Around The Yankees Galaxy - 9/18/10 - Let's recalculate WAR

Jscape continued the conversation about the validity of WAR on Thursday night, and we all debated an interesting point:

Since CC Sabathia is able to pitch deep into games, he winds up keeping inferior middle relievers out of the game longer, or out of the game altogether. 

Fangraphs says CC has been worth 4.3 WAR in 2010, but they calculate this by comparing him to a replacement level pitcher who's thrown the same number of innings.  I don't know how much sense that makes, because if you give a replacement level pitcher 31 starts, how many innings do you think he'd pitch?  150? 160? 

CC has thrown 217 this season.   

If Albert Pujols went down with an injury tomorrow, the #3 spot in the Cardinals lineup would still come up about as often as it did before.  And most relief pitchers tend to be brought into a game based on the score, the inning, and the save rule, rather than how good they actually are.  

However, a starting pitcher can pitch many innings, or few innings, based on how good he is at preventing the other team from scoring and how economical he is with his pitches.   Some pitchers are significantly better at combining both of these traits than others, and so I tried to find a better way to caculate how much a CC Sabathia or a Roy Halladay is really worth, considering that there is no replacement player who's going to average 7 innings per start.

First, I calculated each individual pitcher's winning percentage based on Bill James' Pythagorean winning percentage formula (if you're not familiar, you can read up on it here). 

Let's use AJ Burnett as an example.  His FIP is 4.60 this season, and the average team is scoring 4.41 runs in 2010.  Based on that run differential, if you put him on the mound for 9 innings and give him average run support, you'd expect to win 47.9% of those games.  Then we compare this to pitcher with a replacement level FIP, which I calculate to be 5.19 in 2010, and do the same calculation  You'd win 41.9% of those games.

Burnett has pitched 168 1/3 innings over 29 starts, an average of 5.80.  How many innings would a replacement pitcher average?  Well, I really don't know.  I couldn't find an MLB average for IP per start to base an estimate on, so I used 5.0, which I think is a reasonable.  (If anybody knows where I can find the actual number, please let me know.)

So multiply Burnett's actual number of starts by his average of 5.80 IP per start and then divide that total by 9 (because there are nine innings in a game), and then finally multiply by his Pythag winning percentage to get a total of about 9.  Repeat this calculation using Burnett's actual number of starts but the replacement's average IP and Pythag winning percentage, and you get 6.75.   

So, considering that Burnett will not only allow fewer runs than a replacement pitcher, he's also likely to pitch more innings per start, he's probably worth about 2.2 more wins than a replacement pitcher, at least by this calculation.  By comparison, Fangraphs has him at 1.3 WAR.

I did this calculation for every pitcher who currently qualifies for the ERA title.  Their adjusted WAR is bolded in the right-hand column, next to their Fangraphs WAR.

My methodology isn't perfect, but I do think it's headed in the right direction.  We already know that a starting pitcher who can pitch enough innings to qualify for the ERA title is worth a lot, because he's got to be very durable and at least half-decent when it comes to run prevention.  There simply are very, very few true "replacements" for somebody like that.   

So I don't know if Roy Halladay is really worth 10.6 more wins than a replacement, or if CC is worth 7.4 more, but I do know that there are only five pitchers who average 7 innings per start this season.  So they're probably worth more actual team wins than Fangraphs shows.