by Martin Zweig
TABLE 14
TEN-DAY ADVANCE/DECLINE RATIO GREATER THAN 2-T0-1:1953 to 1996
Percentage Change In Market
3 Months Later 6 Months Later
Date
S&P 500
ZUPI
S&P 500
ZUPI
1/26/54
+7.1
+5.2
+16.6
+15.2
1/24/58
+3.4
+4.4
+11.8
+15.4
7/11/62
-1.2
-1.3
+12.3
+11.3
1/16/67
+7.3
+11.0
+10.0
+20.7
12/4/70
+9.5
+19.8
+13.2
+21.8
1/10/75
+15.4
+18.2
+30.6
+38.9
1/6/76
+10.5
+19.4
+10.7
+19.0
8/23/82
+14.5
+24.2
+26.4
+38.4
1/23/85
+2.6
+2.9
+8.6
+11.0
1/14/87
+6.3
+3.1
+18.3
+9.6
2/5/91
+7.8
+13.6
+11.2
+13.9
$10,000 becomes: $22,068
$30,451
$47,591
$68,953
Return/Period = +7.5%/qt.
+10.6%/qt.
+15.2%/half
+19.2%/half
In other words, had you patiently waited for prices to “explode” over some two-week period, and then stepped in and bought at what seemed like “high” levels at the time, you would have made abnormally large profits in the months that followed. Thus, strong momentum tends to persist. Momentum also tends to be greatest at the beginnings of bull markets. Indeed, seven of the listings in table 14 came near the starts of bull markets. Others were “second legs” of old bull markets.
As indicated in the results shown in table 14, the market must push off with a lot of firepower to get off the ground. To put it succinctly, if the tape can’t ignite, conditions aren’t right.
UP VOLUME INDICATOR
A second momentum indicator uses the ratio of up volume to down volume. Up volume comprises the total volume of all stocks that rise on a given day, and down volume totals all that decline. Again, we’ll ignore the volume for stocks that are unchanged on the day. The up and down volume figures are available daily in The Wall Street Journal, on quote machines, and in the Barron’s weekly statistical section, as well as in the financial sections of most major newspapers.
I have found that when 90% or more of the volume (ignoring unchanged volume) is on the upside in a given day, it is a significant sign of positive momentum. In other words, when daily up volume leads down volume by a ratio of 9-to-1 or more, that tends to be an important signal for stocks. The significance of this ratio was first spotted many years ago by Lowry’s Reports, Inc. However, over the years I’ve developed my own ways of interpreting the up/down volume.
Since 1960, there have been only 124 days in which the 9-to-1 ratio was on the down side—that is, where down volume exceeded up volume by more than 9-to-1. That averages to fewer than four such very poor days per year. Conversely, there have been only sixty-eight very positive days in which up volume topped down volume by at least 9-to-1. Thus there have been only about two 9-to-1 up days per year.
The 9-to-1 down days do have some predictive ability but are not nearly as strong an indicator as the 9-to-1 up days, so let’s just concentrate on the latter. Why is a 9-to-1 up day significant? Simply because it graphically shows the powerful thrust of the market. Every bull market in history, and many good intermediate advances, have been launched with a buying stampede that included one or more 9-to-1 up days. *
The most spectacular one-day reading ever was 42-to-1 on August 17, 1982. Not so coincidentally, it heralded the strongest bull market in nearly five decades. Just three days later there was a 32-to-1 up day, the second greatest in the thirty years for which I have figures. That merely verified the enormous momentum that helped to blast off that bull market.
But even a 9-to-1 up day can fail now and then. In other words, while a 9-to-1 up day is an impressive positive sign, it does not guarantee a great leap forward. More thrust is required to enhance the bullish odds.
However, the 9-to-1 up day is a most encouraging sign, and having two of them within a reasonably short span is very bullish. I call it a “double 9-to-1” when two such days occur within three months of one another. As an indicator, this calculation falls into one of two types. The first and most bullish is when there are no intervening 9-to-1 down days. The second is when one or more 9-to-1 down days occur between the up days. The latter condition implies not as much thrust as does the former, but here too the record provides great comfort to the bulls.
As you will see in table 15, there have been sixteen double 9-to-1 signals since 1960. In 1962 and 1975 there was a third 9-to-1 up day within several weeks of the first two. After the August 1982 double, there were two more 9-to-1 days in October (which I consider a repeat signal), an amazing display of thrust that was soon followed by another double 9-to-1, completed in early January 1983. In early August of 1984, there was an unprecedented string of three consecutive days in which up volume topped down volume by 9-to-1 or better, and about a week later there was a fourth such day.
