Clemens vs. Carpenter
September 12, 2005 Leave a comment
Someone had an interesting idea on Baseball Primer today to look at Roger Clemens’ and Chris Carpenter’s expected W/L record based on how many runs their teams score for them and how many they allow; that is, how much better is Clemens’ actual record compared to his expected record, and how much better is Carpenter than he would be expected to go.
If he were a league average pitcher, Clemens’ expected W% would be .388 and his expected record would be 8-13 versus an actual record of 11-7. In other words, Clemens has 3 more wins and 6 less losses than one would expect. So that would make him +9.
If he were a league average pitcher, Carpenter would have an expected W% of .567, with an expected record of 14-11. He is actually 21-4, 7 wins better than expected and 7 losses better as well. That would put him at +14, or 5 wins better than the Rocket.
Obviously, some notable flaws exist in this study. First of all, I’m not sure there is any evidence that a pitcher has much influence over how much he wins or loses. Secondly, the less runs a team scores, the smaller an influence it has on its expected W%, so Clemens is put at a certain disadvantage in that sense. For example, with the Cardinals offense, Clemens would have an expected record of 16-2 vs. 10-8 (if the decisions for both are constrained at 18, Clemens’ actual number of decisions) for a league average pitcher, which would put him at +12.
That’s still not enough to pass Carpenter—and it’s the reason I think that Carpenter deserves the Cy Young—but the race is closer than the original study would indicate.