It is hard to keep this book in any one category, particularly so when the author is a philosopher, a literary essayist and most interestingly a quant trader. The message of this book is the futility of creating complex mathematical models (and in believing them) to forecast future events. The real life is too complex and random to fit in any models.
Most of the book revolves around the limitations of Gaussian bell curve. He shows with examples how the predictions based on bell curve could go wrong when applied to scalable variables such as stock market returns, bank defaults, personal wealth etc. When one single random event can erase decades of gains (like personal wealth etc) it is time to stay away from forecast pundits.
He also talks about the typical statical biases such as confirmation bias, survivor ship bias, pattern finding, sampling errors etc. While these may look too basic we are in recession today because the CDO models used had all these statistical biases built in. (note that the book came well before the current recession).