by Martin Zweig
Table 18 lists all the signals from Davis’s test of the Four Percent Model beginning May 1966. In this table we have assumed that you sell short on the sell signals (a comprehensive discussion of short selling appears in chapter 14) and go long on the buy signals. I don’t necessarily advise that you do that, but I want to show what the profits would have been had you sold the market short on the sell signals. One could just as easily assume that a luckless investor had always bought on the sell signals. The results of those sells then would have shown considerable losses. But of course that would be doing just the opposite of what the wise speculator would do; namely, follow the trend.
TABLE 18
FOUR PERCENT MODEL VS. VALUE LINE COMPOSITE INDEX: May 6,1966 to March 20, 1996
Signal
Date
Value Line Index
Profit (%)
Days
$10,000 Growth
SELL 5/06/66
133.09
15.9
168 11,591
BUY 10/21/66
111.92
33.6
371 15,488
SELL 10/27/67
149.55
-2.2
63 15,142
BUY 12/29/67
152.89
-4.5
42 14,463
SELL 2/09/68
146.04
.3
56 14,510
BUY 4/05/68
145.57
12.3
112 16,300
SELL 7/26/68
163.53
-2.9
42 15,831
BUY 9/06/68
168.24
5.2
126 16,653
SELL 1/10/69
176.98
3.1
119 17,172
BUY 5/09/69
171.46
-8.0
35 15,798
SELL 6/13/69
157.74
8.3
126 17,114
BUY 10/17/69
144.60
-3.2
35 16,564
SELL 11/21/69
139.95
8.8
105 18,022
BUY 3/06/70
127.63
-4.6
14 17,192
SELL 3/20/70
121.75
21.7
70 20,918
BUY 5/29/70
95.36
-6.9
28 19,475
SELL 6/26/70
88.78
-1.9
21 19,100
BUY 7/17/70
90.49
-3.7
28 18,458
SELL 8/14/70
87.45
-9.3
14 16,742
BUY 8/28/70
95.58
2.0
56 17,082
SELL 10/23/70
97.52
-1.4
42 16,837
BUY 12/04/70
98.92
19.6
175 20,130
SELL 5/28/71
118.27
3.5
84 20,843
BUY 8/20/71
114.08
-2.2
56 20,381
SELL 10/15/71
111.55
5.1
49 21,410
BUY 12/03/71
105.92
13.2
154 24,236
SELL 5/05/72
119.90
6.2
189 25,741
BUY 11/10/72
112.45
.6
42 25,899
SELL 12/22/72
113.14
25.3
203 32,460
BUY 7/13/73
84.48
1.6
28 32,986
SELL 8/10/73
85.85
-1.2
28 32,595
BUY 9/07/73
86.87
3.6
56 33,754
SELL 11/02/73
89.96
12.1
63 37,821
BUY 1/04/74
79.12
-.9
84 37,501
SELL 3/29/74
78.45
9.1
70 40,914
BUY 6/07/74
71.31
-6.5
14 38,263
SELL 6/21/74
66.69
20.2
91 46,003
BUY 9/20/74
53.20
-6.3
14 43,115
SELL 10/04/74
49.86
-11.8
7 38,039
BUY 10/11/74
55.73
-5.5
42 35,957
SELL 11/22/74
52.68
1.1
42 36,340
BUY 1/03/75
52.12
45.2
203 52,766
SELL 7/25/75
75.68
5.8
112 55,813
BUY 11/14/75
71.31
-5.0
21 53,019
SELL 12/05/75
67.74
-5.7
28 49,982
BUY 1/02/76
71.62
19.7
98 59,808
SELL 4/09/76
85.70
-1.4
77 58,992
BUY 6/25/76
86.87
-1.3
56 58,245
SELL 8/20/76
85.77
-2.8
35 56,629
BUY 9/24/76
88.15
-4.1
14 54,309
SELL 10/08/76
84.54
-2.3
49 53,044
BUY 11/26/76
86.51
4.0
133 55,159
SELL 4/08/77
89.96
-5.5
77 52,130
BUY 6/24/77
94.90
-3.1
56 50,493
SELL 8/19/77
91.92
-.5
84 50,263
BUY 11/11/77
92.34
-1.5
56 49,517
SELL 1/06/78
90.97
-3.4
70 47,813
BUY 3/17/78
94.10
10.6
105 52,899
SELL 6/30/78
104.11
-4.5
28 50,496
BUY 7/28/78
108.84
4.4
56 52,718
SELL 9/22/78
113.63
11.8
77 58,944
BUY 12/08/78
100.21
14.9
308 67,738
SELL 10/12/79
115.16
1.3
42 68,626
BUY 11/23/79
113.65
9.6
98 75,178
SELL 2/29/80
124.50
11.5
42 83,801
BUY 4/11/80
110.22
29.3
210 108,321
SELL 11/07/80
142.47
-4.1
7 103,926
BUY 11/14/80
148.25
-7.3
28 96,313
SELL 12/12/80
137.39
-5.0
14 91,483
BUY 12/26/80
144.28
4.9
196 96,004
SELL 7/10/81
151.41
12.3
84 107,810
BUY 10/02/81
132.79
