FanGraphs unveiled splits to the world, giving us yet another reason to spend countless..."/> FanGraphs unveiled splits to the world, giving us yet another reason to spend countless..."/>

Marlins’ relievers and a warning about splits


At the beginning of the week, FanGraphs unveiled splits to the world, giving us yet another reason to spend countless hours on the site (not that I already do that or anything). With this new brand of info, I’ll no longer have to dig up wOBA weights to calculate wOBA for player splits. Neither will I have to use OPS as an approximation to overall hitting talent. The possibilities are boundless for tiny research done by the likes of me.

I figured today I’d start looking at some stuff regarding splits, and the first item on the menu in my mind was lefty/righty splits for bullpen pitchers. Since guys in the pen are leveraged better in terms of platooning, I figured it would be worth a look to see who the Marlins should and shouldn’t use against righties and lefties.

A Quick Warning

Sure, I could check the career splits for each of our relievers and leave it at that, but that would likely be inaccurate. As with all stats, what we observe is not always what we should expect in terms of true talent. Remember, observed results is true talent plus some luck. Thus, as with our seasonal projections, we need to do a projection of the splits based on past performance and a league average. Dave Cameron of FanGraphs goes over this a bit more here.

An Example

Let’s start by taking a look at the Marlins’ much-maligned Renyel Pinto. In his career, Pinto has put up a 4.81 FIP against lefties in 351 batters faced and a 4.75 FIP against righties in 579 batters faced (not including IBB). This displays essentially an even split, but we can’t really just trust that at face value. Instead, we’d have to regress this to the mean.

In order to do this, I took a three-year weighted average (5/4/3) of the FIP differentials for left-handed pitchers versus lefties and righties from 2007 to 2009. I changed it so that I took the run values of the FIP components above the run value of the ball in play (so that balls in play would be measured at 0 runs above average) and took them out of batters faced/PA instead of innings pitched. Here are the components’ values in runs above/below average:

HR: 1.44
BB: 0.34
K: -0.23

OK, so for Pinto, you would get a runs/PA compared to average of 0.039 runs per PA versus lefties and 0.042 runs per PA versus righties, a split of -0.0029 runs per PA measured as lefties runs minus righties runs. But the league average split for lefties is -0.0186 runs per PA measured in the same way. The Book, a source which I often cite thanks to its thorough research, says that at 450 PA versus lefties, you can regress a left hander 50% to the mean. Pinto has had 351 PA versus lefties, meaning that we should regress him 56.2% to the mean. Doing such a regression gives a projected split between performance against lefties and righties of -0.0117 runs per PA compared to the average value of a ball in play (remember, for pitchers, negative runs are good). For an idea of how meaningful that is, if you assume the average value of PA per inning of around 4.29 PA/inning, you get a difference of 0.45 in terms of FIP!

What does that mean for this season? Well, let’s use CHONE’s projections as an example. Pinto is projected at six homers allowed, 37 walks and hit by pitches, an 54 strikeouts in 60 innings. Those 60 innings should translate to around 258 batters faced, giving a projected 0.034 runs per PA above the value of a BIP. Assuming a similar distribution of batters faced in his career, Pinto should face lefties in around 38% of his BF and righties around 62%. Given that distribution, what could we expect as Pinto’s performance versus lefties? Doing this math:

0.034 = 0.38*x + 0.62*(x + 0.0117); where x = runs per PA versus lefties

gives you a value of 0.0267 runs per PA versus lefties, compared to 0.0381 runs per PA versus righties. To get an idea of what that means in FIP, that’s approximately a 4.23 FIP versus lefties and a 4.67 FIP versus righties, assuming average PA/IP.

The following is a table containing the rest of the splits of the expected bullpen cast. I’ll show it in terms of both runs per PA and FIP.

NameHandRuns/PA v LHBFIP v LHBRuns/PA v RHBFIP v RHB
Renyel PintoLHP0.02684.240.03854.69
Leo NunezRHP0.03504.550.02394.12
Burke BadenhopRHP0.03344.490.007213.48
Dan MeyerLHP0.01283.700.03104.40
Brian SanchesRHP0.02764.270.02254.07

Some interesting things from those numbers:

– If you wanted a good reason not to start Burke Badenhop, there’s your reason. He displays the highest projected split among all the pitchers the team has. Primarily as a reliever, the Hopper has been league average versus lefties, which is near replacement level for a reliever. One would imagine it would only get worse for Badenhop as a starter.

– Maybe Brian Sanches should be our closer? Not only does he project as a better pitcher than Leo Nunez (FIP of 4.10 for Sanches versus 4.26 for Nunez), but he does not display the split differential that Nunez displays.

Dan Meyer exhibits similar platoon splits as Badenhop. He may not be the best choice for sending out in the eighth inning as setup. Sanches may be the best man for the job.

– Pinto may be as good against same-handed pitching as Sanches is against opposite-handed pitching. I find this funny.

Keep these splits in mind when we watch Fredi Gonzalez manage the pen. I have a strong feeling like he’s going to mess it all up anyway.