Table 15 measures the Dow Industrials’ advances after the double 9-to-1 days. At the bottom of the third column you will see that three months later, on average, the Dow’s compounded return was up 7.0% per quarter. The next column shows that, six months after these signals, the Dow rose an average of 12.8%, and the final column shows a 17.2% gain one year later. The table reveals that the Dow was up every time but twice three months after these signals, and every single time but once six months and twelve months later.
TABLE 15
DOUBLE 9-T0-1 SIGNALS WITHIN THREE MONTHS VS. DOW JONES INDUSTRIAL AVERAGE: January 1,1960 to March 20,1996
Date
Dow
% Change 3 Mo. Later
% Change 6 Mo. Later
% Change 12 Mo. Later
11/12/62 624 +8.5
+15.9
+20.2
11/19/63 751 +6.9
+9.1
+16.5
10/12/66 778 +6.9
+8.6
+17.4
5/27/70 663 +14.6
+17.8
+16.5
11/29/71 830 +11.8
+17.0
+22.8
9/19/75 830 +1.7
+18.1
+19.9
11/1/78 828 +1.6
+3.3
-1.0
4/22/80 790 +17.3
+20.9
+27.5
3/22/82 820 -2.4
+13.2
+37.0
8/20/82 869 +15.1
+24.3 *
+38.4*
1/6/83 1071 +3.9
+14.0*
+20.2*
8/2/84 1166 +4.4
+10.6
+16.0
11/23/84 1220 +4.7
+6.3*
+20.0*
1/2/87 1927 +20.4
+26.4
+4.
10/29/87 1938 +1.0
+4.9
+8.4
6/8/88
2103
-1.9
+2.4
+19.8*
$10,000 = Compounded return $29,304
$62,737
$91,855
+7.0%/qtr
+12.8%/half
+17.2%/year
As usual, the results are more extreme when measured against the ZUPI, as seen in table 1
6. Three months after the double 9-to-1 signals, the broad market advanced 10.3%, rising every time but twice. Six months later the ZUPI gained an average of 17.1%, and a year later the unweighted average was up 21.2%.
Now observe the bottom line of the upper half of table 17, which compares the performance of the double 9-to-1 signals to that of an investor who merely bought the Dow Industrials and held them to the end of this study in 1983. Such an investor would have made only 4.9% a year in capital appreciation (plus something less than 4% a year in dividends, not shown on the table). Going across the top two lines of the table you’ll see a summary of what had been reported in table 15, namely, the three-, six-, and twelve-month returns on the Dow following the double 9-to-1 signals. The third and fourth lines show the Dow’s performance for all periods since 1960 not following the 9-to-1 signals. These results are virtually all negative. Obviously, the Dow did a heck of a lot better in periods after these double 9-to-1 signals.
TABLE 16
DOUBLE 9-T0-1 SIGNALS WITHIN THREE MONTHS VS. ZWEIG UNWEIGHTED PRICE INDEX (ZUPI): January 1,1960 to March 20, 1996
Date
ZUPI
% Change 3 Mo. Later
% Change 6 Mo. Later
% Change 12 Mo. Later
11/12/62
72.10
+12.8
+19.2
+22.3
11/19/63
86.95
+5.2
+7.8
+16.2
10/12/66
100.64
+16.0
+28.1
+47.5
5/27/70
89.44
+8.1
+12.4
+40.9
11/29/71
109.13
+18.3
+16.3
+9.6
9/19/75
65.16
+1.0
+28.7
+32.6
11/1/78
99.35
+5.6
+8.5
+3.9
4/22/80
100.76
+25.2
+35.9
+46.2
3/22/82
124.47
-2.0
+12.3
+48.3
8/20/82
126.35
+27.5
+40.7 *
+61.3*
1/6/83
169.91
+8.9
+24.6*
+23.8*
8/2/84
182.81
+7.8
+17.8
+27.1
11/23/84
195.96
+11.0
+13.5*
+21.4*
1/2/87
274.78
+16.0
+14.8
-15.4
10/29/87
219.95
+8.6
+16.1
+14.7*
6/8/88
260.10
-1.1
-2.3
+13.9*
$10,000 = Compounded return $47,684
$110,073
$147,045
+10.3/qtr.
+17.1%/half
+21.2%/year
TABLE 17
DOUBLE 9-TO-1 PERIODS VS. ALL OTHER PERIODS: January 1,1960 to March 20, 1996
3 Mo. Later
6 Mo. Later
12 Mo. Later
Vs. Dow:
9-to1 periods:
$10,000 = $29,304 $62,737 $91,855
Return = +7.0%/qtr. +12.8%/half +l7.2%/year
All other periods:
$10,000 = $16,804 $7,891 $5,361
Return = +.4%/qtr. -.5%/half -3.2%/year
Buy-and-hold:
$10,000 = $49,243 $49,243 $49,243
Return = +1.2%/qtr. +2.4%/half +4.9%/year
Vs. ZUPI:
9-to-1 periods:
$10,000 = $47,684 $110,073 $147,045
Return = +10.3%/qtr. +17.1%/half +21.2%/year
All other periods:
$10,000 = $9,478 $4,106 $2,012
Return = -.1%/qtr. -1.7%/half -8.0%/year
Buy-and-hold:
$10,000 = $45,427 $45,427 $45,427
Return =
+1.1%/qtr.
