According to classical economic theory, mature markets at equilibrium should reach a point of optimum efficiency. After a sufficient time, Adam Smith’s invisible hand ought to have slapped the inefficiencies out of any market. Surely, we could be so bold as to presume to apply such reasoning to professional baseball. Baseball is nothing, if not competitive; each side continually seeking out even the most meager of advantages. Do we have little more to learn about the stratagems of baseball? Do the myriad of conventional baseball statistics adequately and accurately quantify the contributions of players?

People have clutched on to various approaches to the game, but since the 1970’s Kansas native Bill James has made it a point to apply statistical analysis to assess the effectiveness of common baseball strategies. One of Bill James’s first heresies was the discovery that attempting base stealing rarely contributed to run scoring unless it was successful more than 70% of the time. Bill James is the founder of sabermetrics, “the search for objective knowledge about baseball.” (The term derives from an abbreviation for the name of the group that followed in James’s footsteps, the Society for American Baseball Research.)

James and others who have adopted his philosophy of critically applying statistics to baseball have developed radically different measures of player performance. For example, players have traditionally been evaluated on the basis of batting average, the number of hits divided by the number of at bats. Batting average is the dominant hitting statistic published in daily papers. Runs are the currency of baseball and sabermetricians have found other parameters that are far more correlated to a player’s contribution to runs scored. Getting on base, on base percentage, even by drawing a boring walk, is critically important. Indeed, the sum of on-base-percentage and slugging average (OPS)1 is far more highly correlated to the runs a team scores than is batting average. Hence, in seeking to evaluate baseball talent, in choosing young players to draft and nurture and others to trade for, OPS and other more exotic statistical measures like “runs created,” represent better criteria than simple batting average.

However, baseball ownership and management are conservative by temperament and generally grossly uneducated in statistics. James’s observations and the work of other sabermetricians were generally viewed as the oddball conclusions of baseball player wannabees, four-eyed, shallow-chested nerds who baseball players used to pick on in grade school. Sabermetricians might conjure up a clever insight now and then, but they really could not contribute to baseball strategy in any meaningful way. However, money has a way of shaking things up and big money shakes thinks up vigorously.

The growing ubiquity of computational capability has made the application of statistics to baseball easier. However, if the importance of statistical ways of looking at baseball players had been universally accepted, computing power would have been found. The real jolt into baseball has been a consequence of the high cost of baseball talent. In 1967, the average salary of a baseball player was $19,000 per year. It grew to $144,000 by 1980, $598,000 by 1990, and $1,900,000 by 2000. Even adjusted for inflation, this represents tremendous growth. Moreover, there is a 6-to-1 ratio in the payrolls of the wealthiest and poorest team. These facts place a high premium on judging and assessing baseball talent.

Thus begins the story of Billy Beane as told by Michael Lewis in his book Moneyball: The Art of Winning in an Unfair Game. As a young man, Billy Beane was a baseball scout’s dream. He had a decent batting average, but more importantly he was big, strong, well built, and fast. He looked like a scout’s image of a baseball player.

The New York Mets persuaded him, against his better judgment, to forgo a baseball scholarship to Stanford and join the Mets organization. Beane’s playing career fizzled primarily because he was too aggressive a batter, but he learned a few important lessons. One, baseball scouts were typically old baseball players who assessed talent as much on appearance as on numerical performance. They seem to all have the dream of discovering and molding the next great baseball talent. Two, Beane began to appreciate that really good hitters seemed to have an innate patience that allowed them to draw a lot of walks and only swing a pitches they can handle. Beane also observed that it is virtually impossible to teach patience, at least by the time players reach professional baseball. Three, it is not until players have played a number of years of college baseball that the sample size of plate appearances or innings pitched is sufficiently large that players could be reliably evaluated. Beane would rarely seek a player like himself right out of high school, even he looks like a baseball player.

After a mediocre career as a player, Beane moved to the front office of the Oakland A’s eventually rising to general manager. Beane was constrained by an ownership unwilling carry a large payroll and he made a virtue of this necessity. Embracing many of the insights of sabermetricians and hiring statistically competent Ivy League graduates to develop better statistical methods of player evaluation, Bean was able to acquire players who by conventional analysis and by perhaps by appearance were undervalued.

Beane uses batting statistics to estimate likely run production and derive dollar costs per run scored. It is then possible to numerically assess whether the acquisition of a particular player will reduce or increase the team’s net cost per run scored. Like an astute investor, Billy Beane exploits the inefficiencies created by traditional baseball measures to find players whose performance would likely exceed market expectations. When a player’s performance becomes apparent after playing with the A’s, his salary demands would grow larger than the A’s could afford and Beane would trade him off for undervalued players from other teams. After the middle of the season, some high payroll teams with slim playoff hopes would be looking to shed some salary, and Oakland could pick up bargain players.

It is estimated that the difference in talent between teams amounts to about a run a game. Luck adds perhaps four runs a game. The impact of the difference in talent between the teams is overwhelmed by luck in any particular game. However, over a 162 game season luck tends to average out and talent generally shines through. Once the playoffs begin, luck tends to dominate again. There are too few games, too small a sample size, for the best team to be assured victory in any particular series.

In 2002, Oakland posted the same 103 wins as the New York Yankees, a team boasting the largest payroll in baseball. There also were teams such as the New York Mets and the Baltimore Orioles carrying large payrolls that could not manage to win half of their games. In the American League West, there was an exactly inverse correlation of player payroll and performance with the parsimonious Oakland A’s leading the division and the high-spending Texas Rangers struggling with less than a 45% winning percentage.

Despite the conspicuous low-budget success of Oakland, other baseball teams have been slow to adopt Beane’s approach. Baseball Commissioner Bud Selig has tried to equalize the resources available to teams to prevent high-spending teams from perpetual dominance. The Oakland A’s are thus an embarrassment. Of course, as sabermetrics becomes more widely adopted, the differential in payroll will become more important. However, for now most of the baseball intelligentsia continue to regard Oakland as a fluke. Such intransigent owners and managers will be left behind as a more critical analysis of baseball takes hold. Beane remains elated at the remaining stubbornness, confident it will insure the continued availability of undervalued players.

1. On base percentage is the number of times a player safely makes it to base divided by the number of plate appearances. Slugging percentage is the number of total bases divided by the number of at bats. The term “percentage” is used when “fraction” is more accurate.

Leave a Reply

You must be logged in to post a comment.