Relative Strength and the TSE
Relative Strength has recently gained popularity due to its success as a basis for speculation. Relative strength itself is simply a measure of how well a particular security has performed relative to other members of its peer group or to the market as a whole. The main focus of this article is on the performance of relative strength approaches as applied to the TSE 35, but other markets will also be discussed. Relative strength as applied to Canadian mutual funds is also investigated. Finally, some criticisms are leveled against relative strength approaches in general.
Relative strength is usually determined by comparing a stock's one-year total return to its peer group. This process requires the definition of a peer group and then the sorting of the members of the group from high to low one-year total returns. The usual relative strength approach has the investor buy the top relative strength candidates (i.e. those with the highest prior returns), hold them for a year and then repeat the process. James O'Shaughnessy provides comprehensive one-year relative strength results for U.S. stocks in his valuable book What Works on Wall Street (See Table 1). He shows that this approach has performed well when applied to Large Stocks (typically the top 16% of stocks with capitalizations in excess of $150 million U.S.). The results are not quite as good when small stocks are included. Nonetheless, the simple one-year relative strength approach enjoys a 1.1% to 2.7% yearly advantage in U.S. markets and as a result has enjoyed a strong following.
Canadians, particularly those with self directed RRSPs, will no doubt wonder if the success of relative strength carries over from U.S. to Canadian markets. Steve Foerster has written on this subject in the Canadian Moneysaver (See September 1998, Momentum Strategies: Catching a Wave). His article describes a comprehensive 5-year study based on the TSE 100 which determined that a strategy using relative strength proved successful with an average annual gain of 27% vs. 14.5% for the index itself. In Steve Foerster's study relative strength was determined by averaging the previous four quarterly price changes with a double weight given to the most recent quarter's results. Furthermore, the portfolio was rebalanced each quarter instead of each year unlike the other approaches mentioned in this article. The success of this approach has spawned a significant following that includes a fan website at tserelativestrength.webjump.com.
The results of applying relative strength to the TSE 35 are also encouraging (See Table 2). The method used to determine these results is straight forward and similar to that used by O'Shaughnessy. On December 31 the top performers of the TSE 35 from the previous year are purchased, held for a year and then the process is repeated. Table 2 shows the results of this approach for portfolios composed of between one and ten stocks. For instance, the column labeled 6 provides the results for portfolios formed by selecting the top 6 relative strength stocks each year. For each portfolio the average, minimum and maximum annual return is shown as well as the growth multiple of the portfolio over the December 31 1988 to November 1 1999 time period. For comparative purposes the last column contains the results for the TSE 35 total return index over the same period of time. It should be noted that 1999 has been a phenomenal year for this approach with the help of Nortel Networks which has gained 116.12% to November 1. If this unusual year is not included the approach still beats the TSE 35 total return index although not by quite as much.
There has been some interest in trying to obtain advantage from a combined relative strength and value ratio approach (See The Value Ratio Approach, Canadian Moneysaver October 1999). Table 3 shows the results of first selecting the top 10 Value Ratio stocks in the TSE 35 (i.e. stocks with the lowest but positive P/E-to-Yield ratios) each year and then choosing stocks from this list that have had the best one-year total returns. This combined approach proves to have been toxic to returns in nearly all cases. The exception being the one stock portfolio which maintains rough parity with the regular value ratio method. As a result, it would appear that the value ratio and relative strength approaches are poor bedfellows.
It looks like buying last year's winners works for stocks. Does the same hold true for mutual funds? To investigate the effect of relative strength on mutual fund performance I used the FundMonitor.com database. This database splits all mutual funds available to Canadians up into over 50 subtypes (i.e. Large Cap US Equity, Canadian Bond etc). For each subtype the funds are ranked according to their one-year total returns and then split into quartile groups. Quartile one (Q1) is composed of the top 25% of the list while quartile four (Q4) contains the bottom 25% of the list (Q2 & Q3 fall in between). The average performance of each quartile is then tracked over the next 24 months. The results for each subtype are subsequently averaged and the conclusions shown pictorially in Figure 1. The line through the center of the figure represents the average performance of all Canadian funds. The relative performance of the quartile groupings is displayed with over performing quartiles above the center line and under performing quartiles below the center line. As can be seen, the funds that performed the worst in the previous year subsequently do better than the top performers. Furthermore, the subsequent relative over performance of the laggards is quite large in the first few months but trails off to near zero after about 24 months. In absolute terms, the one to twelve month annualized performance advantage of the Q4's vs. the Q1's was found to vary between 1.5% and 0.6%. It should be noted that this work is subject to a 'survivorship bias'; the list of funds used does not include those that have disappeared through failures or other means. A future article will report on an empirical determination of the disappearance rate which for Canadian funds is approximately 10%/year. With such a large loss rate it is possible that the results shown here are skewed. For instance, if Q4 contains more funds that disappear than Q1 the over performance of the Q4s shown in Figure 1 may disappear.
Despite the strong performance of various relative strength approaches they are not, in my opinion, suitable for investors. One of the biggest problems with them is that they are not based on the fundamentals of companies but only on price action. Relative strength approaches will happily put the investor into the BRE-Xs of the world based only on the premise that there will be a greater fool who will come along to buy the shares at a higher price. Such an approach may be profitable in bull runs but is highly suspect in down times. Furthermore, since the method is based on price, there is little to help the investor to 'stick to his guns' if the market turns down, even temporarily. As a result, investors may be discouraged by temporary market fluctuations and end up buying high and selling low. Additionally, one has to wonder why one-year relative strength used instead of, say, six-month relative strength. This question was touched on by Robert Haugen in his book The Inefficient Stock Market and some of his results are summarized in Table 4. Here negative figures represent subsequent under performance and positive numbers over performance. It is apparent that one must select the relative strength period well. Furthermore, out performance of certain relative strength time periods changes over the long haul. It is my fear that such natural variations may extinguish the strong performance of the one-year relative strength approaches described in this article. Finally, relative strength methods usually encourage frequent trading which is itself a strong drag on performance and puts the investor in a more speculative frame of mind. All of these factors and my own psychological predisposition makes me very leery of recommending relative strength approaches.
First published in January 2000.
|Disclaimers: Consult with a qualified investment adviser before trading. Past performance is a poor indicator of future performance. The information on this site, and in its related newsletters, is not intended to be, nor does it constitute, financial advice or recommendations. The information on this site is in no way guaranteed for completeness, accuracy or in any other way. More...