SYSTEM DESIGN
On System Development Part 2
by Mark Vakkur, M.D.
Last month, this author introduced the initial steps that someone, novice or veteran, should take when developing a stock market system. This time, he explains the final steps.
Last time, I posited that successful trading consists of systematically putting the odds in your favor. Since our only guide to the future is the past, I showed how the trader must find historical relationships and apply them to the markets going forward. I outlined four steps. First, historical data must be gathered; I used monthly closing data of the Standard & Poor's 500 to illustrate. Second, the data must be explored to discover relationships between variables. The third step is to explore those relationships further by setting up a spreadsheet (or other comparable software). The system developed in the article for illustrative purposes began with seasonality since the stock market, as measured by the S&P 500, makes most of its gains during certain months, especially the November?January period, the March?April period, and July. The worst month by average return from 1950 to 1995 is September, which is also the only month with a negative average return. Ignoring dividends, I illustrated this effect with the following simple systems:
These strong correlations can be exploited via simple systems. Investing only during the preelection and election years and then moving to cash for the other two years of the quadrennial cycle would result in a total return (ignoring dividends) of $733,600 from our hypothetical $10,000 invested in 1950. Note this is superior to the November?April strategy and is simpler, requiring only one round-trip trade every two years (the capital-gains consequences of this strategy would also be superior, but the effect of taxes are ignored in this article). On the other hand, investing only in the postelection and midterm years would have turned $10,000 into $56,300, much worse than buy and hold.
- Buy and hold would have compounded $10,000 invested in 1950 to $372,300. Investing for only six months of the year, from November 1 through April 30 and then moving to cash for the rest of the year, would have turned $10,000 into $703,900, almost twice as much as buy and hold, with only half the risk! Doing the opposite (investing in the S&P 500 in May through October, then moving to cash for November to April) would have only increased the $10,000 to $58,700. The simplest strategy -- that of avoiding September by moving to cash during this single month -- would have led to a return of $624,100.
- The next variable was the Presidential election year cycle. This cycle is of interest because the market tends to make most of its advances in the preelection and election years, which have returned an average of about 18% and 10%, respectively, since World War II. In the period under consideration, there has never been a down preelection year. Conversely, the postelection and midterm years are the worst times to be invested in the stock market; 1990 and 1994, the most recent midterm years, were both poor years for the stock market.
- Tracking interest rates is the final variable. This variable is the relationship between the trend in the yield on the 30-year Treasury bond as measured by the difference between the yield and its six-month moving average (positive values mean the yield is rising; negative values indicate falling interest rates). From 1950 to 1995, the monthly change in the S&P 500 averaged 0.74%. However, when the 30-year yield was below its moving average, the average monthly return on the S&P 500 was 1.56%, more than twice as much; note this is 20.4% annualized. Whenever the yield exceeded its moving average by more than 0.1%, the average monthly return of the S&P 500 was -0.70%. Neutral values (between zero and 0.10% difference) fell somewhere in between, but nonetheless were very strong: 1.43% on average.
1/1950-11/1997
Optimized monthly system
Buy & hold
Difference
$10,000 becomes
$3,828,167
$577,630
$3,250,537
Average
13.5%
9.0%
4.5%
Standard deviation
18.8%
14.6%
4.3%
Risk-adjusted
11.7%
8.0%
3.8%
Max gain
69.3%
57.0%
12.3%
Max loss
-46.3%
-41.3%
-5.0% FIGURE 1: COMPARISON. The preferred method begins with an average annualized rate of return, compared with buy and hold. This alone, of course, tells you nothing about the stability of your system. To do this, the system's standard deviation must be determined. A system that swings wildly from fantastic gains to horrific losses will have a huge standard deviation.
Mark Vakkur is a psychiatrist and a stock trader. Excerpted from an article originally published in the July 1998 issue of Technical Analysis of STOCKS & COMMODITIES magazine. All rights reserved. © Copyright 1998, Technical Analysis, Inc.