Monday, September 19, 2011

Hula Hoops

For the past few years, the quant-shop style of stock-picking has been one cutest fads in investing. The general idea is that you base your buy and sell decisions on hard numbers, which supposedly 'takes the emotion out of investing'. A whole new website has been founded to make money from this idea - to make money for the site owners, that is, not investors - and this article is a recent example.

The article's author claims that he has run a value-oriented screen based on the "Graham Number" of a universe of stocks, the universe happening to be "stocks that are rallying above their 20-day, 50-day, and 200-day moving averages". He then compares the performance of his stocks over the past month with the S&P 500, and, lo and behold, his selections have outperformed the index. He would like us to think that this suggests that his supposedly value-based screening method might be worth looking at. Unfortunately, it suggests nothing of the sort. One could just as easily say that by selecting stocks which "are rallying above their 20-day, 50-day, and 200-day moving averages", the author has chosen a group of momentum stocks. Momentum stocks can indeed be expected to outperform over the very short term, because that is what "momentum" means. In other words, his experiment has not one independent variable, but two, and there is no way to un-confound their effects. In plain English, his test comparison says nothing about anything.

There are also a few other problems with this supposed value screen. The author tries to name-bash us from the very beginning by noting that "Benjamin Graham, the 'godfather of value investing' and one-time mentor of Warren Buffett, developed an equation to calculate the fair value of a stock, known as the “Graham Number.” (Notice the name Warren Buffett, which as far as I can tell has nothing to do with anything in the article.) Unfortunately, Ben Graham was also the man who said that value systems should help one pick stocks for the long term, when an undervalued stock should rise in price or otherwise produce a return reflecting a more 'correct' value. He didn't seem to be talking about a scale of one month. (Remember the voting machine and the weighing machine?)

But there's more. Graham's method is based on the assumption, among others, that a stock which you buy when it's undervalued is likely to rise in price, as the market notices the distortion and corrects it. That means it only makes sense to compare two values of a parameter - the value of the screen results, for example, with the value of the members of the S&P 500, if the difference really does represent a distortion, which one can hope will be corrected. Unfortunately, four of the eight stocks produced by the screen are utilities stocks, which have value parameters almost all of the time; therefore their low price should not be seen as a distortion, and there is no reason to expect the market to correct this distortion by raising the price. There is no reason even to assume that current price represents a low price for their income stream. The screen gives us essentially no information about them; it compares them with nothing relevant.

It's no wonder that money management and investing journalism are among those professions where a non-too-talented amateur usually does better than most of the professionals.

Labels: , , ,


Post a Comment

<< Home


to my broker review site