Posted by
Andrews on Tuesday, September 01, 2009 1:14:46 PM
I have written before about my fascination with the stock market. Part of that fascination is with those who think they will be able to "beat the market" by finding the right model to predict stock prices.
What amuses me is how many rely on some version of moving averages (MACD). I was just looking at the Dow, and for the past 3 months, it seems the Dow completely ignored conventional wisdom, at least using a 15 day moving average. For nearly two months it remained well above the moving average without even a hint of downward motion, then briefly crossed, ignored the average again, and then dropped well below and stayed there.
Of course, the MACD proponents will say I am using the "wrong" period of time. Just like the AGW advocates, every stock wizard thinks that MACD only fails because we pick the wrong time period. or else, maybe MACD is "too simple" and their slightly more complex model will work.
Now, I am not saying that mathematical analysis is pointless. As a general indicator, it is very valuable. As a way to decide how much hedging one needs for a safe portfolio, the numbers are generally good. That is the same case with most models. So long as we recognize that they
ARE models, that the predictions are generalizations and can't be trusted too greatly.
The problem comes when people trust models too much, in any field, be it stock prediction of AGW. Once you forget the model generalizes and conceals some degree of error, you fall into error yourself. And you end up with day traders losing their shirts working some absurd system, or AGW advocates explaining that a model that can't predict tomorrow's weather, or calculate yesterday's, is going to tell us precisely what is coming in a decade.
Actually, with stocks it is even worse, as human volition enters into the question. At least it is conceivable we could eventually understand and predict the mechanistic processes of weather. But with human volition involved, as I wrote in "
The Limits of "Scientific" Management" there is just no way to make such predictions. Human will always makes a shambles of such attempts.