+2.3%/half
+4.7%/year
The second half of table 17 shows similar figures for the ZUPI. The first two lines there summarize table 16, showing the three-, six-, and twelve-month returns on the ZUPI following the signals. The next two lines show the results for all other periods. In all spans not falling within one year to 9-to-1 signals, the investor who bought the broad market as measured by the ZUPI would have lost 8.0% per year, a huge difference from the 21.2% a year he would have made had he listened to the signals. The last line shows the returns on buy-and-hold on the ZUPI, which worked out to 4.7% per year.
Or, if you prefer to think of dollars instead of percentages, the investor who bought the ZUPI and held for a year after double 9-to-1 signals would have seen a $10,000 investment appreciate to $147,045. That does not even count interest when out of the market or dividends while in stocks. By contrast, the unfortunate investor who insisted on buying only in the non-double 9-to-1 periods would have seen his $10,000 shrink to $2,012, an enormous disparity when compared with the dollars made by following the signals.
THE FOUR PERCENT MODEL INDICATOR
I’ve now given you two indicators that use momentum, one monitoring advances and declines and the other measuring up/down volume. These two indicators, as used here, look only for tremendous bursts of momentum that occur occasionally. They lead to terrific returns on the upside. However, these models are limited because, first, their buy signals are rare, and second, as designed, these indicators do not give sell signals. In other words, most of the time they don’t tell you much because the market does not explode very often. What we need is a model that always is on either a buy or a sell signal. Of course, such a model will not give the spectacular returns over shorter periods the way the first two indicators do. But they will tell the investor the safest course on a continuous basis. Remember, though, that no indicator or model is right all of the time. In fact, we’ll see that the model I’m about to develop is right only about half the time. But the profits derived from this model are excellent.
The Four Percent Model for the stock market was developed by my close friend and colleague Ned Davis. It works as follows. First, it uses the Value Line Composite Index, which is quoted regularly on Quotron machines and is found in your daily newspaper or in Barron’s. You’ll recall that the Value Line Index, as calculated by Arnold Bernhard & Co., is an unweighted price index of approximately seventeen hundred stocks and is very similar to my Zweig Unweighted Price Index. In fact, over long stretches the two give nearly identical results. One could just as easily apply this Four Percent Model to my ZUPI. However, the Value Line data are more readily available and easier to follow.
All you need to construct this model is the weekly close of the Value Line Composite. You can ignore the daily numbers if you wish. Just look in the Saturday or Sunday newspaper to find the weekly close, or in Barron’s, which hits the newsstands on Saturdays. This trend-following model gives a buy signal when the weekly Value Line Index rallies 4% or more from any weekly close. It then gives a sell signal when the weekly close of the Value Line Composite drops by 4% or more from any weekly peak. Note: I mean 4% of change, not four points.
For example, if in Week One the Value Line Index closes at 200, it would require a reading of at least 208 to generate a buy signal. Let’s say that happens in a week when the Value Line Index closes at 209. The buy signal continues as long as there is no 4% or greater drop in the weekly Value Line Index. Suppose the Value Line continues to rally, possibly with some small dips along the way—none of which is greater than 4%—until it reaches a closing weekly high of 240. At that point suppose the index begins to fall. It would have to drop by 4%, or to a level of 230.40
or less, to generate a sell signal. Assume that happens in a week when the Value Line closes at 229. The sell point would be 229, and the model would remain on a sell signal thereafter until there was a 4% or greater rally.
That’s all there is to it. In just about a minute a week, with the aid of a calculator (or if you can remember your long division), you can calculate the Four Percent Model.
This model is designed to force you to stay with the market trend. The market can’t rise for very long before you’re forced into a buy signal, nor can it fall very much before you’re forced into a sell signal. Of course, by using only weekly data, the market may move by more than 4% before you get a change in signals. Occasionally, the market may make a big move in a given week and you may be buying 6% or 7% or 8% above a weekly high, or perhaps selling that much below a weekly low; but nonetheless, you’re still in gear with the major trend.
The virtue of this Four Percent Model, or any trend-following model, is that if the market makes a very large move, you will be on the right side of the bulk of it. But there is no free lunch in the stock market. Although you are guaranteed of being on the right side of major moves, you may get whipsawed over very short-term movements. If the market were to zig and zag by moves only a little bit greater than 4%, you might be zagging when you should be zigging and zigging when you should be zagging. That will cost you some money, but the long-run results of the Four Percent Model clearly show that it is worth that cost.
