Ride a black swan
As the follow-up to Taleb's landmark Fooled By Randomness (a book whose ideas are sadly more widely referenced than acted on), it's been getting plenty of press - for example, this by Oliver Burkeman in the Guardian - so I don't need to repeat its arguments here (basically it's all to do with non-normal distributions for many events and phenomena, not least asset price movements). It covers more ground than the earlier book and is perhaps a bit less coherent, but it's still a pretty great read with a much-needed message. Remarkably for a 300-page polemic about statistics and orthodoxies, it's never dull. Actually, I'm wondering whether I should be worried that I found the closing technical section - which Taleb recommends that most readers can happily skip - the most interesting and provocative.
Taleb firmly places his chips with Benoit Mandelbrot, to whom the book is dedicated, and his work on scaleable distributions. For more on this and Mandelbrot's other work in economics, see my interview with the man himself from 2003.