in light of these seasonal tendencies is, why do they occur? As noted, it has nothing to do with economics, except possibly for some biases created by tax transactions around Christmas or New Year’s. The most compelling explanation I’ve come across is that people are affected emotionally around holidays. Most of us feel better prior to a holiday. Why not? It’s nice to know that a three- or even a four-day holiday is at hand. You can look forward to relaxation, time with the family, a visit to the country, or whatever. Under those circumstances, it would not be unreasonable for people to be more optimistic and thus more prone to buy than to sell stocks, thereby creating an upward bias in prices. If this tendency is true prior to holidays, it should also be true on Fridays. That is, people should be in a more upbeat mood approaching a regular weekend than they would otherwise normally be. We’ll see that this is, indeed, the case.
223 WINS VERSUS 8 LOSSES
We’ve seen that there’s an overwhelming tendency for stock prices to rise the day prior to a holiday. The day after Thanksgiving also has a tremendous upward tendency, possibly because it’s really part of a four-day holiday period for most people. On that day, which is always a Friday, people still have the weekend to look forward to. The day after Christm
as is more difficult to categorize, in part because the day of the week will vary. However, Christmas is usually the most upbeat of all holidays, and usually there’s still a dose of good cheer right after the holiday itself. Even so, as noted, the tendency on the day after Christmas is not nearly as strong as the pre-holiday days or the day after Thanksgiving.
Given these holiday tendencies, I have added a few new wrinkles to devise an even more potent trading strategy. It’s based on the premise that the market should go up during the aforementioned holiday periods—indeed, it does so seven-eighths of the time. But if the market does not do what it’s supposed to do prior to the holidays (or immediately after Thanksgiving and Christmas), then that in itself is a negative sign and increases the probability that stocks will fall in the short run. Thus, the general trading strategy is to observe prices during the pre-holiday period, and if, at the close of the day, the market is unchanged or down—defying the normal tendency—then one should sell short the market for the day after the holiday. This rule would apply to Easter, Memorial Day, July 4th, Labor Day, and New Year’s.
Obviously, the strategy would be somewhat different around Thanksgiving and Christmas, because we want to be long on the market on the day after those two holidays. For Christmas, though, we’ll adopt a rule similar to the other holidays. Namely, if the market is flat or down on the day following Christmas, then at the close of that day we would short the market for one day following—that is, the second day after the Christmas holiday.
For Thanksgiving, the rule is slightly different. Overall, the two days around Thanksgiving—that is, Wednesday and Friday—have extraordinarily strong seasonal tendencies. In only six of the past thirty-three years has the market failed to advance by at least one-half of one percent for the two days combined. Thus, I would consider a gain of less than one-half percent over that two-day stretch to be “inferior.” If the Zweig Unweighted Price Index rises by less than .50% during the Wednesday and Friday around Thanksgiving combined, then we’ll short the market at the Friday close and hold that short position for one more day.
Table 28 shows the results of this expanded trading strategy, which calls for the use of short sales—or negative bets—for one day following subpar performance if it occurs during the normally strong holiday period. For example, the first line of the table shows the 34 cases since 1952 involving Easter. As noted earlier in table 27, there were 5 cases when the market went down prior to Easter and 3 cases in which it was unchanged. In those eight years, you would have shorted the market just at the close prior to the Good Friday holiday and held through the close of Monday following the Easter holiday. Had you done so, you would have made money on the short side in all 8 cases.
The good results, for the short seller, on these Mondays would have given you a total trading profit for the Easter period in thirty-three of thirty-four years. You never would have lost money, and you would have broken even once. You can’t do much better than that. In a total of forty-two trading days (of which eight were on the short side) a $10,000 investment would have appreciated to $11,634, a gain of .45% per holiday, or an annualized gain of 99.2%.
The results for the other holidays are similar. The addition of 9 short-selling attempts around Memorial Day winds up producing a holiday profit in thirty-two of thirty-four years, with only two losing periods. For July 4th, Labor Day, and Thanksgiving the returns become truly spectacular as each holiday escapes with a perfect 33-to-0 record. In other words, in 99 holiday spans encompassing those three holidays, you would have made money 99 times and never lost! At Christmas time there were 27 winners, 5 losers and I tie, but that still works out to a nice 88.8% annualized profit around that holiday. New Year’s checks in with 32 wins and 1 loss.
TABLE 28
OPTIMAL HOLIDAY STRATEGY ZWEIG UNWEIGHTED PRICE INDEX: January 1952 to June 1985
Direction of Market
Holiday
Total Days Invested *
Up
Down
Unch.
