22 July 2008

Was Anyone Surprised By This?

This recent Freakonomics post shows that forecasters have a done terrible job at predicting the price of oil.

That should not be surprising.

Why should the price of oil be predictable? It depends on factors that are barely measurable by humans such as future consumer sentiment, future government action, future weather and future geo-politics.

Nicolas Taleb has described this in great detail over the last few years. He calls the mis-application of mathematical models to data which they cannot hope to model the Ludic fallacy .

In some cases people make a jump from analyses of recent history to prediction of the future. Businees books such as Good To Great come to mind.

Ronald Fisher who was good at prediction said this about pundits.

In scientific subjects, the natural remedy for dogmatism has been found in research. By temperament and training, the research worker is the antithesis of the pundit. What he is actively and constantly aware of is his ignorance, not his knowledge; the insufficiency of his concepts, of the terms and phrases in which he tries to excogitate his problems: not their final and exhaustive sufficiency. He is, therefore, usually only a good teacher for the few who wish to use their mind as a workshop, rather than a warehouse.- R.A. Fisher. Eugenics, academic and practical. Eugenics Review, 27, 95-100, 1935.

For those who care, here is an attempt at explaning of oil price trends http://europe.theoildrum.com/node/4007

No comments: