Using Ball in Play Data to Identify a Sleeper

I spend an inordinate amount of time playing with splits. It’s hard to come up with two new topics to write about each week, so I export a lot of data to Excel and click and sort around until I find something interesting. During the relatively boring late slate of football games yesterday afternoon, I was looking at ball in play data.

It occurred to me that there should be a pretty strong correlation between ERA on balls in play and OPS allowed on balls in play (there is). It then occurred to me that guys who have a relative ERA that’s much lower than their relative OPS might be getting a little lucky. When I say relative, I just calculated each pitcher’s z-score for both statistics, which were adjusted for number of batters faced since they are rate stats. Here are those potentially lucky pitchers.

Player	ERA	ERAz	OPS	OPSz	OPS-ERA
Corey Kluber	2.46	0.86	0.725	-1.08	-1.935
Cole Hamels	2.04	1.37	0.692	-0.36	-1.729
Mark Buehrle	2.94	0.07	0.735	-1.46	-1.527
Jon Lester	2.18	1.32	0.681	-0.18	-1.504
Zack Greinke	2.47	0.75	0.709	-0.67	-1.422

I said “potentially lucky” before the chart because there’s another factor that needs to be taken into account before calling these guys lucky. Here’s the chart again with that other factor included.

Player	ERA	ERAz	OPS	OPSz	OPS-ERA	K%
Corey Kluber	2.46	0.86	0.725	-1.08	-1.935	27.5%
Cole Hamels	2.04	1.37	0.692	-0.36	-1.729	24.2%
Mark Buehrle	2.94	0.07	0.735	-1.46	-1.527	13.1%
Jon Lester	2.18	1.32	0.681	-0.18	-1.504	24.9%
Zack Greinke	2.47	0.75	0.709	-0.67	-1.422	25.4%

Of course it makes sense that guys who can regularly get guys out without the ball being put in play via the strikeout would have a better relative ERA. Strikeouts mean fewer balls in play and the ability to strand runners that hit their way into scoring position more often than their ERA might indicate.

The problem here is that we really don’t get any useful fantasy knowledge out of this. Kluber, Hamels, Lester and Greinke are good and not lucky. Breaking news! And Buehrle is a regression candidate. That’s also old news. Podhorzer told us that in May. A look at his monthly splits also tells us the regression has been on since July. But what about the flip side? Has anyone been unlucky?

Player	ERA	ERAz	OPS	OPSz	OPS-ERA	K%
Jake Odorizzi	3.75	-1.00	0.661	0.18	1.174	24.4%
Hector Noesi	3.48	-0.73	0.647	0.44	1.178	16.6%
Shelby Miller	2.83	0.23	0.587	1.59	1.360	16.1%
Kyle Gibson	3.74	-1.19	0.651	0.40	1.589	13.3%

Ah, there we go. One of these is not like the other. These are the guys who have an OPS allowed that is much better than their ERA relative to the other pitchers in the sample, which is starters with 100+ IP this season. The premise is that these guys might have had a little bad luck. But Noesi, Miller and Gibson don’t have the ability to get strikeouts very frequently, which makes it harder to strand runners.

But Odorizzi’s ERA is exactly one standard deviation below the mean despite his OPS allowed being slightly above average. And he has the ability to miss bats, which should be helping him have a relative ERA that is lower than his relative OPS allowed. I want to point out that Odorizzi’s slightly above average home run rate doesn’t factor into his ERA here because this is just ERA on balls in play. Home runs are’t considered ‘balls in play” so that issue isn’t a problem in this specific exercise.

With an ERA of almost 4.00 for the year and an ownership percentage of 29.2% on ESPN.com, it’s doubtful that Odorizzi will be all that expensive come March. But his ERA seems to have some room to improve, and it would pair nicely with an already solid strikeout rate. Be the person to spend a buck (or a few if necessary) to roster him next draft season.