Martin Zweig Winning on Wall Street
Page 17
I count both the number of bullish ads and the number of bearish ads. However, I’ve found that the number of bullish ads is a more effective indicator since bearish ads tend to be fewer in number and are not as good a sample size. Graph M (pp. 144–145).shows the four-week average of bullish ads in Barron’s since 1974. The most striking factor is that back in 1974 there were periods in which there were actually no bullish ads whatsoever. That also marked the bottom area of the worst bear market since the Depression. By contrast, near market peaks in 1976, 1978, early 1980, and 1981, the number of bullish ads increased to about 20 per week on a four-week average. In 1976 prices moved sideways for the rest of the year, while in the other three cases the market plunged. In mid-1980 and in early 1993 the indicator also reached the 20 area, but the market went higher for a few more months, primarily because monetary conditions were still positive.
In 1983, before the market went into an intermediate decline, the peak of bullishness among the Barron’s ads was about 16 per week on a four-week basis. That was pretty high, but not ultrahigh as seen in the earlier cases. Of course, that range was reached as early as late 1982, but the market still kept on rising because monetary conditions were excellent and most other factors were still quite good. But, by the middle of 1983, as the other evidence began to turn more negative, stocks couldn’t buck that trend much longer and began to fade. In 1987 the number of bullish ads reached record proportions, giving warning of the October crash. By the end of 1987, when the stock market stabilized, the number of bullish ads was down to subdued levels.
Note that, as with all indicators, one must use Barron’s ads as part of a bag of tools. It’s too much to expect one indicator always to be correct. What you look for are tendencies. When enough indicators are showing tendencies in the same direction, it’s time to believe that those indicators, working in concert, will be correct.
Now look back at graph M and review the cases when the four-week average of bullish Barron’s ads dropped to about 7 or less. This was seen in fall 1975, mid-1982, mid-1984, and late 1984. In 1982 stocks embarked on their greatest bull market advance in more than fifty years. In the other three cases, intermediate declines ended and bull market advances renewed.
As a rough guide, I would suggest getting cautious when the number of bullish Barron’s ads reaches about 13 per week on a four-week basis, and I would tend to get more bullish when the figure drops to about 7. If monetary conditions are bullish—say, according to our model in chapter 4—I would raise the range toward about 16 on the negative side and perhaps 10 on the positive side. When monetary conditions are bearish, I would drop the range a few points. In that case somewhere around 10 to 11 might be considered dangerous, while 4 or fewer might be positive. Again, these are only rough guides, not absolutes.
As for the data, I’m the only one I’m aware of who actually takes the time and trouble to count these ads—I’ve been doing it since about 1972. Whenever the number becomes too extreme, I mention it in my advisory service, The Zweig Forecast. Other than subscribing to my service, about the only way to obtain these figures is to count the ads in Barron’s yourself. Of course, if you count them, you might come up with a slightly different number than I would. The reason is that some of the ads are vague, and some interpretation is necessary to determine whether they are bullish, neutral, or bearish. By the way, I ignore the neutral ads. Whatever biases I may have have been consistent for the thirteen or so years during which I’ve been keeping count.
GRAPH M
Ned Davis Research
The consistency of any bias is important in an indicator such as this. For example, if you were to start counting the number fresh, you might average 3 more bullish ads per week than I would. That might lead you to an opinion that optimism is too heavy, while I would take more time to reach that same conclusion. It’s just a fact of life that whoever counts these ads will have some bias in the interpretation.
The same is also true for Investors’ Intelligence’s survey of the advisory services. But only two people, the late Abe Cohen and the present editor, Michael Burke, have been doing the counting there. So once again, their biases have been consistent. If I tried to count the sentiment among the 140 or so advisors, I would surely come up with a different figure than they do. It’s not the absolute figure that’s important, but rather the deviation from whatever is normal, and normal includes the bias of the counter.
SECONDARY OFFERINGS
I’m trying to give you a flavor of various sentiment indicators, and, as with most endeavors in this book, I’m trying to keep it simple. I am deliberately avoiding indicators that deal with short-selling statistics and puts-and-calls option trading, for example, because they are too complex, and they have been subjected to great distortions in recent years for various reasons, primarily the advent of options trading in the 1970s and of futures trading in stock indexes in 1982. Besides, there are usually a lot of calculations involved in these particular sentiment indicators. I’d rather talk about indicators that need very little computation. In this vein, a worthy indicator, especially for calling market tops, is the number of secondary distributions, or secondary offerings.
A secondary offering is an issue of stock being sold by a company that is already public. It is not an initial public offering, or new issue as it’s commonly called. Because there already is stock out, this subsequent distribution of additional shares is classified as “secondary.” Another type of secondary distribution is when a very large block of stock, held by an insider or possibly another corporation, is marketed through a brokerage house in the form of a secondary distribution because it may be easier to sell that way than to have it hit the floor of the exchange. In such a secondary, a whole network of brokers at different brokerage firms is able to drum up interest in the distribution, much as with an initial public offering. It makes no difference whether the secondary is being sold by the company itself or by an individual or other corporate holders. A secondary is a secondary. You’ll find them listed in a table in the back pages of Barron’s each week.
