Report Testing and the Relative Strength Report
By David Vomund
Last month we discussed a mechanical trading technique that combines the Weighted Action List report with AIQ's market timing model. This technique has shown an average annual rate of return of 21% since 1992 and is already up 7.2% in the first quarter of this year. This month, we will apply the same mechanical test to different AIQ screening reports. The reports we will test are Persistence of Money Flow, Moving Average Crossover, and Relative Strength.
We tested a simple mechanical trading strategy. Using a database of the stocks that comprise today's S&P 500 index, we ran the AIQ reports every time there was an AIQ market timing buy signal (an Expert Rating of 95 or greater). Only the first signal in a string of buy signals was used and no confirmation was applied to the market timing signals. The top five stocks that appeared on each report were purchased and held until a market sell signal was registered, at which time all positions were moved to cash.
To obtain percentage return figures, we made the following assumptions:
The portfolio started with $25,000 with the buy signal on 09/10/92.
All profits/losses were reinvested in the next block of trading with each of the five stocks receiving an equal amount of capital.
The buy and sell points on the stocks were the opening price the day after the market timing Expert Rating.
Commissions were factored in using deep discount brokerage rates ($33 per stock trade).
Slippage, dividends, and money market interest received were not factored in.
Table 1 lists the yearly results of using each report as a means of selecting which stocks to buy. We see that over the testing time period, the S&P 500 gained an average of 12.9% per year.
The results for the Persistence of Money Flow report were poor, losing money in four of the five years. This was not surprising since it is consistent with our earlier testing of this report which was documented in the May, 1996 Opening Bell Monthly. This study revealed that Persistence of Money Flow does not perform well under short-term trading conditions. The market timing signals which governed the holding periods for the current study resulted in a fairly short-time average hold of 48 days. Persistence of Money Flow is most effective with a six month to one year holding period and is best applied to small company growth stocks rather than the S&P 500 type of stocks. In our May, 1996 study, the stocks selected from this report outperformed the S&P 500 by about 13% per year. We plan to update the Persistence of Money Flow study and report our findings next month.
The results were also disappointing for the Crossover of Two Moving Averages report. We had many requests for this report before we included it into the software package. Our default values for the moving averages of 21 days and 100 days was not as a result of statistical research. If you were one of those requesting this report, we encourage you to let us know which moving average values should be used.
The last report was Relative Strength - Strong. The long term section of the report was used. The results were spectacular. The average yearly gain was 32.8%. The worst year was a 20% gain in 1993. It even performed well in 1994's difficult trading year.
Live time results using the Relative Strength screening strategy would be different than our backtested results because the makeup of the S&P 500 index has changed over time. In our test, we are probably buying some stocks that were not part of the S&P 500 index at that time. As can be seen from the poor performance of the other screening reports, this is not enough of a factor to make all screening techniques attractive. It is clear that buying S&P 500 stocks with strong relative strength is a winning strategy.
Trade by trade details for the Relative Strength mechanical trading strategy are found in Table 2. This table lists the average return of the five stocks that are purchased for each trade along with the S&P 500's return. Again, the percent change figures are based on the opening price the day after the market timing Expert Rating. Whereas the S&P 500 gained an average of 2.7% during each market timing buy signal, the top five stocks on the Relative Strength - Strong report gained an average of 9%.
The largest portfolio loss was the 8.4% drawdown in early 1993. The drawdown in between signals at times were larger than this figure. For example, the July 9, 1996 signal turned into a 4.6% loss. At one point during that trade, however, the portfolio had lost 11.5%. Anyone applying this strategy should incorporate a stop strategy. It may be appropriate to move to the sidelines if the portfolio drops by 15%.
With impressive results on the long side, we wondered how the test would work for shorting stocks. A final test was run which used the Relative Strength - Weak report to short stocks. When we received a market timing sell signal, the top five stocks on the long term section of the Relative Strength report were shorted until a market timing buy signal was registered. The results are found in Table 3. Unfortunately, the results are not as favorable as was the case on the long side. The average return of the weakest S&P 500 stocks when AIQ was on a sell signal was about the same as the S&P 500 index. Picking S&P 500 stocks with poor relative strength is not an effective strategy for picking short candidates.
In our testing which combines AIQ's market timing model with various screening reports, we found very impressive results when using the Relative Strength - Strong report. The strategy of buying the strongest stocks when a market timing buy signal is registered and moving to cash when a market timing sell signal is registered has consistently outperformed a buy-and-hold strategy (see Figure 1). Keep in mind that this is a backtested strategy, however, and is only a five year test. We'll maintain a paper portfolio this year using the current S&P 500 structure at the time of the buy signal and will report the results in a future Opening Bell.