Spotting Free-Agent Bargains in Baseball

SMITH BRAIN TRUST — By using advanced baseball statistics, it's possible to predict the monetary value of baseball players' future contributions — and avoid awarding massive multiyear contracts to players whose best days are behind them, according to new research from the Robert H. Smith School, at the University of Maryland.        

Sean L. Barnes and Margrét V. Bjarnadóttir, assistant professors in the Smith School's department of decision, operations and information technologies, examined the performance of impending free agent major league players from 1998 to 2014. They matched the players' statistics against publicly available data about their free-agent contracts. (Players tend to be eligible for free agency, a fully free market for their skills, after six years in the big leagues, although the authors also included players signed to multi-year extensions before they were technically free agents.) To measure performance, Barnes and Bjarnadóttir focused especially on Wins Above Replacement (WAR), a metric developed by statisticians that for non-pitchers combines several measures of offensive, defensive, and base running contributions. For pitchers, WAR factors runs allowed by the pitcher and innings pitched, and corrects for such things as the quality of the pitcher’s defense and park effects. It then rates the player's contributions relative to an idealized average player.

Barnes and Bjarnadóttir created two models: One was designed to estimate the yearly salary that a player would attract on the free agent market. This model included past performance but also such off-the-field factors as previous salary. The second model looked purely at predicted performance, translated from baseball statistics into dollars. Tested on a subsample of the players in the study, the two models were able to "predict" who would overperform and underperform their market value.

The tools in the paper can be used on current and future free agents, and potentially offer a way for small- and medium-budget teams to hold their own against major market titans like the New York Yankees. Payroll matters, the data showed, but the difficulty of predicting future player performance meant it was far from determinative.

"Teams often pay players based on their past performance or what the market says they are worth," Barnes said. "And the market is likely to overpay players. What we were getting at was the goal of paying players for their expected future performance."

Barnes and Bjarnadóttir singled out the best and worst contracts during the period studied.  For everyday position players, Ichiro Suzuki, signed by the Seattle Mariners in 2003 for $13.9 million annually, produced the most "surplus value": The authors estimated that his contribution to the team was worth $22.6 million per year. (The model had estimated that a team would pay him $12.8 million, not far from the mark.) Among pitchers, Chris Sale, signed by the Chicago White Sox in 2012 at $6.6 million annually, provided $20.4 million in performance to the team per year during his contract, making him another bargain.

The biggest contract flops will be familiar to baseball watchers: Among hitters, Ryan Howard, the first baseman whom the Philadelphia Phillies paid $25.8 million a year starting in 2011, was in fact worth only $6.2 million annually based on his projected performance, the authors found. Among pitchers, the Pittsburgh Pirates' A.J. Burnett had the worst price-performance ratio. Signed in 2013 to a deal worth $11.2 annually, his contribution to the team was valued at -$0.4 million yearly (relative to an average substitute).

More surprising, especially to Yankees fans, was Derek Jeter's place at No. 2 in the most-overpaid position-player category. Paid $17.9 million annually, starting in 2010, his performance was valued at only $6.7 million. "You can't find one Yankees fan who thinks he was overpaid, because of all of the championships he won, but according to the key performance statistic, WAR, he does not fare well," Barnes said.

In general, the data showed that mega-contracts were a bad idea from an efficiency perspective. Of players with $100-million-plus contracts, only two — the outlandishly talented pitcher Randy Johnson, signed by the Seattle Mariners in 1999, and the prodigious hitter Albert Pujols, signed by the St. Louis Cardinals in 2004— produced numbers that made such outlays worthwhile. The data also suggested wariness about big contracts for 30-something players. Overperforming players tended to be younger players on the verge of breakout careers.

Only one player signed when he was older than 30 made the Top 10 high-value list for position players or pitchers: David Wells, a hefty journeyman signed by the Yankees for his year-40 season, for a paltry $1.6 million per year. He went on to produce $13.1 million in value that year.  

Teams are paying attention to advanced statistics, the study found: On the open market, simple numbers like batting average and home runs did not affect players' salaries nearly as much as WAR, which account for a player’s overall contributions.  Teams' back offices have studied their Moneyball, it is clear, but the new paper takes the statistical tools available to baseball teams to a new level.

The researchers did not examine the current crop of free agents, but Barnes, a Cubs fan, says he has high hopes for second basemen Ben Zobrist (signed by the Cubs to a four-year, $56 million deal) and right fielder Jason Heyward (8 years, $184 million), despite Heward's mega-contract. "Neither has an especially high batting average, and neither hits an exceptional number of home runs, but both have been in the Top Ten in WAR for the last few years," he says.

READ MORE
"Great Expectations: An Analysis of Major League Baseball Free Agent Performance," by Sean L. Barnes and Margrét V. Bjarnadóttir, has been provisionally accepted at Statistical Analysis and Data Mining.