-.8
105 107,007
SELL 1/15/82
/>
131.80
3.1
77 110,278
BUY 4/02/82
127.77
-2.3
56 107,784
SELL 5/28/82
124.88
2.9
84 110,857
BUY 8/20/82
121.32
64.3
343 182,184
SELL 7/29/83
199.38
-1.6
56 179,279
BUY 9/23/83
202.56
-4.2
28 171,791
SELL 10/21/83
194.10
-1.7
35 168,959
BUY 11/25/83
197.30
-4.4
70 161,568
SELL 2/03/84
188.67
6.0
182 171,305
BUY 8/03/84
177.30
-.9
119 169,837
SELL 11/30/84
175.78
-3.0
42 164,716
BUY 1/11/85
181.08
5.6
112 173,867
SELL 5/03/85
191.14
-4.1
35 166,708
BUY 6/07/85
199.01
-0.6
70 165,628
SELL 8/16/85
197.72
-1.0
84 163,894
BUY 11/08/85
199.79
18.5
245 194,238
SELL 7/11/86
236.78
0.7
42 195,657
BUY 8/22/86
235.05
-6.5
21 183,029
SELL 9/12/86
219.88
-4.5
49 174,747
BUY 10/31/86
229.83
13.5
167 198,264
SELL 4/16/87
260.76
-2.7
57 192,850
BUY 6/12/87
267.88
2.9
98 198,386
SELL 9/18/87
275.57
26.7
91 251,386
BUY 12/18/87
201.95
10.1
154 276,830
SELL 5/20/88
222.39
-5.9
21 260,424
BUY 6/10/88
235.57
-2.4
63 254,111
SELL 8/12/88
229.86
-3.0
56 246,406
BUY 10/07/88
236.83
-4.1
35 236,345
SELL 11/11/88
227.16
-3.9
56 224,856
BUY 1/6/89
236.07
11.8
280 251,389
SELL 10/13/89
263.93
6.7
154 268,232
BUY 3/16/90
246.23
-5.8
42 252,675
SELL 4/27/90
232.04
-4.1
14 242,315
BUY 5/11/90
241.59
-1 2
77 239,407
SELL 7/27/90
238.64
19.7
126 286,570
BUY 11/30/90
191.55
22.3
210 350,475
SELL 6/28/91
234.36
-4.2
56 335,756
BUY 8/23/91
244.21
-4.1
91 321,990
SELL 11/22/91
234.26
-3.3
35 311,364
BUY 12/27/91
242.09
4.5
98 325,375
SELL 4/3/92
253.00
-0.1
119 325,050
BUY 7/31/92
253.36
-4.4
63 310,748
SELL 10/2/92
242.25
-2.9
28 301,736
BUY 10/30/92
249.27
14.2
518 373,873
SELL 4/1/94
284.85
0.0
133 373,873
BUY 8/12/94
284.68
-1.2
91 369,386
SELL 11/11/94
281.14
1.3
49 374,188
BUY 12/30/94
277.52
13.8
301 425,826
SELL 10/27/95
316.04
-4.0
35 408,793
BUY * 12/1/95
329.31
4.2
133 425,962
The fourth column in the table shows the percentage of profit on all of the signals, both the sells and the buys. Also shown are the calendar days during which each signal was open and, in the right-hand column, the cumulative value of an initial $10,000 portfolio. The results here are theoretical because no one could have actually bought and sold the Value Line Index over this period. Stock index futures began to trade on the Value Line Index in 1982. Since then, one could have closely approximated the returns on the actual Value Line Index. Over the span since 1966, you could have had some approximation of the Value Line Index by buying diversified mutual funds that had broad-based portfolios, or by buying a diversified portfolio of stock heavily weighted toward medium-and smaller-sized companies.
GRAPH I
Ned Davis Research
Graph I (pp. 98–99) shows the buy and sell signals on the Four Percent Model plotted against the Value Line Index back to 1978. The B’s on the graph show the buy signals while the S’s show the sell signals.
Table 19 sums up the results of the Four Percent Model. There were 61 buy signals. Of these 61 buys, only 30 were profitable, just 49%. However, those 30 profitable buys produced average profits of 14.1% per trade. Conversely, the 30 losing trades lost only 3.5% per trade. This is a perfect example of cutting your losses short and letting your profits run, the ideal strategy for the speculator … and not such a bad idea for the traditional investor either.
