Download Moneyball The Art of Winning an Unfair Game Michael Lewis Books

Product details Paperback 320 pages Publisher W. W. Norton & Company; 1st edition (March 17, 2004) Language English ISBN-10 0393324818

Moneyball The Art of Winning an Unfair Game Michael Lewis Books Reviews

• Read this book only if you are prepared to realize that much of what you thought you knew about baseball is nonsense. This book is an amazing eye-opener about a then radical new way of managing a pro baseball team that allowed the dirt-poor Oakland A's to win as many games as the fat-cat NY Yankees. Using detailed statistical analysis created by baseball fans like Bill James who wanted to know how to make better teams in their fantasy baseball leagues, Oakland GM Billy Beane drafted or traded for players other teams considered sub-standard or worn-out and Oakland became a post-season threat despite having the second lowest payroll in the major leagues. Although the baseball establishment reacted with horror and contempt to having its time-honored methods of choosing players challenged, the approach used in Moneyball has been widely adopted by many teams including the Boston Red Sox who won the World Series shortly after doing so. Since reading this book I laugh every time I hear an announcer use the phrase, "productive out", knowing that over the long haul it's teams that don't trade outs for bases that win more games. The Moneyball approach remains controversial with many fans and baseball industry insiders--it's more fun to watch someone bunt a runner to second than it is to watch that hitter draw a walk--but the numbers show that over time the walks are more valuable to a team. Regardless of how much you agree with Bill James and Billy Beane, this is a terrific book that will make you really think about how the game of baseball works.
• Terrific read. For some reason, I didn't read it until 2016. It has a different perspective now after 13 years. We have already seen the affects of Moneyball and Sabermetrics on baseball. Theo Epstein seems to be the greatest beneficiary of the theory, the man who used sabermetrics and had a budget to spend. Two World Series victories, one for the Red Sox and another for the Cubs after many decades of losing. The Cubs didn't win the World Series in 2016; Theo Epstein was the winner. I liked reading Moneyball to follow the theories and careers of those who were the big players in the book. The afterward by Lewis gives a lot of insight into the reaction of baseball's old boy network, the old guard who rejected change.
• I decided to read this book after I got a nagging feeling that I am the only statistician in the US who hasn't read it. This review is aimed at those who are interested in quantitative performance evaluation or quantitative modeling in general.
  
  First, I'd like to thank Lewis for a book that is worth reading even for someone who has never learned the rules of the game and never will. Even if I had had a glimmer of interest in baseball per se before I read the book, it would have been extinguished for good anyway (read on to learn why).
  
  According to Lewis, for about a hundred years American baseball existed as a peculiar social club composed of GMs, scouts, baseball writers and commentators, and players who would often become one of the former. The first reason why The Club was special was that it was rarely run as a business. Consider Walter Haas, a former owner of Oakland Athletics. Haas was a great-grandnephew of Levi Strauss and coincidentally served as CEO of Levi Strauss & Co for almost 20 years. He was not shy of giving it back to the American society or at least to the most indispensable part of it, professional baseball players. As a result, in 1991 Oakland A's enjoyed the highest payroll in baseball while losing money for the owner. Unfortunately, that commendable manifestation of social responsibility was halted in 1995 when the new owners switched to a pro-business stance and slashed the payroll drastically.
  
  The new goal for the GM became to keep the team in the major league, while minimizing the ratio of payroll to the number of wins. The quest for how to run the most cost-effective team resulted in a number of fascinating discoveries. First, consider the process of drafting young players from high schools and colleges. The old scouts were used to some glaringly subjective player selection criteria, such as how cool the player's body looks ("selling jeans") or how concordant his facial features are with a great future in baseball ("The Good Face"). The "internal compass of an old baseball guy", directed them to look at the player and "imagine what he might become". They believed that they could judge and predict the player's performance simply by watching him. To be fair, the old school scouting worked fine but only assuming that the team's pockets were very deep. Indeed, that was often the case thanks to Walter Haas and other individuals who possessed similar amounts of wealth and social responsibility.
  
  Of course, different stats (foot speed, batting average, etc) were also considered, but none of the old guard had the desire, ability or means to assess how predictable such metrics are given the current information about the player. For instance, do young power hitters tend to develop precision over time or not? Are the stats of a high school player informative enough or does it make sense to ignore them and focus on the college players who have a longer history? Are there metrics that confuse skill and luck and contain no useful information even about the past performance, let alone the future?
  
  Further, suppose the future values of a measurable characteristic (e.g. foot speed, fastball velocity) can be predicted well, and we are fairly certain that luck and skill are not mixed up. So far so good, but exactly how much does that particular skill contribute to the ultimate objective, winning a game at the lowest cost? Say, if the future foot speed of the player is guaranteed to be as good as it has been, how much the team should be willing to pay for the player? If the answer is known, the GM can then perform some arbitrage by selling players whose characteristics are overpriced and buying those whose characteristics are underpriced.
  
