Top critical review
63 people found this helpful
Provocative, entertaining, and grossly flawed
on 14 April 2017
I will summarise the book in one paragraph and then give positive and negative comment and a more serious critique of his message.
Forecasting is extremely difficult and usually little better than guessing. Most experts in social sciences, especially finance and economics, overrate themselves, proclaiming abilities they simply don't have (he uses the word "fraudster" many times). Risk analysis, and indeed all statistical work that relies on Gaussian assumptions, doesn't work because Gaussian assumptions don't hold in most real world situations, which are subject to "unknown unknowns". He cites the example of a casino which used mathematics to control its losses to gamblers but still lost a lot of money when an irreplaceable performer in their main show was maimed by a tiger, and when the owner's daughter was kidnapped. it nearly lost its licence when an irresponsible employee hid a bunch of tax returns instead of filing them. These, of course, are the black swans of the title.
It is easy to read and entertaining, with an engaging human touch. Its insights are largely valid and important. It is a useful provocative challenge to conventional thinking and a slap in the face to self-important economists who take themselves too seriously.
It is repetitive and could have been written in 50 pages. By the mid-point, the constant rubbishing of the entire scientific and mathematics community as fraudsters felt like a tiresome rant. Towards the end it felt like listening to the charismatic leader of a religious sect explaining that only he knew the truth and all those who didn't listen to him were damned. Indeed, the author is even more arrogant than those he attacks.
The world of finance has known for a long time that its models don't allow for black swans. The question is whether to use existing models which work most of the time, or some alternative. Taleb isn't clear about what his alternative is though he quickly dismisses other peoples' attempts to allow for black swans in existing models. He says his method "develops intuitions from practice", relies on "skepticism", and "sophisticated craft", "seeks to be approximately right across a broad set of eventualities", "respects those who have the guts to say I don't know", and uses "messy mathematics and computational methods". None of this is specified, it is just motherhood and apple pie.
He gives a hint of what he means for investing. "I worry far more about the 'promising' stock market, particularly the 'safe' blue chip stocks, than I do about speculative ventures - the former represent invisible risks, the latter offer no surprises since you know how volatile they are and can limit your downside by investing smaller amounts."
It is worth taking a good look at this logic. He is saying he invests in risky assets but because they are risky, he doesn't put much money in them. Where does he put the rest of it then? Presumably in the safe blue chip stocks which he just said he doesn't like. Notice the comment "you know how volatile they are". He spends most of the book preaching that you never know how volatile things are. What he means here is that he knows they are very volatile. Reading carefully, his strategy amounts to this. The market generally undervalues some risks and overvalues others and, using his intuition, he does the opposite in order to benefit from other peoples' errors. This is a sound approach. It is also a common and well known one. The hard part is how to identify the errors. He thinks you can do it by allowing for black swans, since most trading models don't allow for them. But that means he is actually relying on all those existing models, albeit in a contrarian manner. If everybody stopped using them, his strategy would disappear. He has made the case that black swans are important and difficult to account for. It does not follow that the market would work better if existing methods were ditched.
Finally, his rude dismissal of all maths that uses Gaussian distributions is crass. For example he points out that words are distributed according to a power law, not Gaussian. That is true, and yet speech recognition, and all the language based artificial intelligence that is rapidly becoming powerful and useful, is based on mathematics that uses Gaussian distributions at its core.