Taylor's Reviews > The Book: Playing the Percentages in Baseball

Good for what it is: an attempt to use rigorous mathematical modelling to determine the most statistically sound plays in baseball. It attempts to answer questions like when are the best and worst times to steal, hit and run, sac bunt, etc. It goes into some depth about how best to create a batting order and when to use a relief pitcher for maximum effect. There is even a chapter on if there is such a thing as being "clutch" and how long a hot/cold streak lasts.
But, while I don't question the math, I do question their sample size. Most of the data used is from the 1999-2004 seasons, which seems small for deriving general advice based on statistics. That the data used is from smack in the middle of the steroid era makes me a bit skeptical of some of their conclusions. Furthermore, some of the techniques used are less rigorous than others. For example, when trying to determine whether a batter "owns" a pitcher (and vice versa), the author uses the concept of "families" of players, the qualifications of which were never really explained all that well and the relationships seemed to be a little arbitrary. That good math gets piled on top of that to prove a point seems a little bit like building a house atop a sand dune.
Like I said in the beginning of this, it is good but not great. There are useful pieces of information to be gleaned here. But, other books have covered this side of baseball better, in my opinion. The Hidden Game of Baseball for instance.