Taking the 61 buy signals together, they produced an average gain of 4.7% per trade. The buy signals were in effect for 80 calendar days on average, or something more than eleven weeks per trade. That’s not really too many trades. It’s within reason as far as commissions and portfolio turnover are concerned. When that 5.0% profit per trade is annualized, it works out to 16.2%. From 1966 to March 20, 1996, had you merely bought and held the Value Line Index, you would have made only 2.7% per year. These calculations ignore dividends both on buy-and-hold and on trading. Obviously, the addition of dividends would add to the return in both cases.
The results on the sell side are similar. Assuming that you had sold short on the sell signals, you would have made money 28 times in 61 trades, a success rate of 46%. That may not sound like much, but on those 28 successful trades on the short side, the average gain was 9.6%. (Alternatively, had you insisted on buying in those 28 cases, you would have lost an average of 9.6%). By contrast, of the 33 cases in which the short-selling speculator would have been wrong, his average loss would have been only 3.5%. It nets out to an average gain per trade on the short side of 2.4%, with an average holding period of 48 days, or 7 weeks. The annualized profit on the short side was 13.5%. That means that the investor who kept on buying during the sell signals would have lost 13.5% per year in those spans.
The third section of table 19 combines all trades, irrespective of whether they were buys or sells. Fifty-two percent of those trades lost money, but the average loss was only 3.7%. The 48 percent that made money showed average gains of 11.8%. The average gain per trade over 122 total trades was 4.1%. That equal
s an annualized return from trading with the Four Percent Model of 13.3%, far in excess of the 2.7% return that one could have garnered from merely buying and holding over 30 years. Obviously, this very simple model works well.
Occasionally, the losses are somewhat more than I would prefer, mainly because of the drawback of using only weekly closing prices. However, there’s a plus side, primarily its simplicity and the fact that you need not hover over a Quotron machine daily, worrying about whether the model will flip or not. Sometimes, by examining too many trees, one loses sight of the forest—not a good idea. Had you traded both the long and the short side since 1966, an initial $10,000 would have grown to $425,962, exclusive of dividends. That’s not bad for an extremely simple model.
You can use this model just as presented. Or you can alter it to your own liking. There’s no law that says you have to wait for a 4% change. For example, if you want fewer trades and fewer signals, you can increase the 4% rule to, say 5% or 6%. You’ll probably have a slightly lower return on a gross basis, but you’ll save something in transaction costs and avoid some of the signals. Conversely, if you are more short-term-oriented, you might cut the rule to, say, 3% or even 2.5%, and have more trades, probably a higher gross return, but greater transaction costs. I feel the 4% rule is a nice trade-off between excessive turnover on the one hand and solid returns on the other.
You could also apply a similar rule to some of the other major averages such as the S&P 500, but it won’t work as well as it does on the Value Line or on the Zweig Unweighted Price Index. That’s because the major averages are not as volatile as the Value Line or ZUPI and generally don’t go up as much in bull markets, nor fall as much in bear markets.
Let’s sum up how the Four Percent Model works. All you need is the weekly close on the Value Line Index. If the index rises by 4% or more, it triggers a buy signal. If it drops by 4% or more, it’s a sell signal. About half the signals will be unprofitable. However, the profits on the good signals overwhelm the losses on the poor signals. As a result, you’ll make solid profits in the long run by staying in gear with the trend.
TABLE 19
SUMMARY OF FOUR PERCENT MODEL VS. VALUE LINE COMPOSITE INDEX: May 6,1966 to March 20, 1996
Type of Trade
Profit per Trade
Number of Trades
Average Days per Trade
Annualized Profit
Buys (long)
Losses 31 -3.9% (51%)
Gains 30 +14.1% (49%)
Net 61 +4.7% (100%) 80 +16.2%
Sells (short)
Losses 33 -3.5% (54%)
Gains 28 +9.6% (46%)
Net 61 +2.4% (100%) 48 +13.5%
Total
Losses 64 -3.7% (52%)
Gains 58 +11.8% (48%)
Net 122 +4.1% (100%) 64 +13.3%
Results of all trades:
$10,000 became $425,962 in 30 years (+13.3% annualized return).
Annualized return for buy-and-hold = +3.2% ($10,000 became $25,727).
CHAPTER 6
Combining Monetary and Momentum Indicators—The Only Investment Model You Will Ever Need
In chapter 4 we developed the Monetary Model using interest rate and Federal Reserve indicators to forecast the market. Its excellent results verify the rule “Don’t fight the Fed.” In chapter 5 we developed three momentum-type indicators, one of which, the Four Percent Model, gives continuous bullish or bearish signals. Its results are also excellent and honor the theory “Don’t fight the tape.” Given those solid results, it would be reasonable to combine both monetary and momentum indicators to obtain a superior model, one that will follow both the Fed and the trend of the tape. In this chapter we will develop such a model.