Projections Evaluation - MLB.com Pitchers (3/18/08)
Here’s my review of the Pitchers’ projections available at MLB.com. I’ll use the benchmarks laid out in my 3/17 post on Pitchers.
- Total Innings Pitched (IP) for all Pitchers, all teams - can’t judge this; unfortunately, the MLB.com summary projections don’t include a Pitcher’s projected IP.
- Total IP for a team - see previous item.
- Total #Wins = 2,180, which is less than the benchmark of 2,430.
- The #Saves as a percentage of the #Wins (1,237/2,180) = 56.7%, which is greater than the range of 49-52%. If the #Wins were corrected (i.e. - Total #Wins set to 2,430), the #Saves as a percentage of the #Wins (1,237/2,430) would be 50.9%, which is within the benchmark range of 1,190-1,265.
- Total #Earned Runs allowed - see item #1.
- Overall Earned Run Average (ERA) - since IP for individual Pitchers aren’t readily available, I used Strikeouts as a proxy for IP. I realize this proxy leaves something to be desired, but even so, the resulting estimate of 4.03 is significantly lower than the benchmark of 4.25-4.50.
- Total Walks Allowed - can’t judge this easily, since the Walks Allowed are not included in the summary projections.
- Total Hits Allowed - can’t judge this easily either, since Hits Allowed are not included in the summary projections.
- Aggregate Walks+Hits per IP (WHIP) - again, using Strikeouts as a proxy for IP, the overall WHIP estimate is 1.33, which is below the benchmark of 1.37-1.41
Conclusion - MLB.com’s Pitchers’ projections are more difficult to generalize than their corresponding Hitters’ projections. Closers will be overvalued relative to Starting Pitchers because Saves are within the benchmark range, while Wins are understated. Both the ERA and WHIP appear to be low in aggregate. When viewed as a complete set, the MLB.com projections overstate the relative value of Hitters, while understating the relative value of Starting Pitchers, and, to a lesser extent, Closers.
Until next time,
The Sherpa
Tags: , MLB.com, Pitchers' projections, The Sherpa