I have to admit, it wasn't until all the Twitter feeds came pouring into yesterday's live chat that I realized just how bad Phil Coke has been of late: 17 ER over his last 15.1 IP. His ERA has ballooned from 2.97 to 5.05 in that stretch.
Some may want to blame this on his over-use, as his 59 appearances are the most on the team by a good margin (Rivera is next with 52), and his 51.2 IP in relief is third behind Alfredo Aceves (61) and Rivera (53). And maybe that has something to do with it, but not very much.
Most of what's happened to Coke over these last 20 outings is just good old statistical correction. As I've stated before, particularly when evaluating relievers, I prefer to look at Fielding Independent Pitching (FIP) rather than ERA. As you may recall, FIP is dependent upon the three things pitchers can control: strikeouts, walks, and home runs, and is then is adjusted to an ERA-like scale.
Unfortunately, I can't find game-by-game logs of Coke's FIP, nor do I have the free time to calculate it at the moment. However, at every point this season that I looked at his numbers, his ERA was outperforming his FIP - by a lot. These past twenty outings have served to correct that gap, so much so that after yesterday, Coke's ERA (5.05) is now worse than his FIP (4.86).
So what's the good news/bad news here? The good news is that the numbers seem to indicate that Coke has at worst leveled off, at best is due for a small improvement. The bad news is that where those numbers stand right are not all that good. The good news is the Coke's strikeout (7.14 per 9) and walk (3.14 per 9) rates are slightly better than the league averages. The bad news is his home run rate (1.57 per 9) is a half home run worse than the league average, and that's what is killing both his FIP and his ERA.
Another number to look at is Coke's batting average on balls in play (BABIP). Evidence has shown that a pitcher has little control over balls in play (which is why FIP is a valuable statistic) and that most pitchers end up having a BABIP close to the league average. Currently, Coke is at .234, far better than the league average of .304. This could suggest that we haven't seen the bottom for Coke yet; if his BABIP regresses to the mean over the final five weeks he could be in for a few more rough outings.
However, given Coke's K and HR rates, I would expect his BABIP to be a bit low. His K rate is higher than the league average, meaning there are fewer balls in play against him. Furthermore, 6.1% of batted balls off Coke are home runs, as opposed to 4.0% for the league. As such, hitters are making contact less against him compared to the league, but when they do, the ball is traveling over the fence far more often. While the latter certainly isn't a good thing, I do think that indicates that Coke's BABIP likely won't get close to the league average by season's end. Besides, any BABIP regression to the mean may be a good thing for Coke, as it could indicate that his gigantic HR rate is dropping off a bit.
So what does it all mean? Phil Coke isn't as good as he appeared to be through most of the summer and isn't as bad as he appears to be right now. He gives up way too many home runs, and Yankee Stadium may have something to do with that (18.2 AB/HR at home, 25 on the road). If he can get his home runs down, his K and BB rates suggest he is a relatively effective pitcher.
Relievers are highly volatile due to the relatively small number of innings the pitch. One or two bad outings, like yesterday or Coke's 0.1 IP 6 ER disaster in Chicago on 8/1, can have a major impact on a reliever's statistics. Despite being a former starter with a decent arsenal of pitches, and despite his numbers being good against right handed batters for most of the season, Joe Girardi has insisted upon using Coke as a match-up lefty for most of the year, with 30 of his 59 appearances lasting less than an inning.
Coke has been pretty high up in the bullpen pecking order for most of the season. He'll need to show some improvement over these last five weeks to justify keeping that status in October. If he can keep the ball in the ballpark more often, he has a good chance at making those improvements.