Tuesday, January 11, 2011

ESP and Investing

There's a huge controversy bubbling through the psychology world about ESP.  They New York Times has covered it nicely, both here and here.  A researcher has demonstrated that a group of college students have the ability to predict random appearances of porn on computer screens.  No, I'm not kidding.

Here is a response to the paper. I nominate this paper as the best piece of finance writing I have seen in years.  No, I'm not kidding.

If you took a statistics class in college, you learned this stuff.  You just didn't understand it.  (I got a D in Professor Schervish's intro class...the indignity of this experience only got worse when several years later Lars Hansen, world famous econometrician and my dissertation adviser, found out!)  Remember hypothesis testing?  We reject the null hypothesis that blah blah blah.  Hypothesis tests, those used academics the world over, assume you know nothing about the data before you establish and then test your hypothesis. The test requires different methodology when you know something about the data.

So, why is the rebuttal to the ESP paper great reading?  I for one would prefer to contemplate statistics methodology with visions of erotic images on computer screens to the usual urns filled with black and white balls.  More importantly, the authors carefully and lucidly explain the key methodological issues behind statistical tests that all of us too frequently ignore.

Please, read the paper.  Then, think about it the next time someone tells you they've back-tested their investment strategy, and the results are statistically significant.

3 comments:

  1. Marc,
    Thanks for posting the response. When we lived in Durham, we lived down the street from the Rhine Research Center where did many of the famous experiments.
    I have always found back-testing of strategies to be suspect at best.
    Best,
    Bruce

    ReplyDelete
  2. great post marc...hope all is well. Adithya

    ReplyDelete