We commonly evaluate pitchers by the number of so-called “earned” runs they allow (ERA). That said, many of you know that a pitcher’s ERA at any given time is a poor, or at least misleading indicator of either their true ability or likely success going forward.
There are many reasons for this. One of them is the nature of the “earned” run, which depends on the whim of the official scorer and these days, the vehemence with which the player later protests the decision. Another is that pitchers have a hard time controlling the results of balls which are put into play (BABIP), as hitters occasionally “put them where they ain’t,” and even more frustratingly, bloop them over where the fielders are.
This is one reason why knowledgeable fans monitor Fielding Independent Pitching (FIP), the metric that rates how well pitchers manage the events largely within their control: strikeouts, walks, and home runs. FIP correlates better with future ERA than past ERA, partly because strikeouts and walks affect scoring, and partly because high-strikeout pitchers induce weaker contact and strand more runners on base.
But that doesn’t help us with the problem of balls in play. Having one pitch put into play by itself is not a big deal. The problem is when those hits cluster together, resulting in runs. No pitching staff wants to give up nine doubles, but if those doubles are distributed one per inning, the opposing team will strand every one of those runners, and the pitching staff will look unfairly dominant.
Not surprisingly, hits do not cluster together the same way for all pitchers in all appearances. Some pitchers will constantly work themselves out of jams, while others can throw six shutout innings and be dinked-and-dunked to death in the seventh, making it a good overall outing look subpar.
How do we evaluate the runs a pitcher should have given up with the walks, hits, and home runs they actually allowed? Fortunately, Dave Smyth did that work for us with the BaseRuns formula, which is often described as assessing the “cluster luck” of a pitcher or a team. I don’t understand calculus, but Dave does, and he’s let us know that a pitcher’s expected runs allowed can be estimated through BaseRuns, the formula for which is in the appendix below.
Even better, a poster on the Fangraphs Community forum recently proposed a method to normalize the results of the formula and determine the amount of positive or negative “luck” that a pitcher has experienced in a season. It’s such a good idea that I thought I would borrow it.
So, through last Saturday, here are the BaseRuns “cluster luck” figures for Brewers starters in 2014:
|Starter||IP||Actual R/9||Projected R/9||Luck|
The columns, in order, list the pitcher, their innings pitched, their actual runs allowed per nine innings, their “projected” runs per nine innings (what they should have allowed, according to BaseRuns), and finally the positive or negative amount by which they outperformed their expected runs, which I’ve deemed “luck,” for lack of a better term.
Every pitcher is going to over- or underperform their run expectation a bit, so we needn’t be concerned about players with small luck factors, such as Kyle Lohse (3% unlucky), or Wily Peralta (4% luckier than expected). We can be a bit more concerned about Yovani Gallardo (7% better than deserved). Then again, Brewers fans are perpetually concerned about Yovani Gallardo, so that is nothing new. Marco Estrada’s performance should have been even worse than it was (6% luck factor), and Jimmy Nelson, in an extremely small sample predating Tuesday night’s performance, had been both (1) pretty terrible and (2) fortunate it was not worse.
But look at poor Matt Garza. By BaseRuns, even including his fractional-inning disaster against the Nationals, Garza should be allowing runs at the exact same rate as Kyle Lohse (3.60 runs /9 innings); instead he’s been stuck with 4.78 runs / 9 innings, which is almost an additional run for each seven-inning start. This has been frustrating to watch, but would seem to bode well down the stretch. Provided Garza remains healthy, the Brewers have essentially gotten a second Kyle Lohse through the Garza signing. I can’t see why any knowledgeable fan should be disappointed with that.
Briefly, let’s look at the cluster luck for Brewers relievers:
|Reliever||IP||Actual R/9||Projected R/9||Luck|
Because relievers pitch far fewer innings, their luck factors are larger, and we need to treat these ratings with some caution. Nonetheless, we can see that Tom Gorzelanny has lucky so far, giving up zero runs when he should have given up 2.27 runs / 9. (Mind you, this is still quite good.) We also see that Wei-Chung Wang was a bit unlucky (-14%), but terrible regardless; that Will Smith and Zach Duke have had some favorable cluster luck, while Rob Wooten was a bit unlucky; and that Francisco Rodriguez is pretty much just an average reliever at this point.
But the alarming pitcher on that chart should be Brandon Kintzler. For a reliever who generally pitches an inning or so, Kintzler has not been particularly good this year, allowing 3.71 runs / 9 innings. More importantly, his luck factor of 60% is off the charts: he should have allowed 6.39 runs per 9 innings, and we’re not even getting into the runners he inherited and stuck on somebody else’s tab. Kintzler doesn’t strike guys out, and his only real value is as a sinker-baller who gets guys out on balls in play. Perhaps it is unfair to pick on Kintzler for a 34-inning sample, particularly when he in fact did minimize the damage from those hits. On the other hand, Kintzler basically has one job, with the notable exception of last night, he’s been getting shelled. If the Brewers do manage to get Jim Henderson back soon, or trade for a new eighth-inning guy, Kintzler almost certainly should be the reliever on the bubble.
Follow Jonathan on Twitter @bachlaw.
All statistics from Fangraphs; calculations by me.
The formula for BaseRuns:
BsR = A*B/(B + C) + D
For pitchers, the variables are defined as follows:
A = H + W – HR
B = (1.4*TBe – .6*H – 3*HR + .1*W)*1.1
C = 3*IP
Where TBe = 1.12*H + 4*HR