The Bolsheviks of baseball's information revolution are the three dozen staffers of Baseball Prospectus, a think tank-website that subjects the game's conventional wisdom to the cold light of sophisticated statistical scrutiny. Among the most valuable statistical insights put forth by Baseball Prospectus is the concept of replacement level. It's an economics principle that, as applied to baseball, quantifies the expected level of production a team will receive from a readily available, inexpensive player—picked up through free agency or waivers, promoted from Triple A or obtained at minimal cost from another club—when a starter is lost.
Baseball's talent distribution resembles a bell curve, with a handful of stars, a larger number of average players and an abundance of scrubs who are interchangeable and available to any team. "If any team can readily acquire a type of player at minimal cost, that player provides no competitive advantage," says Keith Woolner, a writer for BP. "The important question is, how much better is he than the freely available talent?"
To answer that question, Woolner, a software product manager in Cary, N.C., who holds a bachelor's in mathematics from MIT and a master's in decision analysis from Stanford, developed a metric called VORP (Value Over Replacement Player). Woolner outlines VORP as follows:
Define a baseline team with nine theoretical players at the league average for batting average (in the NL in 2003, .262), OBP (.327) and slugging (.417) and, using a model called Marginal Lineup Value, estimate how many runs one player, say leftfielder Barry Bonds (.341, .529, .749), would add to the baseline team's production when surrounded by the other eight players in the average league lineup.
Bonds raised an otherwise average team's expected production from 4.64 to 5.47 runs per game, a Marginal Lineup Value rate (MLVr) of .832; over a 162-game season Bonds would add 134.8 runs to an average team. Using MLVr as a basis, VORP then compares Bonds not with an average player, but with a scrub, accounting for playing time and position and considering, in addition to BA, OBP and SLG, stolen bases and caught-stealings and in essence answering this question: How many runs better were the Giants in 2003 because Bonds, and not a waiver pickup or minor league call-up, was in leftfield? VORP says 114.6 runs better.
[This article contains a table. Please see hardcopy of magazine or PDF.]