Another Look at James Shields and Hittability

By R.J. Anderson //

After the Rays clinched a playoff berth, my hope was to focus on the postseason rather than these final two series against inferior opponents. And yet, nothing like James Shields enduring a rough start to send me back into the regular season analysis for (presumably) one last post before the playoffs begin. I’m not going to copy and paste what I’ve said before because that does not advance the discussion. If you want my complete thoughts, read the old posts. Otherwise, let’s try a few different approaches with Shields.

Note: The data excludes Friday night’s start.

Counts

One of the popular opinions about Shields goes something like this: He becomes super-duper hittable when he falls behind. Does this pass the sniff test? Yes. Because Shields’ fastball is thought of as flat and becomes a fast-moving projectile whenever he throws it in a predictable manner. The league average pitcher sees his batting average on balls in play increase in counts where the batter has the advantage (i.e. more balls than strikes) and decrease in counts where he has the advantage (i.e. more strikes than balls). For his career, Shields is the opposite:

Split		BABIP Ahead	BABIP Even	BABIP Behind
Lg. Average P	.289		.296		.306
Shields Career	.316		.318		.292

Keep in mind that the average pitcher has a BABIP of .297 whereas Shields’ career BABIP is .315. Let’s adjust those numbers a bit by subtracting the BABIP in the above splits from their overall BABIP:

Split		BABIP Ahead	BABIP Even	BABIP Behind
Lg. Average P	-.008		-.001		.009
Shields Career	.001		.003		-.023

Where the average pitcher is seeing his BABIP drop, Shields is seeing his rise. And where the average pitcher is seeing his BABIP rise, Shields is seeing his drop drastically. This does not fit with what we would assume, because we know that Shields does indeed ride his fastball more heavily when behind. For as long as FanGraphs keeps track of pitch types by count, Shields’ highest percentages are usually when trailing in the count. Perhaps batters are simply hitting more home runs, or hitting the ball harder when Shields falls behind relative to the league average pitcher:

Split		ISO Behind	HR/PA% Behind	OPS Behind
Lg. Average P	.196		2.91%		.971
Shields Career	.251		4.52%		.975

That much is true, although one has to wonder how much of that ISO difference is tied to the added home runs. That the OPS are nearly identical despite Shields’ known reluctance towards walks is interesting, although I’m not sure what to make of it at this point, although this does lead me to the other area of interest.

Bases

At some point, whenever a pitcher consistently outperforms his peripherals, one must ask whether we’re missing something. Tom Glavine, for instance, is renowned as someone who pitched differently with runners on base. As previously noted, Shields seems more willing to give batters a pitch to hit in tricky situations than walk a batter. It’s a double-edged sword. Nobody wants a pitcher who gives up home runs, but nobody wants a pitcher who walks many batters either.

State		BB%		H%	XBH/H%	SO%	Non-SO Out%
Nobody on	4.46%		25%	35.4%		20.2%	49.3%
On first	4.63%		27.6%	39.1%		17.7%	49.4%
RISP		7.88%		23.1%	37.9%		19.4%	49.6%
First empty	10.3%		20.8%	40.9%		18.1%	50.8%

That’s a lot of data to digest. Shields is more willing to walk a batter with first base empty (and at least one other runner on) than in the other situations. As The Book tells us, that’s the proper situation to hand out a free pass. Meanwhile, batters are less likely to get a hit when first base is empty (indicating Shields might go into nibble mode) however, when they do get a hit, it is generally of the extra base variety.

Let’s extend this breakdown with a look at home runs allowed:

State		HR/PA%
Nobody on	3.2%
On first	4.1%
RISP		2.7%
First empty	2.7%

Perhaps there is the canary in the coal mine. Shields’ BABIP with a runner on first is the highest of any of his other base splits with at least 300 plate appearances. Not only does he allow more hits, but he allows more home runs. Here’s yet another look at base situations and their ISO:

Split		SLG		BA		ISO
RISP		.429		.261		.168
Empty		.430		.264		.166
Men on		.461		.276		.185
First Base	.500		.295		.205
Second Base	.382		.236		.146

That idea that the runner on first base situation results in Shields getting lit up shows up again.

Conclusion

So, the data says Shields gets killed when there’s a runner on first and that he’s actually pretty good when behind in the count. Truthfully, I’ve never noted whether he pitches from the stretch when there are runner(s) on bases other than first. Meaning, perhaps he struggles with the stretch. Defensive positioning might play into it too, but I cannot place any amount of confidence in that being the biggest (or second biggest) reasoning.

I’m not entirely certain what to pull from all of this other than Shields’ supposed hittability issues go deeper than him falling behind or being stubborn about giving up a free pass in certain situations.

This entry was posted in On the Field, Skill talk and tagged . Bookmark the permalink.

One Response to Another Look at James Shields and Hittability

  1. Pingback: Keeper Conundrums: Starting Pitchers, Part Two | FanGraphs Fantasy Baseball