About John Vos

I have less than 250 friends. Not enough to be on Facebook. I'm mentally incapable of condensing my message into a few sentences, let alone 140 characters. No Twitter either. I promise to join LinkedIn as soon as I'm in desperate need of a job that values networks more than skills and experience (not at this moment). I got banned from eroticdating.com for reasons that are unimportant. The only way you can contact me is by sending me an email: Firstname.lastname1@telenet.be (First name : John, Last name : Vos)

SCIENTIST BEHAVING BADLY – A Review of Models. Behaving. Badly. by Emanuel Derman

The book’s subtitle is “Why confusing loose-association metaphysics with a sound argument can lead to absurdity, in finance and in philosophy”. Just kidding, that’s the subtitle of my review.  First I should say that I deeply respect professor Derman. He stands for the ‘D’ in the once-popular BDT model (Black-Derman-Toy in full) for interest rates. I loved his previous book, ‘My Life as a Quant’, an engaging, candid and humble account of his career, first as a physicist, then as a quant in finance. His lively, detailed account of the genesis of the BDT model left a deep impression on me, the more so because it painted a much more realistic picture of the protagonists, or by extension quants in general, than the straw men critics tend to make of them.

When I bought his latest book, I was worried, though. Not because of the title (most models come with problems in some respect). Rather, it was the subtitle ”why confusing illusion with reality can lead to disaster, on wall street and in life” that got me worried. I was afraid it would be yet another unfounded attack at quants, yet another mindless reduction of the complex causes of the financial crisis to “a religious belief in flawed models”. The book surpassed all my fears.

I can be short about the strong points of the book. I give one star for Derman’s passionate presentation of the most successful theories in physics (the second star is because I’m a generous man by nature). The reader be warned, though; not all of physics shares the spectacular predictive accuracy of quantum mechanics and particle physics. The study of complex systems, for example, is much closer to ‘modeling’ than to ‘theory’ (in Derman’s definitions of those two words). The difference between ‘models’ and ‘theories’ (as Derman defines them) is not so neat as his juxtaposition of finance to particle physics suggests.

Previous reviewers complained about the shallowness of Derman’s analogies and arguments and the lack of focus in the book, pointing out that its central theme seems to be Spinoza’s misty philosophy rather than finance. I share those reviewers’ complaint, but I don’t consider it the book’s weakest point. I think every sane person understands that Derman’s comparison of a financial derivative (e.g. stock option) to “love as a derivative of pleasure” in Spinoza’s philosophy is a case of amateur poetry, and certainly not supposed to explain financial derivatives by means of a far-fetched analogy with the psychology of love.

The weakest point is that a professional quant whose name is attached to some famous quantitative models, hence whom you would reasonably trust to give a fair account of financial theories, lapses into the same form of straw man thrashing that you can expect from minor geniuses as Taleb and Triana (Derman co-authored some recent papers with Taleb; Sleep with the dog and catch his fleas?), only now with the stamp of approval from a real expert with a real reputation among professionals (it’s much easier to win a reputation among laymen; toss around a bunch of technical words and you’re ipso facto considered an expert).

His treatment of the Capital Asset Pricing Model (CAPM) is a nice illustration. He gives a clear and fairly accurate explanation of the model, but I doubt a reader who’s not well versed in statistics will understand the term ‘expected return’ correctly as a return on average, with a lot of dispersion around that average.  That misunderstanding is crucial to the acceptance of Derman’s “simple test” proving CAPM wrong. He compares the one-year (!) return of Apple (one security!) with the return of the S&P, and observes that Apple’s small beta grossly underestimated the stock’s excess return. Let’s get this straight: a sample of one observation, observed over an extremely short observation period, is supposed to invalidate the CAPM??! And it’s not even a random sample; Derman undoubtedly singled out Apple because it was a top performing stock over that particular period. Shame on you, professor! You can get away with that in a book written for a lay audience, but try it in an academic paper, and reviewers think it must be a joke.

Another illustration of how deep an intelligent man can fall when not constrained by his peers to follow the rules of logic are his comments about the Efficient Markets Model (EMM). According to the EMM, he says, the best estimate of a security’s value is its current price. No objection from my side. But then he continues to offer a counterargument: “Anyone with hindsight can see that the market is sometimes wrong about the value of an asset”. With hindsight, you’ve said it, professor. Foresight is a bit more difficult. But didn’t you notice the egregious logical error? The current price is the best estimate; that’s not the same as an infallibly exact prediction. The best estimate only means it’s better (on average!) than any other estimate. Just like the CAPM, there’s a lot of dispersion here, explicitly recognized by the model. Whether market prices are effectively the best estimate, that’s an interesting question. But it needs a little bit more sophisticated statistical tests than simply referring to market crashes, as is invariably done in the popular press as a way to ‘prove’ the falsehood of market efficiency. I presume Derman is wise enough not to be disappointed if the academic literature doesn’t include his ‘proof’ as a serious counterargument to market efficiency.

After reading the book, I still don’t know how confusing illusion with reality can lead to disaster. All Derman says about the topic is that (some) quants tend to believe in the literal truth of their models, without naming any.  He does name Fischer Black (and himself, of course). But only to generously exculpate him from the sin of blind reliance on mathematical models (deservedly, in my opinion). Here’s a question I’d like to ask professor Derman:

Why should we believe other quants do make foolish mistakes Fischer Black and you never made?

Pray tell, professor: what reason have we to believe quants are prone to confusing illusion with reality?

 

Emanuel Derman: “Models.Behaving.Badly.: Why Confusing Illusion with Reality Can Lead to Disaster, on Wall Street and in Life”, Free Press, July 24, 2012

 

(This is a slightly different version of the review I posted on amazon)