A Pitch F/X Look At Ryan Cook's Platoon Split

Last week, in this space, I used the great Baseball Savant Pitch F/X tool to gain insight into (if not answer) a simple yet important question: Why does A.J. Griffin allow such an extraordinarily high proportion of fly balls? Seeing as Griffin's tendency in this direction is a limitation on his effectiveness (if a non-drastic one, given the spacious confines of his home park, his outfield defense, and his other skills), breaking down what exactly was going on with it on a Pitch F/X level was somewhat illuminating.

This week, then, I'm going to do something similar: apply the Pitch F/X data from Baseball Savant to break down another significant weakness in another good A's pitcher.

Ryan Cook is good. He has a 2.38 ERA, and FIP says that's unlucky, pegging him at 2.27 (xFIP, ever the sourpuss, says he's at 3.78, but it was 1.39 too high last year, so what does it know, right?). He's struck out three times as many batters as he's walked, has a league-average groundball rate, and hasn't allowed a homer all season. On a stuff level, he looks the part, with the traditional late-game relief arsenal of a mid-90s fastball and mid-80s power slider.

But Ryan Cook has a problem. This year, lefthanded batters have hit .279/.380/.343 off him. His fellow righties, by contrast, are essentially hitting like pitchers against him, with a paltry .165/.200/.225 line.

Some of it might be weird BABIP luck, but there's no doubt Cook is far less effective to lefties than righties. In his career, he has a K/BB of 3.79 against righties--more than double his 1.86 to lefties. He makes up for this a bit by getting more grounders against southpaws--51.4% to 41.2%--but his weakness against lefties separates him from being a truly dominant relief pitcher.

So, what's causing this discrepancy? And what, if anything, might we be able to recommend Cook do to help minimize his platoon split?

First, let's look at his pitch selection against lefties and righties (NOTE: All data presented in this piece is 2013 data, not 2011-12). Cook throws four pitches: a four-seam fastball (92-98 mph), a two-seam fastball (92-98), a slider (79-87), and a changeup (85-88).

There are some obvious differences here. Cook is mostly a four-seam/slider pitcher to righthanders, occasionally mixing in a two-seamer, whereas he's mostly a four-seam/two-seam pitcher to lefties, only occasionally going offspeed to either the slider or the changeup.

There's a lot we can take away from that. Here, I want to quote an excerpt from the "Splits" entry in the awesome FanGraphs Library:

There has been lots of work done over the last couple years to determine if certain pitches are more effective against same-hand or opposite-hand batters. The general rule of thumb is that pitches that move horizontally (e.g. sliders) work best against same-handed hitters, while pitches that move vertically (e.g. curves, changeups) are most effective against opposite-handed hitters.

Four-seam fastballs: Small platoon split

Two-seam fastballs: Large platoon

Cutter: None

Sinkers: Medium platoon

Sliders: Large platoon

Changeups: Small reverse-platoon

Curveballs: Small reverse-platoon

In other words, pitchers should try and attack same-handed hitters with stuff like sliders, sinkers, and two-seam fastballs, while they should use changeups, curveballs, and cutters against opposite-handed batters. These are sweeping generalizations, though, and the optimal breakdown may vary depending on a pitcher’s movement and pitch repertoire.

We need to keep that last sentence squarely in mind, but setting that aside for a second, Cook's usage patterns to lefties don't make a ton of sense at first glance. Okay, he breaks out the changeup, which is clearly his least-trusted pitch, only to lefties and not to righties, which makes sense given the reverse-platoon tendencies of changeups, and he uses the slider a lot less to lefties as well, which also follows from the general consensus about pitch type splits. But he de-emphasizes the four-seam fastball for the two-seam fastball against lefthanders, which seems to be a very strange decision in this context. That emphasis on the two-seamer helps explain the increase in ground balls Cook gets against lefthanders, but it may also be a major part of his (relative) ineffectiveness to them. We'll test this out in a bit.

The other point I want to emphasize is that we can already see why Cook doesn't strike out as many lefties as righties--he doesn't throw them as many offspeed pitches. Against righthanders, he's at 37.8% usage of offspeed offerings (almost exclusively sliders), whereas he's at just 23.8% against southpaws. Offspeed pitches almost invariably generate more empty swings than heaters, so reducing them is likely to lead to a reduction in strikeouts.

