Stumbling On Wins: Two Economists Expose the Pitfalls on the Road to Victory in Professional Sports [Hardcover]

David J. Berri (Author), Martin B. Schmidt (Author)

Book Description

Publication Date: March 18, 2010

The next quantum leap beyond Moneyball , this book offers powerful new insights into all human decision-making, because if sports teams are getting it wrong this badly, how do you know you're not? Sometimes the decisions that teams make are simply inexplicable. Consider: sports teams have an immense amount of detailed, quantifiable information to draw upon, more than in virtually any other industry. They have powerful incentives for making good decisions. Everyone sees the results of their choices, and the consequences for failure are severe. And yet... they keep making the same mistakes over and over again... systematic mistakes you'd think they'd learn how to avoid. Now, two leading sports economists reveal those mistakes in basketball, baseball, football, and hockey, and explain why sports decision-makers never seem to learn their lessons. You'll learn which statistics are connected to wins, and which aren't, and which statistics can and can't predict the future. Along the way, David Berri and Martin Schmidt show why a quarterback's place in the draft tells you nothing about how he'll perform in the NFL... why basketball decision-makers don't focus on the factors that really correlate with NBA success... why famous coaches don't deliver better results... and much more.

Editorial Reviews

Review

As seen on The New York Times' Freakonomics blog, ESPN.com's True Hoop, and Slate.com.

 

"In Stumbling on Wins, sports economists Berri (Southern Utah Univ.) and Schmidt (College of William and Mary) follow up their The Wages of Wins (with Stacey Brook, CH, Jan'07, 44-2764) with more modeling and number-crunching applications. The holy grail remains the same: understanding and improving decision making on the court, field, and ice and in the front offices of North American professional team sports. Summing Up: Recommended. Sports and sports economics collections at all levels. Reprinted with permission from CHOICE, copyright by the American Library Association.

About the Author

David J. Berri is associate professor of economics at Southern Utah University. He has written extensively on sports economics for academic journals, newspapers, and magazines, including The New York Times.

Martin B. Schmidt, professor of economics at the College of William and Mary, specializes in sports economics and macroeconomics. His writing has appeared in the field’s leading academic and general interest journals, including The New York Times.

Berri and Schmidt coauthored The Wages of Wins and maintain a popular blog, The Wages of Wins Journal, which discusses the economics of sports decision-making (dberri.wordpress.com).

Product Details

• Hardcover: 256 pages
• Publisher: FT Press; 1 edition (March 18, 2010)
• Language: English
• ISBN-10: 013235778X
• ISBN-13: 978-0132357784
• Product Dimensions: 6.3 x 0.8 x 9.1 inches
• Shipping Weight: 1 pounds (View shipping rates and policies)
• Average Customer Review: 3.9 out of 5 stars  See all reviews (38 customer reviews)
• Amazon Best Sellers Rank: #559,829 in Books (See Top 100 in Books)

More About the Author

Discover books, learn about writers, read author blogs, and more.

Most Helpful Customer Reviews

22 of 25 people found the following review helpful

3.0 out of 5 stars Disappointing: nothing new, a limited unvariate analysis of a multivariate dynamic system June 2, 2010

By Peter G. Keen VINE™ VOICE

Format:Hardcover|Amazon Vine™ Review (What's this?)

This is disappointing and I don't recommend it, either for sports fans or anyone with an interest in the growing and substantial research on fallibilities in human judgment, decision making and use of information. Its basic point is one that is well established, that conventional measures of performance in professional sports are misleading predictors of future performance; examples are pitchers' ERAs and NBA points per game. It repeats the many-times made observations about how often NFL first round quarterback draft picks are bombs. That's well presented and thoroughly documented but in more detail than the use of the findings warrants. Its main point is that overreliance on the wrong data leads to bad economic decisions by managers who should know better. I don't recall any item in the analysis that has not been covered elsewhere. Examples here are: (1) Field managers and coaches in baseball, football and basketball have little impact on team performance, (2) Statistically, it makes sense to go for it on fourth down, (3) Trading up to get a high draft pick is generally a bad deal, economically and in terms of finding the best talent, (4) NBA draft position is a poor predictor of career performance, (5) The NBA "hot hands" streaks are a myth and (6) Isiah Thomas was a truly, truly lousy general manager of the Knicks. Agreed. Agreed.
The main weakness of the book seems to me that it largely relies on data about individual performance for its core evidence and though it alludes to the context of teams, it is very univariate in its analysis. The authors emphasize this but only in a single footnote. (The regression-based methodology examines only the strength of the linear relationship between two independent variables....) There is no multivariate perspective: the balance of a team's roles within a deliberate defense-offense strategy, career injuries, character burnouts and dysfunctions, owners, franchise history and the many contextual economic factors such as player caps and guaranteed contracts. It's almost all about individual stars only. The selected examples seem casual and arbitrary. In a book that argues that the role of the coach is minimal, for instance, there is no mention of, among others, Bill Belichick, Bill Walsh and Scotty Bowman, as counterexamples or confirmatory ones to be explained through the same methodology of data analysis. None of these are even mentioned once and Isiah Thomas and Alan Inverson are given far too many pages. There is no consideration of what I suspect explains the dynamics of performance, which is the nature of sports teams as a social organization comparable to that of most Fortune 1000 firms and very much driven by short-term career, management and market pressures. Given that the univariate analysis shows that no figure explains more than a fraction of the variance in the linear regression analysis, then there simply must be a multivariate methodology for any conclusions to be drawn about the nature of the decision making process.
Perhaps I am incorrect in my assessment here; I have only limited skills in statistical analysis. The authors have strong resumes and the book thoroughly documents its sources and provides excellent footnotes. It's well presented. But all in all, it seems far too casual in its framing of a complex issue, doesn't add new insights and is overloaded with micro-level data that belabors the case that decision making in pro sports is largely screwed up. Read more ›

