Baseball

This week pitchers and catchers started reporting for spring training in Arizona and Florida, marking the end of winter and the approach of a new baseball season. In preparation for the 2005 fantasy baseball season, I’ve been working on a new statistical tool to evaluate players. Now that spring training has started, it’s time to unveil the G-Coefficient.

The G-Co is steeped in the grand history of fantasy player evaluation pioneered by the A-Coefficient, and later the M-Coefficient. The G-Co attempts to evaluate a player’s relative value under a salary cap/points earned system of fantasy baseball. Essentially, the amount of salary cap space determines the expected points he must produce in order to be considered a “good value.” A series of calculations involving the players projected points for the season, his salary (based on CDM Budget Baseball values), and how he compares to other players generate a score (the G-Co) to evaluate the player’s value.

The key to this exercise is the accuracy of the projected points a player is expected to earn. For offensive players, I relied on the Bill James’ invented “Brock” projections tool. For pitchers, I made use of the “ZIPS” projections from Baseball Thinkfactory.com. Projections are, of course, fraught with hazard. Injuries, age, and playing time all affect the quality of projections, so the science is certainly inexact.

To test the G-Co’s success, I plan to use the actual player data for the 2005 season to determine if the G-Co would have generated the best fantasy team under the assumption of perfect projections. I will report the results in November.

Some rules of thumb for the G-Co:

• A score of 50 or better indicates excellent value.
• A score of 30 or less indicates poor value.
• A large number of players score in the 30-49 range, suggesting that small changes in the projections could have profound effects on a player’s relative value versus other players.