Customer Reviews

Curve Ball: Baseball, Statistics, and the Role of Chance in the Game

5 star:		(2)
4 star:		(10)
3 star:		(1)
2 star:		(0)
1 star:		(0)

 

 

Jim Albert and Jay Bennett share two traits that make them the perfect authors for this type of book (1) they are both baseball fans who know the game and have seen many games and much statistics from many angles and (2) they are both professional statisticians who understand probability and the subtle aspects that chance can have on statistics. By being professional statisticians they also know how sophisticated statistical techniques can add to ones ability to seriously address questions of strategy and comparison of player performance. That is what they accomplish in this book, teaching some basic probability and statistics along the way.
They also make it very interesting to the baseball fan by raising interesting baseball questions related to players that the fans relate to, namely the stars that the fans follow and the great clutch hits and clutch defensive plays that we baseball fans have imprinted in our memories, like Mazeroski's game winning home run in the 1960 World Series, or Willie Mays' famous over the shoulder catch of Vic Wertz's long fly ball in the 1954 series, or Bobby Thompson home run that won the 1951 playoffs for the Giants.

In the very beginning Albert and Bennett distinguish themselves from the sports statisticians that are hired by the teams. The sports statisticians collect the data and present it in various ways. However, this is merely exploratory data analysis. Albert and Bennett point out that a numerical difference in a hitting statistic such as on base percentage between Chuck Knoblauch and Kenny Lofton may be a real difference in ability but may also be a small enough difference to be merely due to chance. Finding ways to analyze the baseball data to make probabilistic inferences like answering the question of whether Lofton is better at getting on base than Knoblauch is the focus of what professional statisticians do and is the theme of the book.

In the course of reading the book you will learn many things about baseball. Some may agree with previous notions and some will be surprises. You will learn about the massive amount of major league baseball data available, about SABR a society for baseball research and more. You will be opened up to the hinden world of professional statistics where probability models have been used for over a century to handle military, engineering, energy, environmental, agricultural and medical problems. These same tools in recent years have been used to handle baseball questions also.

They start with simple table top baseball games like All Star Baseball to introduce concepts. They then move on to baseball data and probability. Then they look at statistical questions, situational effects in Chapter 4, hot hitting in Chapter 5, methods of measuring offensive performance in Chapter 6, more sophisticated measures in Chapter 7, simulation models in Chapter 8, measures of clutch play and team value in Chapter 9, ways to predict performance in Chapter 10, analyzing World Series results in Chapter 11 and final comments in Chapter 12.

This is a great book for any one who loves baseball and baseball statistics. It also is a great way to learn and become interested in the techniques of the professional statistician.

For statisticians that teach statistics, it provides a wealth of interesting examples to help illustrate important statistical concepts in basic or even advanced courses, including the value of Bayesian methods, the need for overdispersion models (e.g. batting averages) and the value of linear and nonlinear prediction models.

This is a book that I was excited to buy but unfortunately I did not enjoy it as much as I had hoped for. The two main reasons for this are 1) the lack of major insights and 2) the huge quantity of typos (I stopped counting after around 20). The copy editor for this did an absolutely terrible job, I'm afraid to report (writing this guarantees a typo somewhere in my review :) ). Some of the players' names are spelled incorrectly, and some of the numbers in the charts are inconsistent. This is very distracting.

The book is divided into 12 chapters, starting with a fairly trivial look at tabletop baseball games. The authors devote much attention to evaluating offensive performance, comparing various measures such as batting average, SLG, OBP, linear weights, total average, runs created and a few other more obscure ones. There is also some discussion of clutch hitting and a look at "Did the best team win the World Series?"

One of the other problems I found was that the authors stop short of providing actual statistical formulas and get into hand-waving a few times. I see that they have all the academic credentials but it seems as though they took the "book sales are inversely proportional to number of equations in the text" relationship to heart. I felt the book suffered because of that.

I'm not sure who I could recommend this book to. I have a feeling that some of the concepts might be too advanced for kids, and not deep enough for those who have a decent understanding of statistics. If you are into baseball stats but you don't know much about real statistics you will definitely find some new concepts in here, so I guess that is the audience.

I'm not sure who the target audience for this book is. At the beginning the introduce the most basic of statistics concepts as well as the most basic of baseball concepts. And at the end they seem to assume lots of knowledge of statistics. As both a baseball fan and a mathematician, I felt the beginning of the book very slow, but I also worry that someone who isn't knowledgable in both already might have trouble following some of it.

That said, the book was very well written, and posed some interesting ideas and questions. I wish it had been longer, as the last few chapters were really getting me into sabermetrics!

As a statistician and baseball fan, I had high expectations for this book. Generally, those expectations were met, although I came away from this book feeling like an opportunity had been lost. The biggest problem with the book is that the authors can't seem to decide how much knowledge to assume of their readers. The first 100 pages or so are presented at a sub-high school level, while the last few chapters assume the reader to have taken higher-level college courses in statistics. Also, I do not expect any book to be written and edited perfectly, but the typos actually become an occassional distraction from the text.

On the whole, though, I would still recommend this book--it is by far the best contemporary statistical breakdown of the game of baseball. It is an especially good complement to Michael Lewis's "Moneyball," which is a more anecdotal presentation of similar material. If you enjoy baseball at all, and have even a passing interest in batting averages, ERAs, and HRs, you will be entertained by this book and will probably learn a lot, too.