2 of 2 people found the following review helpful

3.0 out of 5 stars interesting but not enough for a book December 27, 2010

By bottomofthe9th

Format:Hardcover|Amazon Verified Purchase

Interesting in parts, but this is obviously a blog converted into a book, rather than content so substantive that it was suited to a book. Most of the inefficiencies discussed are relatively common knowledge by now among statistics-savvy fans, although I did think the analysis on how kickers' value comes more from their kickoffs than field goals was interesting and new.

1 of 1 people found the following review helpful

3.0 out of 5 stars Linke "Freakanomics" for sports fans August 19, 2010

By D. Greenbaum VINE™ VOICE

Format:Hardcover|Amazon Vine™ Review (What's this?)

"Stumbling On Wins" is a very entertaining book - it essentially takes a statistical, economic approach to sports in an attempt to answer common barroom debates such as which coaches are the best and do things like streaks and hot hands really exist.

This approach to sports isn't new - baseball has been stat-crazy for years, and "Moneyball" took it to the next level by providing (sometimes surprising) statistical value to various aspects of the sport. "Stumbling On Wins" expands the formula to other sports, such as pro basketball and football to answer questions about the real value of coaches and players.

The book is filled with nuggets of interesting data that have the potential to change how you look at sports. For example, in professional football, kickers really do matter (not terribly surprising) but what is surprising is that their skill on kickoffs is actually more valuable to a team then their field goal accuracy. Sound crazy? Well, there's stats to back it up.

Other areas of interest in the book cover things like the relative value of various pro athletes (underpaid and overpaid) which are always fun to debate, and especially for basketball fans, a baseball-level-geekery of analysis which is eye-opening to say the least (hint: rebounds are more important then you think.)

I would recommend this book to anyone with an interest in pro sports, from the casual fan to the fanatic. It is an enjoyable, fairly quick read and like the best books, it challenges how you think.

3 of 4 people found the following review helpful

3.0 out of 5 stars Not bad BUT... July 15, 2010

By t.g. randini VINE™ VOICE

Format:Hardcover|Amazon Vine™ Review (What's this?)

Most of the articles in here have merit and are interesting and are better written than Berri/Schmidt's first volume.

But...

Don't take their basketball value formula too seriously. Berri overrates the fetching of the ball after missed field goal attempts... wrongly assuming the correlation of rebounds to wins is causation.

Of course, rebounding is correlated to wins! What causes rebounds? Missed field goal attempts. What causes missed field goal attempts? Great individual and great help (team) defense. Unfortunately they do not relate defense-to-wins to any appreciable degree and thus greatly overvalue the ball fetching value of the Cambys of the league.

Dennis Rodman was an amazing player early on in his career mainly for his defense... as he grabbed more rebounds he got somewhat lazier on defense (to grab more rebounds) thus diminishing, not increasing his value per possession.

Someday these authors will realize this.

5 of 7 people found the following review helpful

1.0 out of 5 stars Stupid Regression Tricks August 1, 2010

By Aaron C. Brown TOP 500 REVIEWERVINE™ VOICE

Format:Hardcover|Amazon Vine™ Review (What's this?)

A book like this can do three good things. It can be fun to read, well-written and amusing. It can show deep love and understanding of a sport, giving insights to participants and fans. Or it can illustrate quantitative reasoning, finding non-obvious conclusion from data.

