This board has been migrated to our new platform! Check out the new home page at discussion.fool.com or click below to go directly to the new Board on the new site.
This was posted over on the Foolish Four board (#22017). I tried searching the net for the original article but I could not find it. The author though seems to publish quite a bit. The article is on the FF but it applies to the workshop screens.Mined, All Mined: Watching Out for Funny Numbersby William J. Bernstein | The clerk at the Seven-Eleven knows you're there even before she looks up. Because every night after work you stop by for the paper, a decaf, and five lottery tickets.Since the payoff averages only about half of the ticket price, this hasn't done much for your investment strategy. But that's OK, because you have a plan. Since day one you have kept meticulous records, and some very interesting patterns are popping up.Thursdays, for example, on average you almost break even. And the third Thursday of each month, bowling night, has actually turned a tidy profit. But wait! Bowling nights in the summer are particularly lucrative, netting several hundred percent. You've found your way to easy street: Bet the farm on summer bowling nights. Maybe you'll even publish a best-seller describing your sure-fire technique. Now, anyone with an IQ above room temperature can see the flaw--your payoffs are purely the result of random variation. There is no good reason why the season or day of the week should influence your luck. You're guilty of data mining--sifting through a series of chance results, and picking out the associations with the happiest outcomes. Statisticians and pointy-headed finance types will dismiss your analysis as the inevitable sorry wages of in-sample analysis. In other words, you have found associations peculiar to the limited amount of data available to you. The cure? An out-of-sample study. In the above case, another several years' worth of summer bowling night lottery purchases. Goodbye Maui.If you intentionally search for associations with stock market returns, you will come up with some pretty amazing specimens. Perhaps the best known is the "Super Bowl" strategy--stocks lead the year after a team from the old NFL wins the game, while T-bills win out when an old AFL team prevails. Even better, tune your equity exposure to butter production in Bangladesh, which "predicts" 75% of S&P 500 returns.Which gets us to The Motley Fool. For those of you who have just returned from missionary work in Burkina Faso, this wildly popular website offers several easy recipes for beating the market, all comprehensible to the average cocker spaniel. Perhaps the most famous is the Foolish Four: 1) Start with the 10 highest-yielding stocks in the Dow Jones Industrial Average.2) Pick the 5 with the lowest share prices.3) Drop the lowest-priced stock.4) Double up on the second lowest-priced stock.Step 1, of course, is the venerable Dogs of the Dow. But Steps 2 through 4 are head-scratchers. Why should the share price of a stock have anything to do with its return? And why should there be such a large discrepancy between the returns of the Dogs with the lowest and second-lowest share price? These make about as much sense as summer bowling night lottery purchases. Enter Gary Karz, who runs the Investorhome website and knows data mining when he sees it. With a bit of digging, Mr. Karz actually found seven Foolish portfolios. Two had outperformed the market, two had underperformed it, and most ominously, two had been closed. Karz contacted academicians Grant McQueen and Steven Thorley, who had previously examined the Dow Dividend strategy. (Bottom line: It beats the market by a hair, but only before taxes.) The result, "Mining Fool's Gold," is a scholarly examination of the Foolish Four strategy, published in Financial Analysts Journal, as well as online. It's first and foremost a lucid and funny primer for small investors on the dangers of data mining, and how to spot it. Among the warning signs:1) The presence of multiple strategies, some of which are dropped.2) Strategies that make little theoretical sense, like the Superbowl method, or one using raw share-price data.3) Overly complex trading rules. The authors find the Foolish Four guilty on all counts. Yes, it beats the Dow by 2.95% annually for the July 1973-June 1996 period. But apply the acid test--an "out-of-sample" analysis using a 1949-72 test period--and it wins by only 0.32%. And with 40% more risk. It actually underperforms the simple Dogs strategy for 1949-72.The Fools are not alone. James O'Shaugnessy, author of the wildly popular How to Retire Rich, and What Works on Wall Street, similarly mined stock selection strategies over the past 50 years and came up with a number of sure-fire methods, ranging from a simple Dog-like portfolio to more complex strategies. Following publication, the author started several funds based on his methods. Thus far, the out-of-sample results have not been encouraging. Since their inception in late 1996, only one of his four funds--O'Shaughnessy Cornerstone Growth OSCGX-- has beaten its peers or the appropriate Barra Index. O'Shaughnessy will certainly retire rich, but his shareholders may not. (And really, 1.46% in expenses for a Dogs-of-the-Dow OSDGX fund? Gimme a break.)Even the most respected academicians cannot occasionally avoid temptation. Wharton's Jeremy Siegel, in (Stocks for the Long Run), famously points out that if you ignore 1975-83, small-cap stocks don't outperform. Well, yes. Make me 25 years younger and 40% faster and I can play cornerback for the Broncos. C'mon, Jeremy, real academicians don't vaporize whole decades. So next time somebody offers you a convoluted market-beating strategy based on historical returns, hide your wallet.
Best Of |
Favorites & Replies |
My Fool |