Very few secondary distributions are seen once a bear market gets rolling. People are not interested in buying more stock at those times; moreover, if prices are depressed there is less interest by corporate or individual holders in selling such blocks. However, when bull markets heat up and speculative froth is in the air, the number of secondaries increases markedly.
This is so for two reasons: First, it’s a lot easier to sell such blocks when the market is roaring, because the public’s speculative appetite is whetted and it’s willing to bite. It’s easy to sell secondaries in a hot bull market, just as it’s easy to sell initial public offerings. Second, it’s more enticing for the selling company or selling shareholders to dump such blocks when prices are high. After all, everyone would rather sell stock when prices are high rather than when they are low.
I track both the number of secondary offerings and the total dollar amount involved. The dollar figures involve too many complications, but the absolute number of offerings, which is simpler, works even better. Graph N (pp. 148–149) shows the three-month average of secondary offerings going back to 1958. When the number of secondaries fades to about 3 per month or less, it indicates very little speculative activity and not much overhanging supply of stock—a relatively bullish condition. Such scores were seen at the bottoms or near-bottoms of bear markets in 1960, 1970, 1974, early 1980, and mid-1982. Other low readings came at good buying points in 1984 and late 1987. It is interesting that after the devastating bear market of 1973–74, the number of secondaries fell to just about zero. After that it took many years before secondaries increased in number toward the norm of the 1960s and early 1970s.
GRAPH N
Ned Davis Research
Generally, when secondaries are averaging fewer than 10 per month on a three-month average, and monetary conditions are favorable, it is bullish. But when monetary conditions are negative, secondaries ought to drop to between zero and
3 per month before the implication is positive.
The number of secondaries is even more valuable in suggesting that speculation has gone too far and that a top is being formed. As seen on the graph, the total of secondaries rose to above 25 per month in 1959, 1961, 1965, 1968–69, 1971–72, 1983, 1986, 1987, 1991, 1992, 1993, 1994, and 1996. Bear markets ensued in five of these cases. In 1965 there was an intermediate decline, the worst sell-off in three years. After a rally of several months, stocks finally buckled and went into a moderate bear market in 1966. In the 1983 case, stocks peaked in midyear, and entered into an intermediate decline that lopped 200 points off the Dow Industrials in less than one year. In 1986 stocks had a moderate correction before rebounding sharply.
As a rule, when monetary conditions are very favorable, I would not get too nervous about the number of secondaries until it rises to approximately 30 per month on a three-month average. But when monetary conditions are bearish, even a figure of 15 or so would be a poor sign.
In sum, the number of secondaries is an excellent barometer of excessive speculation near tops when the figure gets quite high. The indicator is best used for determining such tops. However, when the number of secondary offerings dips to extremely low numbers, it’s a sign of lack of speculative enthusiasm in the market, which implies pessimism. And that is often the harbinger of a market bottom.
ZWEIG’S SENTIMENT INDEX
You would not be able to calculate my Sentiment Index on your own. But I will present it here so that you can see how you might combine numerous measures of crowd psychology into a workable indicator. You could construct your own index, if you wish to take the time, by using even the four simpler indicators discussed earlier in this chapter. You can, of course, add more indicators on your own.
I regularly maintain a list of approximately thirty sentiment indicators, several of them overlapping to form one indicator out of three or four components. Included in this index is the mutual funds’ cash/assets ratio, the advisory sentiment, the number of bullish ads in Barron’s, and the number of secondary distributions, all of which I’ve covered in this chapter. Others that I monitor include the puts/calls ratio, which I invented and first wrote about in Barron’s in 1970–71; half a dozen different measures of short-selling activity; odd-lot buying and selling; insider trading; margin debt trends; initial public offerings; and speculative volume on the AMEX and OTC markets.
I grade most of these indicators on a scale where +2 points is extremely bullish, +1 moderately bullish, 0 is neutral, -1 is moderately bearish, and -2 is extremely bearish. On some of the lesser indicators the scale would only range from +1 to -1. For a few, such as the advisory sentiment, the scale can range from +3 to -3 because these indicators are more telling than the rest. I then convert the ratings to an aggregate reading where 100 is dead neutral on my Sentiment Index.
Theoretically, the Sentiment Index can range from +200 at the bull extreme—which would imply that every single component is extremely bullish—down to the bearish extreme of zero—in which every component would be extremely bearish. There has never been a reading at the extremes. The bullish record was 183 at the bear market bottom in mid-1970. The bearish extreme was a score of 26 in the spring of 1976, when the Dow was a couple of points away from its high. After that it trended sideways for several months and then finally eased into a bear market, which took the industrials down about 250 points before bottoming in February 1978.
Graph O (pp. 152–153) shows my Sentiment Index back to its origin in 1965. Scores above 140 show excessive pessimism and are rated extremely bullish. Scores between 120 and 139 are bullish. Readings between 100 and 119 are neutral to slightly bullish. When the Sentiment Index ranges between 76 and 99 the implication is moderately bearish. Finally, when my Sentiments Index is 75 or less it signifies too much optimism and the interpretation is extremely bearish.
