This was much better than the last book on statistics I read. At least it would either prove something, or state that the math involved was beyond the scope of the book and leave a proposition unproven. It also talked some about the limitations of the ideas and approximations that were being used. It even made an attempt to explain "degrees of freedom."
I like this all the way through the basic hypothesis testing chapters. I liked the approach taken here to non-parametric testing. But that's about the point where things started to get fuzzy. And it was bit fuzzier with Chi squared distributions. And by the time we got to F Distributions, it happened again: I felt like bees were living in my head.
What I take away from this is that I pretty much dislike statistics. It all seems like a huge house of cards to me: alot of very elegant math built on what may or may not be some pretty shady assumptions. That said, I did take away some very useful ideas that might have some application to testing trading systems.