Summer Reading II, III--The Predictors; The Black Swan

I've put these two together because they share a theme, one I'm particularly interested in these days: the problem of prediction.



The Predictors, by Thomas A. Bass, is subtitled "How a Band of Maverick Physicists Used Chaos Theory to Trade Their Way to a Fortune on Wall Street". The subtitle takes most of the mystery out of the telling of the story, as they work through their organizational issues, find funding, generate reams of unworkable models, and finally--in the last chapter--actually start making some bucks (1997 or so). The author had chronicled some of the physicists' earlier (unsuccessful) attempt to devise a casino-breaking roulette calculator, and he was in on the discussions from the very beginning. So, that part of it was interesting.


Also interesting, and one that took me back to my own career efforts in the mid-90's, was their attempt to bring in huge data streams, address the predictive qualities of the data, and develop and operationalize models which would predict specific outcomes from the data.

The author made a lot--too much--out of the notion that these geniuses chose to launch their envelope-stretching endeavor from the puny backwater town of Santa Fe.

Their methods made a lot of sense from a model-development and validation point of view, but I felt they were overreaching. Essentially, they wanted data feeds of most everything in order to develop models predicting--almost everything. They should have focused on a few key trading indices and bent all their efforts to find the occasional opportunities these markets produce, instead of trying to get involved in a whole bunch of different markets and trading all the time.


The book needs a postscript, epilogue, or second edition for today's readers to understand how well the approach worked in the long run, through the vagaries and changes in the marketplace (some of which this group may have, in fact, driven), and especially in response to the calamity that occurred 9/11/01. How well did chaos theory (which seems to have more inspired the model-building effort than drving the actual functional models) anticipate and deal with that one, and its aftermath?



Which brings us to the tougher read and subject of critical analysis, The Black Swan (subtitled "The Impact of the Highly Improbable"), by Nassim Nicholas Taleb. This new volume has been a major success (#25 in Amazon's nonfiction titles, last I looked).

The key notion of the book is that nonlinear processes describe key events in society (also a key idea in The Predictors). Forecasters are deceived by their need to place narratives on phenomena that may not fit well, and statisticians and economists are deceived in their efforts at risk management by using faulty assumptions of normality. The events that disrupt things in a nonlinear way Taleb calls "the black swans" (let's not go into why--suffice to say it yields a catchy title).

Taleb hits particularly hard on those whose job is to control risk but never consider the possibility of new events, ones that have never occurred and thus not easily considered empirically. Chaos theory also comes out here.

Taleb is a man who has given up reading newspapers and magazines and watching the news; he claims the time it freed up has allowed him to read many more books. He is certainly well-read. His style is gruff and off-putting; rather gratuitously, he throws an insult in the first 20 pages at anyone who stands for a woman's right to choose (regarding abortion) and dares to call oneself a humanist. He doesn't suffer fools gladly, and everywhere he looks he sees mostly fools, so he's irritated all the time.

I agree with many of Taleb's arguments and find it to be original and thought-inspiring. So why did I want to disagree with him constantly? It was that style.

I have done many projects to look at future behavior based on past behavior, carefully constructed as those with the Predictors. I would agree that the conditions of normality are usually absent and that that fact is often ignored. The proof of the model is simply how well it predicts, not the theoretical basis, though, in the end, we would always only include factors for which we could conceive some sort of a priori justification (a blind spot--though we tested the others, too). Taleb would hate that narrative externalizing.

The real issue comes from the only type of business Taleb finds interesting--complex and highly-leveraged derivatives. With these, there is no room for an occasional off day. Taleb advocates a highly conservative strategy (e.g., T-bills) while leaving oneself open for the big score. My strategy was always to select a small number of high-likelihood gambles, each providing a reasonable return and avoiding excessive exposure in each.

I have to commend his ending for the main text: "Thank you for reading my book." Dignified, courteous.