Few books do all three. Moneyball has brilliant writing and baseball insight, but no clever reasoning. The Bill James Handbook 2010 teaches important things about baseball and quantitative reasoning, but is weak on literary style. Mathletics: How Gamblers, Managers, and Sports Enthusiasts Use Mathematics in Baseball, Basketball, and Football is fun to read and a great way to learn reasoning, but the author's interest stops at the box score, he never goes to the field.

This book does none of the above. The writing is wooden and often unclear. There is no interest in the games. Worst of all, the arguments are deeply flawed.

The first extended example claims to prove that shooting percentage in basketball is a better indication of player talent than points scored. This is surprising on the surface. When players do things under reasonably standard conditions, kick points after touchdowns, shoot free throws or bat in baseball, with well-defined outcomes, their average performance is usually tracked. But when players do things under widely varying conditions, especially when their team picks the attempts, we use totals instead.... Thus total field goals made for a football kicker, but percentage of points after touchdown; total points scored in basketball, along with free-throw percentage; total RBI's for a baseball player, but batting average (or better, OPS). Totals are also good because they make adjustments for intangibles, for example a good defensive player in many sports will get higher offensive totals, because he is inserted for his defense. Or if a player has good teammates, he will benefit from less defensive attention, but lose out from fewer attempts. Averages can only measure specific skills, not contributions to the team.

To a first approximation, all players on a basketball team should have the same shooting percentage. If a player shoots better than his teammates, he should take more shots, including lower-probability ones, and attract more defensive attention; freeing up better opportunities with less defensive pressure for his teammates. His percentage should go down, and theirs should go up, until equilibrium is reached. The better shooter will show up by getting more points, but not a higher percentage.

This is not exactly true. A point guard, for example, can get a shot every possession, with little work or risk. Getting the ball to the center, with his hands above his waist near the basket, takes work and risks turnovers. So point guards need lower shooting percentages than centers. But the general principle applies, that you expect shooting percentage to vary according the team and a player's role, not according to his talent.

So how do the authors support their surprising contention? By regressing team wins on team shooting percentage. This is known as the ecological fallacy, attributing characteristics of a team on the players that make it up. Of course improving team shooting percentage is good, that's what happens when you put a better shooter in the game. But it is reflected in everyone's percentage, not just the good shooter's.

They compound the problem by validating the result in sample. They never test whether or not it is true, they just assert basketball fans and executives are stupid. I did a simple test. I looked at all NBA players since 1987 who changed teams or positions. For example, Jack Jarrett went from Portland's first backup point guard in 2007-08 to Indiana's starting shooting guard in 2008-09, replacing Marquis Daniels. If something is a skill inherent to a player, Jarrett's 2008-09 performance should be predicted by his 2007-08 performance; but if it results from the role on the team, it should be predicted by Marquis Daniels' 2007-08 performance.

Free-throw percentage is unambiguously player talent, there is no significant statistical effect of the role. The same is true of points. But field-goal percentages are predicted best by about equal combination of player and role. That makes sense, a better shooter will come in and raise the team's shooting percentage, including the percentage for his role, but not up to the full level of his talent, as he will take more shots, and more low-probability shots.

I don't claim this proves anything, it's just as sloppy as the authors' work. It does show that you need to test things, not just run a regression and declare yourself the smartest people in the world.

Another long portion examines racism in NFL quarterback selection. The method is to compare ten white Hall of Fame quarterbacks with four black quarterbacks who had similar per-game stats but shorter careers, only one of whom made it to the Hall. No one can seriously deny that the NFL discriminated against black quarterbacks for most of its existence, but this argument for it is fallacious due to selection bias. By selecting only successful (i.e. Hall of Fame) whites for the comparison group, the authors virtually guaranteed that the black quarterbacks would come up short. What you need to do is pre-select a group of players, based on criteria known before success is determined, and compare the experiences of whites versus blacks.

These are by no means the only errors. The problem with basketball shooting percentage (ecological fallacy, in-sample metrics, no testing) is repeated, and other errors are added, in the construction of "wins" indices for basketball and football. These underlie the arguments in more than half the book, rendering them worthless.

There are some good things about this book. The bibliography cites some excellent books and papers (unfortunately, many of the sound results are butchered by the authors' presentation). It is very short, 138 pages of text and if you don't count long lists barely mentioned in the text, it's probably under 100. The page size is small as well, suggesting the authors had trouble finding enough errors to fill a normal-sized book. Regression appears to be the only technique the authors know how to misuse. Finally, they confine their nonsense to sports. Imagine what would happen if they took their models to Wall Street (whoops, maybe they did). Read more ›