Since splits happened over at FanGraphs, I can’t help but get excited. It’s been a lot of fun projecting these splits, and the practical use of these isn’t half bad either. That having been said, I think Marlins fans would be interested how the team’s players are projected to stack up versus lefties and righties, so I went ahead and figured out the projected platoon splits for each player based on their career splits and the league average. After the jump, we go over the rules and the splits.
The Rules (as written by The Book)
The Book handled platoon splits pretty well, identifying them as a real effect on pitchers and hitters. For pitchers, it was much easier to establish a platoon split, i.e. it was regressed a lot less to the mean split. For pitchers, it took righties 700 PA of facing lefty hitters to regress 50%, while it took only 450 PA for lefties facing lefties. For hitters, however, the regression was a bit more harsh: left-handed hitters facing lefty pitchers had to be regressed 1000 PA for 50% to the mean, while righties facing lefty pitchers had to be regressed 2000 PA to the mean.
Now what means am I regressing to? I’m stealing a line from FanGraphs writer and friend of the Maniac Matt Klaassen from his piece on estimating hitter platoon skill. Matt calculates the league average wOBA split for lefties versus left-handed pitching and righties versus left-handed pitching, arriving at a split of 8.6% for lefties vs. lefties and 6.1% for righties vs. lefties. Given the matchups, this means that on average, lefties had a split between their lefty and righty wOBA that was 8.6% of their actual wOBA, worse against lefties. Meanwhile, righties had a split between their lefty and righty wOBA that was 6.1% of their actual wOBA, better versus lefties.
Two quick examples
Let’s take a look at the weird example of Dan Uggla. You always hear Rich Walz and Tommy Hutton describe Uggla’s reverse platoon split (career .328 wOBA versus lefties, career .360 wOBA versus righties). But that split has only occurred in 638 PA against lefties, essentially one full season for Uggla. What can we expect as of now? Well, the career split for Uggla is a drastic 9.0% and in a reverse direction. But after regressing 2000 PA to the mean of 6.1% in the correct direction, we get a projected split for 2010 of 2.4% better against lefties. Applying that to Uggla’s CHONE-projected wOBA of .355 for the season and you get a projected split of .362 wOBA versus lefties and .353 wOBA versus righties. In other words, you would expect to be safe to put Uggla out versus either handed pitching.
On the other hand, let’s examine the splits for Cody Ross. Ross exhibits your classic split to the extreme: in 504 PA versus lefties, he has a massive .403 wOBA, while in 1143 PA versus righties, he has a diminutive .319 wOBA, giving him a split of 24.2%! But after the appropriate regression, we see a diminishing of that gigantic effect, down to a projected 9.8% split. That gives Cody a projected .362 wOBA versus lefties and .329 wOBA versus righties. In other words, he’s around average versus right handers, but he’s akin to Dan Uggla versus lefties.
Here are the rest
In table form, of course.
|Name||Hand||wOBA vs. LHP||wOBA vs. RHP|
Some interesting notes from these results:
- I chose to do only players I expect to see a lot of playing time and have played at least two years. So nothing on Chris Coghlan yet.
- Hanley Ramirez is really good.
- The duo of John Baker and Ronny Paulino worked out very well for the Marlins last year, and it looks like it may do so once more this season. Check out Paulino’s performance versus lefties; it’s very similar to what we saw last season, when you account that he had a more even split between facing lefties and righties. Keeping Baker away from lefties may also be a good call. Expect almost three wins from these guys again this year, barring a disaster from Baker’s defense.
Keep these numbers in mind next time you watch these guys come to the plate, which thankfully is coming soon. Pitchers and catchers report in four days! Get excited.