VALUE INVESTING
The Value Of Random Numbers

By Stephen Bland (TMFPyad)
January 14, 2005

Part of the human condition is that we seek often to find order in chaos or randomness. And we often find it too though the discovery may well be false. It seems to many people somehow wrong that actual randomness exists in nature or human activity. There has to be a pattern. Or does there?

I recall reading a book years ago on math and gambling where the writer showed how apparent patterns can be discerned in tables of statistical random numbers if you look hard enough at certain parts of the series. Yet no such pattern can exist overall and even if it does for a part of the series examined, it is certain that it won't repeat in the end. This is strengthened in my opinion where people looking for patterns employ the unscientific technique of trying to prove their own pet views, as against the proper scientific approach of trying to disprove them.

Years ago in the early days of The Motley Fool, mechanical share schemes were all the rage. I too was a fan of some aspects of this and designed my own one based on high yield. I deliberately did not base this on back tested results, which in any event were impossible to obtain, but purely on investment logic I believed to work in the market. However a lot of people came up with schemes, usually technically based on momentum etc., showing that outrageously successful results could have been achieved in the past based on back testing. It was then hoped that this idea would extrapolate into similar future performance. Almost all of the schemes failed in the end as the market conditions which had given rise to past success changed radically.

What these naïve researchers were doing in the main was having an idea for picking shares, then looking back over past market data for evidence that it worked. Like patterns in random numbers, if you look hard enough you can find evidence that almost anything will work for a period. But it can be a false discovery. The sceptical amongst us called this process data mining, in a derogatory sense to mean seeking patterns over a past period which fit a preconceived notion.

The paradox is though that even if you have a successful approach, you don't really know whether it is still just a phoney pattern for a time in a series of random numbers. As readers will know, my two preferred approaches to shares are value for short-term trading and its long-term cousin, high yield portfolios for income investors or those who don't wish to trade much or ever.

Now as far back as I can recall personally, which is several decades now, and even further from reading the extensive research on these subject in books from people like James O'Shaugnessy and David Dreman, value and high yield have been exceptionally successful approaches to investing. Nevertheless I have to admit that it is possible that they are the false patterns in random numbers, even though these approaches have been around for a very long time. I don't actually believe that they are false, I am just admitting that it is possible.

Supposing over a person's investing lifetime, they pursued a strategy which indeed turned out to be a false pattern but only long after they had gone. You can seen that it wouldn't then matter whether the pattern was false or true simply because it had worked for them in their life. It would though matter for those who came after and picked up on the strategy, particularly if it turned out to be false whilst they are in the middle of it.

Whatever period we research, it is only a part of an eternal series. The question is whether patterns observed in the period examined are representative of that eternal series or whether they may be false patterns that we happened upon. Even if false amongst the random, the question then for an individual is whether they will persist long enough to work in that person's lifetime. The only answer is that we can never know for certain, for the obvious reason that we don't have the eternal series in front of us from which to draw the ultimate conclusions about what really worked in the end and what did not.

In the real investment world it doesn't matter because people don't have eternity, they have only a finite number of years in which to invest. So what sensible strategy investors tend to do is based on what has worked in the lengthy past, rightly or wrongly,  But I base my faith in value and high yield on something much stronger than just decades of past results, very powerful though those figures are. I base my faith on human nature and now we really are talking of something that is tens of thousands of years old, not just the couple of hundred years of the stock market.

And I know from this that buying something as good or better than something else, but much cheaper than it because of nothing more than highly fallible human sentiment, has got to produce results over time. Build on human stupidity and you build on rock as my Dad used to say.

You can follow my ideas on value and high yield shares in my Value Investor monthly publication.