One of the more persistent pieces of spam to fill my inbox is those little gems that relate to people or software’s ability to predict the market. I suppose this is the joy of having an email address that has the word trading in it. They generally start their pitch with something along the lines of in1983 our experts predicted that the All Ordinaries would be blah, blah, blah. Firstly even a blind squirrel finds the occasional nut and secondly…..bollocks.
I understand the psychological attractiveness of prediction, it brings a perceived sense of order to a chaotic and sometimes frightening environment. People are also apt to believe in the power of the expert, after all we trust many other areas of our lives to experts why shouldn’t we trust our economic future.
To get an insight into the world of prediction consider a study by Philip Tetlock. Who ran a 20 year study on expert prediction. He picked 284 people who made their living commenting or offering advice on political and economic trends and he asked them to assess the probability that certain events would occur, both within their area of expertise and in areas where they had no specialized knowledge. At the completion of the study in 2003 the experts had made some 82,361 predictions.
Tetlock measured his experts on two dimensions. How good they were at guessing probabilities and how accurate where they at predicting a very specific outcome.
After some particularly impressive number crunching Tetlock reached an interesting conclusion. Experts are useless at prediction. The studied experts had performed worse than if they had merely guessed about various outcomes. So experts who spent their lives and derived their income from predicting possible future outcomes in a variety of fields where no better than a lottery. He also found that they were no more reliable than non-specialists in the field who had merely guessed about a given outcome.
As we all know trading is also full of its fair share of experts whose accuracy is about impressive as that found by Tetlock in his study. The best example of this I have come across remains the forecasting record of mutual funds for the period 1954 to 1988. Whilst this study is almost 20 years old it still retains its core message. In this study the cash to asset ratio of funds was assessed each year. The logic was that if a fund went into cash it was bearish for the upcoming year. If it was fully invested it was bullish for the upcoming year. At the end of the study it was found that if you had followed the funds predictions you would have made 269 points on the Dow. However you would have lost 4,456 points. Clearly the correct strategy would be to do the reverse of what the funds were doing.
The question is what is the relevance of this to trading? The answer is somewhat different from my usual refrain of you are nuts if you think you can predict the market. My feeling is that people fall in love with their hunches and this problem is particularly true when we feel we have a plausible to back up the hunch. The more detail we have or think we have the more confident we feel about our predictions.
As humans we hate to be wrong and we will do almost anything to avoid being proved wrong. For traders this takes two generalized forms. Firstly clinging to whatever information supports our original position no matter how tenuous this position has become. Secondly by delaying action which confirms that we were wrong – this generally means that traders hold their positions long past their sell by date.
New traders find it hard to accept that we exist in a world that is unknown to us and that all our trading system does is to help us cope with this lack of future knowledge. When a properly constructed trading system generates a signal it is not making a prediction it is simply using past information to enable us to place a bet. Our trading system could stop functioning the moment we place this bet. It is this possibility that the system may stop working the moment we implement it and the fact that good traders use risk management that indicates that they have no future knowledge whatsoever. It is this willingness to accept that we are wrong and act accordingly that removes any use for prediction.
PS: A well known software company that up until recently had been trying to sell me a piece of market prediction software has recently gone tits up. I suppose they were not able to predict their own demise,
