EL SEGUNDO, CA (Nov. 4, 1999) -- Last week on the message board, Elan Caspi (TMF Elan) posted a request for readers to voice their preferences for the strategies that the workshop will track next year. The results will be reported here next week.
Reading through the responses, several comments got my attention. First, someone suggested that the Spark5 be dropped as a strategy since all of the newer strategies had higher CAGRs. (For those unfamiliar with the Spark strategy, I will discuss it in detail in an upcoming article.) Then there were a number of suggestions for subsets of strategies, wild combinations of strategies, etc., that show returns of 45%, 50%, 55%. The general tone of these posts led me to wonder if too many people are overemphasizing the CAGRs and undervaluing everything else.
Yes, those amazing past returns (and thus the potential future returns) are the main reason I invest in Workshop strategies, and yes, I would never invest in a strategy without looking carefully at those numbers. In particular, I look at the year-by-year results and how a strategy performed during bad times in the market such as the crash of '87 and the Asian crisis a couple of years ago.
However, it doesn't matter whether the strategy returns 35%, 40%, or even 50%. OK, all things being equal, it matters, but at most it's a tiebreaker. Why doesn't it matter to me? There are two major reasons. First and foremost is that we are never going to repeat these returns. They are PAST returns. Second, the amount of data is so small that small changes such as dropping or adding one or two stocks could swing the returns some 10% to 15%. That leaves room for an awful lot of luck.
It's very easy to get caught up in the hoopla surrounding the unprecedented success of the stock market over the last decade and in particular the last four years. After such a decade, it is human nature to put on those rose-colored glasses and assume that this is how the market will act forever.
But it won't. It may continue this way for another year, another three years, maybe even another five years; unfortunately, I am not "wise" enough to know such things. However, you can rest assured that sometime within the next 10 to 25 years, things will change.
OK, you're not convinced. Let's look at this situation with our rose-colored glasses intact. Let's assume that in one of the greatest flukes of history, the years 2000 through 2013 have the exact same returns for the Dow, the S&P 500, and the Nasdaq as they did between 1986 and 1999. Who here thinks that each of our strategies would come back with the exact same returns? Not I! I would be willing to bet, if I were a gambler, that at least one of our great strategies won't even beat the S&P 500 over that period.
Ouch. And why is that? Let's think, what do our backtests really tell us? The backtests tells us that our stock screening methods worked from 1986 to 1999, nothing more, nothing less.
Yes, when I invest my money in the stock market, I feel a lot better investing in a method that has worked for the last 15 years. However, while I do sleep better, I am also aware that there is a very good chance that at least one of these strategies that look so rosy now won't work in the future. Read Todd Beaird's series on statistical significance if you have any doubts.
Even statistically significant screens are, to some extent, dependent on luck. For example, let's take a look at the PEG screen. From 1986-1999, an annual 5-stock PEG screen picked a total of 70 stocks. Except for 1998, it did an almost unbelievable job of picking stocks. But even if we assume that in the future the market will act the same and that the PEG method will prove to be an outstanding method of picking stocks, there will still be an element of luck that contributes to the actual results.
Luck was probably a factor in the PEG having no losing years from 1986-1997. Any number of little things could have affected a stock or two that it selected and thus turned a winning year into a losing year. We are all at the mercy of hurricanes, tornados, earthquakes, fish kills, meteor strikes, corrupt management, public taste, and thousands of other factors that can cause a company to tank while we hold it, even if all the auguries are favorable on the day we buy it. In the second half of 1998, when the whole portfolio nose-dived, luck was probably a factor.
So when you look at two strategies, one that returned, say, an average of 37% and another that returned 32%, do you know how much of that return was skill and how much was luck? Of course not.
Since there is no way to know for sure, I have two recommendations. First, an investor should go through my five rules for finding a good screen. These rules won't tell you how much of the backtested returns are based on luck, but they will help you avoid some of the more questionable screens. Second, diversify so that if one of our many strategies does end up trailing the market in the future, only 10% to 20% of your portfolio will be affected.
Finally, be realistic. I have a simple goal: beat the market. I'd like to kill the market, but anything better than the S&P 500 makes me a happy man, which I guess explains why I'm so happy these days.
Until next time, Fool on!