GRAPH O
Ned Davis Research
Of course, as I’ve indicated before, one should use sentiment numbers in conjunction with the monetary background. Lower numbers than normal are needed for tops when monetary conditions are good, but only moderately low numbers might mark peaks in prices when monetary conditions are poor. Conversely, extraordinarily high scores such as those in 1966, 1970, and 1974 are required to pinpoint bottoms in times when monetary conditions are unfavorable. Readings in the moderately bullish range between 120 and 139 can be sufficient to herald excellent buying opportunities when monetary conditions are favorable, such as the intermediate bottom in the fall of 1975 or the buying juncture at the tail end of 1984, prior to a very spirited rally in January 1985.
The main thing to remember about measuring sentiment is to use several measures and not to value it too much when the numbers are relatively neutral. But when so many of your indicators show excessive pessimism that your index rises toward a high extreme, it’s probably a pretty good sign that the pessimism is overdone and that prices are near a bottom. Likewise, when too many folks are optimists and most of your readings indicate this, it’s time to start thinking about selling stocks. It’s quite useful to know when the crowd is extremely one-sided in its opinion. It is not as helpful to know that 55% are bulls, 45% are bears. The extremes are what really matter.
CHAPTER 9
Seasonal Indicators—A Year-Round Forecasting Guide
I suppose I was destined to be interested in the intriguing seasonal tendencies of the stock market. My youngest son was born on an Easter weekend and my eldest on Memorial Day weekend. My birthday usually falls on the July 4th weekend and my wife’s near Labor Day. Moreover, my mother found out she was pregnant with me on December 7, 1941—Pearl Harbor Day. To be sure, December 7 is not exactly a holiday, but my personal history on that day was just one more reason for me to be interested in the calendar and, more precisely, how it might affect stock prices.
I’ll cover six types of calendar tendencies in this chapter, starting with the most interesting, the market’s action around holidays. Later we’ll cover days of the week, months, month-end tendencies, the presidential cycle, and finally, the effects of year-end tax selling.
If the market were a truly unemotional mechanism, there would be no reason to expect any aberrant behavior around holidays, except conceivably late in the year, around Christmas and New Year’s, when transactions made for taxes—truly an economic purpose—might have some effect. But there is no economic reason to account for abnormal trading patterns around the other holidays. I have gone back and inspected the market’s action around the holiday periods back to 1952, giving us thirty-three to thirty-four observations per holiday. If trading had been normal at these times, the number of days that the market was up would be slightly more than one half, the long-run average for the market. What I found, however, is anything but normal. Rather, price trends near holidays are extraordinarily bullish, shattering the myth espoused by many academicians that stock price movements are random. Moreover, it’s nearly certain that these patterns are attributable to the emotionalism of investors.
There are seven major holidays during the year in which the stock market is closed, and two others that can be dismissed. We’ll throw out Election Day right off. First, a national election day occurs only once every two years, and in those years when only the congressional elections were at stake, the stock market remained open. The exchanges used to be closed only for presidential elections, but even that tradition was ended in 1984. So there is no longer an election holiday as far as stocks are concerned. The borderline case is the so-called presidential holiday in February, a sort of combination of both Washington’s and Lincoln’s birthdays celebrated on a Monday. Years ago, the exchange would sometimes close for both presidents’ holidays, other years only for one. Beginning in 1969 the new holiday format was adopted.
Actually, the market has done very well on the day prior to President’s Day. Of the seventeen occasions of this “newer” holiday, the Zweig Unweighted Price Index has risen 12 times prior to the holiday, decli
ned 3, and was unchanged twice. Ignoring the unchanged days, that’s an 80% success rate. The ZUPI gained an average of .17% per holiday period, for an annualized gain of 28.3%. That’s way above random, although, as we’ll see, it’s not up to snuff vis-a-vis other holidays. Moreover, since this holiday’s history is relatively brief, I’ve decided to ignore it in the forthcoming discussion.
We’ll focus on the seven remaining holidays: Easter, Memorial Day, July 4th, Labor Day, Thanksgiving, Christmas, and New Year’s. In all cases, I measured the market’s activity around the holiday season using my Zweig Unweighted Price Index. The most startling observation is that on the last trading day prior to the holiday, the market had an exceptional tendency to rise, no matter which holiday was involved.
Table 27 shows the price action on the day preceding each of the seven holiday periods. For example, in the 34 observations prior to Easter, the market rose 26 times, fell only 5, and was unchanged 3 times. The ZUPI gained an average of .26 of a percent per day, an annualized pace of 68% a year. As seen in the middle column, had you invested $10,000 on only the one day preceding the Easter holiday over the past thirty-four years, the money would have grown to $10,906. Similar data follow for each of the other holidays.
The holidays with the most bullish tendencies are Labor Day and New Year’s. Labor Day produced the best percentage gains, with the ZUPI up .64 of a percent for the day, an annualized rate of a whopping 180.4%. The market was up 31 times and down only twice on the day before Labor Day. New Year’s had the best percentage of winning days, with the market up 31 times, down only once, and even once. The ZUPI rose .53% on the day, or a per annum rate of 146.3%.