$10,000 Investment
Return per day
Annualized Return
Easter 42 33 0 1 $11,634 +.45% +99.2%
Memorial Day 43 33 2 0 $11,767 +.48% +104.8%
July 4 43 33 0 0 $12,481 +.67% +147.1%
Labor Day 35 33 0 0 $12,657 +.72% +193.6%
Thanksgiving 72 33 0 0 $14,564 +1.15% +161.6%
Christmas 77 27 5 1 $12,680 +.72% +88.8%
New Year’s
35
32
1
0
$11,857
+5.2%
+135.3%
Total: 347 223 8 2 $47,353 +.45% +184.5%
(96%) (3%) (1%) +.67% per period
The number one holiday on an annualized basis is Labor Day, sporting a 193.6% per annum profit. However, the best total gain per holiday comes at Thanksgiving, when the market is up 1.15% per holiday, thanks in part to the fact that one is always long for two days at Thanksgiving. But the annualized gain for Thanksgiving is “only” 161.6%.
From January 1952 to June 1985, there were a total of 347 days when one would have traded the market around holidays, of which 48 days were spent on the short side of the market after the original holiday period turned in subpar performances. In that span, a $10,000 investment would have appreciated to $47,353, a gain of .45% for every day one was invested, or a profit of .67% per holiday period (there were 233 such holiday periods). This works out to an annualized gain of 184.5%. It is also equal to a gain of 4.78% per calendar year. In other words, had one engaged in this holiday trading strategy over the last thirty-three years, you would have made about 4.75% in the average year, despite being invested only about 4% of the total time. You would never have had a losing year!
Of the 233 holiday periods, 223 were winners, only 8 produced losses, and 2 were ties. Eliminating the ties, that works out to a 97% success rate. Some would argue that these returns are only theoretical and that they would be difficult to achieve in practice. If you tried to enhance the overall holiday strategy by selling short at the appropriate times, you could not use no-load mutual funds for this purpose. However, you could trade stock index futures. Again, though, you might encounter problems with the vagaries of movements in the premiums on the futures relative to the underlying stock index. But even if you do not trade stocks actively around the holidays, it must be acknowledged that the price behavior at these times is truly extraordinary and anything but random.
DAYS OF THE WEEK
One of the early studies on the effects of days of the week was by Art Merrill (Merrill Analysis, Box 228, Chappaqua, NY 10514). Between 1952 and 1974 he checked to see the proportion of days that were up or down on the Dow Jones Industrial Average. For all days together, the Dow was up 52.5% of the time. There was no significant difference during midweek. Tuesdays rose 51.8% of the time, Wednesdays 55.5%, and Thursdays 53.5%. However, Mondays lagged the normal tendency quite a bit, rising only 41.6% of the time, while Fridays were up 59.8% of the time. That strong Friday tendency is consistent with the behavior of prices seen prior to holiday weekends.
In other words, investors ought to be in better-than-normal moods prior to a weekend, although that tendency should not be as strong as it is prior to longer holiday weekends, which is, indeed, the case. By contrast, if emotions have any negative effect on any day of the week, it ought to be the first day of the business week, thanks to the “blue Monday” syndrome. It’s safe to say that most of us—if we’re going to get the blahs—are more likely to get them on Mondays.
Another early study, by Frank Cross (Financial Analyst Journal, November–December 1973), verified Merrill’s work on the strong Friday and poor Monday results. Cross used the S&P 500 Composite as his market index and found that, between 1953 and 1970, Fridays were up 62.0% of the time, whereas Mondays rose only 39.5%. The average percentage change was +.12% on Fridays and -.18% on Mondays.
In a somewhat more recent test by Michael Gibbons and Patrick Hess (Journal
of Business, 1981, volume 54, number 4), they checked returns by days of the week from 1962 through 1978. They found that the S&P 500 declined .13% on Mondays and rose .08% on Fridays. Moreover, they found that an unweighted price index (somewhat similar to my own Zweig Unweighted Price Index) fell an average of .11% on Mondays but rose a substantial .22% on Fridays.
So, the Friday and Monday effects on stock prices are interesting, but they probably don’t offer great profit opportunities for investors because of transaction costs. However, other things being equal, if you were going to sell late in the week, you would probably want to wait until Friday’s close or even Monday’s opening. If you were going to buy, you should not do so late Friday or during the day Monday. You are probably better off waiting until Tuesday. Of course, these tendencies might be overwhelmed by more important indicators such as those discussed in earlier chapters.
MONTHS
Table 29 shows the results of another Art Merrill study, this one on seasonal tendencies for months of the year from 1897 to 1974, which we have updated to 1996. The table shows the percentage of the months in which the Dow Jones Industrial Average advanced. It is seen that there are two significant times in which the months do the best: at year’s end, with November at a 59.6% success rate, December the top month with a 71.7%, and January at 65.6%; and then during the summer, with July up 61.6% of the time and August up 64.6%. By contrast, the average for all months to rise was only 57.0%. At the other end of the spectrum, September had the worst tendency, rising only 40.4% of the time, with February the second worst, up only 49.5%.
A different study, by Anthony Tabell of the brokerage firm Delafield, Harvey, Tabell, substantiates Merrill’s earlier findings. Covering the span from 1926 to 1982, Tabell found that, once again, December was up more than any other month. The same held true when we updated the study through March, 1996. In December, the S&P 500 rose 52 times and fell only 18. January was second best, gaining 45 times, followed by March, April, July, and August.