Next, let's take a look at locations.

Okay, that's dizzying and scattered. Let's try something a little more useful:

Well, one thing jumps out here: Cook is out of the zone a lot more to lefthanders (65.4% of the time) than he is to righthanders (49%). Other than that, the actual height patterns seem fairly similar--he's in the middle of the zone the most, the upper third the least, and his out-of-zone pitches tend to be below the waist more than above it.

What about inside/outside?

Cook has no fear challenging righthanded batters on the inner half, but like most pitchers, he doesn't venture inside much to lefties. He isn't in the middle of the plate much, either, which is good, but now we've seen he's predictable in two fashions: he throws lefties a lot of fastballs, and they're usually away. That means lefties can basically wait him out--he's out of the zone to them almost twice as often as he's in it--taking pitches unless, on the off chance, he happens to miss his spot.

Here's a look at his overall outcomes against lefties and righties. They shouldn't surprise you:

So overall, we've established that Cook is an intimidating, dynamic two-pitch flamethrower against righties and a fairly meek one-note nibbler against southpaws. When you consider Cook's role, that does make sense to a degree--often, he may be pitching around a lefty to face a righty, just trying to avoid a game-blowing mistake. While his strategy to lefthanders seems almost mind-numbingly tentative when juxtaposed with his ruthless approach to righties, it does generally limit damage--lefties are slugging just .343 off Cook this year. After all, if he's throwing sinking stuff on or off the outer half, there just aren't going to be a whole lot of extra-base hit opportunities for the batter. Sure, if a manager sends three lefties up in a row, Cook might allow two walks and a sinlge on a bad day, but then the question is why Cook was left in to face three straight lefties in a high-leverage situation in the first place, not why he failed against them.

Against lefties, Cook throws balls 10.9% more--frankly, I'm surprised it's only 10.9%, given that he throws pitches outside of the strike zone 16.4% more often to them. In exchange, he gives up 3.5% swinging strikes, 2.6% called strikes, 1.4% fouls, and 3% fly balls (he also adds 0.7% grounders and subtracts 0.6% HBPs).

So that's his strategy, and I suppose it can be excused, though it can't really be embraced. But that leaves a question: Is Cook's approach born out of necessity or timidity? If he challenged lefties more, would he get hit hard, or is he just afraid to try?

We can get at this by examining his results on pitches inside the strike zone to both lefties and righties.

It does seem that Cook's effectiveness in the zone is significantly reduced against lefthanders. Called strikes and swinging strikes are the best results a pitcher can get, and Cook's getting them 39.1% of the time against righties on pitches in the zone, compared to just 27.5% against lefties. Notice that the groundball-to-flyball rate, which was better against lefties overall, is actually worse here, as well.

Speaking of "good strikes," as I call them (called and swinging strikes), one other thing I want to look at is Cook's pitch types to lefthanders and how they break down in terms of generating good strikes. Is his use of the two-seamer, for example, troubling?

Interesting. The two-seamer grades out slightly better than the four-seamer against both lefties and righties; perhaps Cook should turn it loose a bit more against his own kind, actually. Both fastballs grade out quite poorly against lefthanders, but the offspeed pitches actually show some real promise--on the whole, they're about twice as effective at generating non-contact strikes against southpaws as Cook's fastballs are. Some of that is due to the element of surprise, and some of it is due to Cook throwing so many of those pitches out of the strike zone (nine of the ten good strikes on the changeup were swinging, not looking), but I don't think it's absurd to wonder what would happen if Cook operated on 30% four-seamers, 30% two-seamers, 20% sliders, and 20% changeups (or some such combination) to southpaws. The offspeed offerings seem vicious enough to put some unease in lefthanded batters, something Cook's current approach seems largely unable to instill.

Ultimately, I think this analysis does provide some reasons to justify Cook's current approach to lefthanded batters and the results that approach generates, but it also offers other possibilities for him to consider that may (or may not) yield improved results against opposite-side hitters. Certainly, any strides he could make in this area would help him take a step forward into the game's elite relief pitchers and remove the one remaining chink in his armor.