This one is a book for the Sabermetrically inclined who already have a background in stats. In the first couple of chapters, the authors review some basic concepts through the lens of baseball before getting into some deeper analyses. To be honest, there's nothing in here that you can't get in Baseball Between the Numbers (although to the authors' credit, this book predates BBTN by 6 years) but it's a decent starter's guide. Worth the read, although those with a background in Sabermetrics will probably want to pass.

Very good book. Most academic types write about sports as if they have never watched a game in their life, but not so in this case. Good for beginners and others who already have a working knowledge of sabremetrics.

The world of sabermetrics is a much-explored one. Many topics have already been explored by some stat guy or another in the last 20 years. Some have been explored many times (clutch hitting, for instance). One of the best things about this book is that because the author isn't particularly immersed in sabermetrics, he manages to strike out (NO PUN INTENDED) into new territory and has some somewhat new insight on things that have already been looked at elsewhere.

Baseball game nerds will recognize the game at the beginning of the book as Ethan Allen's All-Star Baseball, one of the oldest on the market. It is, as noted, pretty basic, which is why it's so much easier to use as a model than, say, Strat-o-Matic or APBA (additionally, it isn't really, really awful, so it beats APBA there as well). He does return to the simulation from time to time, which is kind of fun but isn't really necessary: most statheads are well aware of the concepts of probability and random chance.

One area he does really strike some new ground in is in the category of consistency. It's really just an introductory look at the subject (to sum it up shortly, he demonstrates that Todd Zeile in the late 90s was subject to a lot more streaks and slumps than you'd expect by random chance) and as such it doesn't really *prove* anything, but it's definitely an avenue that deserves more look.

The book as a whole is worth nuggets of stuff like this. A full-on stathead will decry the All Star Baseball format because it doesn't truly model the way baseball works (as a guy who's done beta testing for a computer baseball game, I can confirm that people like this exist), but IMO that's a small criticism to make. A bigger one is that there are a lot of ideas tossed out but not a lot of proof made. However, if you already have TangoTiger's "The Book", subscribe to Baseball Prospectus, and have thrown away your Derek Jeter bobblehead because he sux, this is a good book to have on your bookshelf.

For anyone who wishes to understand cause and effect in baseball, this is a very insightful book. The authors clearly and convincingly demonstrate that many of the statistical outcomes we generally attribute to a player's ability are really nothing more than random effects. For example, a player with a lifetime batting average of .300 who hits .280 during a season is said to have had an "off year." The authors show that any player who has a true batting ability to produce base hits in 30% of his at bats (i.e., a .300 hitter) can be expected to hit .280 or less or .320 or higher about one-third of the time. For someone who grew up thinking that all these year-to-year fluctuations were the result of "good years" and "bad years," the very significant impact of randomness came as a rude awakening! But for the serious student of the game, this is a critically important insight.

Similarly, the authors show that a team's win-loss record during any single season may not reflect the team's real ability. Again, in 162 game season, randomness rears its head. It is not that uncommon for a team to win 12 to 15 games less (or more) than its underlying talent would suggest.

As we reduce the number of games in a series (for example, consider the typical best-of-seven post-season playoff format), the effects of randomness are greatly magnified. Thus it is not at all uncommon for the best major league team in any season to fail to win the World Series.

"Curve Ball" is well written, and the authors do a good job of explaining the statistical models they employ. I often find myself returning to this book to refresh my understanding of baseball probabilities.

The one deficiency that bothers me most is the lack of a subject index. Thus the reader is forced to thumb through the book to locate some particular topic of interest. But even so, this is an excellent book that belongs in any good baseball library.

More than any other sport, baseball is a game of statistics. This book goes behind the numbers and looks at where the true patterns are and where the seeming patterns are just the result of chance.

As an example, the book discusses how, in general, hitting is not affected by night play or day play; on the other hand, there is an effect for facing a right-hander versus a southpaw, based on the side of the plate you are on.

Generally well-written, this book only sometimes gets bogged down in statistical calculations and is generally accessible to the nonmathematician. The main flaw in the book is its emphasis on hitting and the relative lack of writing on pitching. While there is plenty of discussion on the value of batting average, slugging percentage or on-base percentage in determining runs scored, there is no similar discussion on ERA or strikeouts and its impact on wins.

The other problem I find with the book is it removes some of the mystique of the game. It's sometimes more fun not to overanalyze things; it's kind of like watching a magic show; if you understand what's happening, you feel smarter but some of the pleasure has gone away. Which is not to say that I'm not going to continue enjoying baseball, but I will look at the game with more scrutiny when it comes to all the statistics that are cited.

Curve Ball: Baseball, Statistics And The Role Of Chance In The Game provides the non-specialist, general reader with a sophisticated by accessible approach to statistics that can greatly enhanced their understanding and appreciation of baseball numbers and the game itself. With their unique and original approach to the subject, Jim Albert and Jay Bennett have successfully collaborated to present a coherent and informative introduction and survey of the many statistics that are a part of the baseball experience. Curve Ball is a "must read" for all baseball fans and enthusiasts.