  It may appear strange that those questions were not raised and addressed until after Billy Beane became the GM of Oakland Athletics in 1997. Apart from the philanthropic attitude of the team owners, there were other reasons why. Before the late 1970s, quantitative analysis of baseball data was both hard for the lack of computers and impractical because the players' salaries were not that high. Most importantly, the eventual adoption of quantitative performance evaluation methods revealed that the baseball data are extremely noisy. To extract some useful insights, one has to generalize from a large body of stats that no living scout or other baseball insider can keep in his head. Likewise, to see whether a given quantitative approach does (not) work, it's not enough to follow the career of a single player or observe the outcome of a single game.
  
  That nourished the perception that quantitative metrics are more or less useless and recoursing to one's "internal compass" works better when it's time to draft players and then decide how (not) to use them during the game. An example of such attitude is the scouts' belief that "if you see it once, it's there", which is a ringing endorsement of making a decision based on the sample size of one. Given how much the outcome of anything that happens on the field depends on luck, that's a kiss of death for good scouting, or at least for a cost-effective version of it.
  
  Undoubtedly, baseball teams vary in skill, but, according to the book, the contribution of skill to the outcome of an average game is about four times less than the contribution of luck. In particular, the number of games in playoffs is so small that luck doesn't cancel out in a series of five games, the worst team in the major league has a 15% chance to beat the best team. Because a team that loses in the playoffs cannot get the coveted World Series title it creates another psychological incentive for having little faith in the quantitative approach. To me it also means that if all of the noise were to disappear (and that's what "sabermetrics" is trying to achieve), the game would become far less fun to watch and talk about. There is not much thrill in observing how a slightly biased coin is flipped. Isn't that another reason why both the insiders and fans preferred to stick to the old ways instead of embracing the "performance scouting" and other insights offered by sabermetrics? If that's the case then The Club turns out to be far more rational and efficient than Lewis portrayed.
  
  Next, I'd like to explore the most important point made by Lewis bad as they may have been, the baseball statistics were probably far more accurate than anything used to measure the performance of people in many other lines of work, so ...? What immediately comes to my mind are the "scouting" (hiring) practices of Microsoft and Google that included the so-called brainteaser interviews that were administered for 10 to 20 years (see Are You Smart Enough to Work at Google? and How Would You Move Mount Fuji?). To me it has always been clear that being good at brainteasers does not predict one's future job performance unless the job amounts to solving brainteasers. Both companies banned the practice in the end, even though some copycats are still lurking around. The point, however, is why it lasted for as long as it did. Apparently, the engineers at Microsoft and Google are no fools. They figured out that the brainteasers fail at achieving their purported objective many years ago. Why continue to use them, then? Moneyball provides a plausible explanation. Companies are not only about doing business but sometimes about being a certain social club. The club entry requirements can be as arbitrary as anything. If they appear, to put it mildly, irrational, that is to be expected every now and then. Therefore, the real question is not "Are you smart enough to work at Google?" but "Are you a good fit for the Google social club?". Regardless of what the answer is, it has nothing to do with one's intelligence or the absence thereof.
  
  Now back to baseball. If you have some quantitative background I could highly recommend "Stein's Paradox in Statistics", a paper published by Prof. Efron in 1977 (incidentally, that was the year Bill James published his first Baseball Abstract). The rather ancient article is quite in line with what I learned from Moneyball the amount of noise in the baseball stats is substantial and the players are a lot more similar to each other than it appears to the public and even the insiders. It looks like my background in Statistics, combined with what I learned from the book, made it impossible to enjoy the game the way millions of fans do. Nevertheless, I managed to derive a lot of value from Moneyball and I wish you the same.
• Very interesting book that flies by. It is so easy to read that you can almost pardon's Michael Lewis' penchant for perpetuating the hyperbole of others and contributing exaggeration of his own. You can easily see how the added flair and drama makes the book more interesting for a mass audience. You can also excuse the fact that the book doesn't actually point out the answer to the original question. Why were the Oakland A's able to win so many games on a small budget? It turns out that was a little bit of skill but mostly luck. They got lucky with their pre-moneyball pitchers Zito, Mulder and Hudson.
  
  Despite the fact that Lewis doesn't highlight this anti-climactic conclusion on his own, the book still has enough information for the reader to get there. It is ironic that this parallels the reason Bill James' departed from Baseball Abstract and that is mentioned in the book. There is too much focus on the details that don't matter. The high level narrative and explanation are more important. Budget matters. You can get a slight advantage by exploiting inefficiencies but this takes a long time to prove out due to limited data points the out-sized impact luck has. There are lots of little nuggets of information like that hiding in this book that make it worth reading.