Mixing Pitches, Situationally

There are a couple of ways to go about doing research, and one of them is preferred over the other. In the preferred course, you identify something of interest, and then you go into the numbers. The other way is the other way, where you go into the numbers and hunt around for something of interest. The former is a lot more targeted, but one could say the latter is a lot more open-minded. For this post, I started with the numbers and went from there.

I recently wrote about trusting catchers with runners on third base. In the course of doing that research, I noticed Jaime Garcia had a much higher rate of low off-speed pitches with a runner 90 feet away, as compared to with the bases empty. I didn’t really go any further in that piece, but I was interested in a follow-up that was built around guys who pitch differently by situation. It stands to reason every pitcher is different and every catcher is different, so not every battery will take the same approaches. This is a starting point: This is a post about fastball rates, and runners being on base or the bases being empty.

Baseball Savant, as always, made things easy. I decided to look at the window from 2013 to the present. Though Gameday doesn’t do a perfect job of classifying pitches, by and large it’s pretty good, so there should be a decent amount of error overlap between different situations. That is, if a pitcher is particularly hard to classify, he’ll be about equally hard to classify with men on or with no one on.

So, the first point: league averages. Let’s establish a baseline:

Bases Empty: 63% fastballs (cutters included)
Runner(s) On: 61%

There’s not much of a difference, but there is a difference, in favor of fewer fastballs and more secondary stuff when runs are more likely to score. In these situations, pitchers might more highly prioritize getting a strikeout. With the bases empty, pitchers might be looking for quick and easy outs. With a runner or multiple runners on, they might then bear down. There are a few ways to try to explain a very small overall frequency gap.

But there are league averages, and there are individuals. I looked at every pitcher who’s thrown at least 1,000 pitches since the start of 2013. The following graph plots their relevant fastball rates. The sample pool numbers nearly 400 and, obviously, there is a very strong relationship we can observe:

In general, as fastball rates go, fastball rates go. There’s an obvious linear relationship. But we’re most interested in the exceptions, and two of them, here, are included. One of them, you’ll recognize as Grant Balfour. The other one of them, you won’t recognize as Phil Hughes, because it’s actually Kyle Gibson. You probably didn’t recognize him as Kyle Gibson, either. It’s OK, you can’t know everything.

Balfour is interesting because, with nobody on, he’s thrown fastballs 69% of the time. But with at least one guy on, he’s thrown 53% fastballs. Gibson is interesting because, with nobody on, he’s thrown fastballs 63% of the time. With at least one guy on, he’s thrown fastballs 73% of the time. We can kind of make sense of the Gibson point: He’s a sinker-baller. In theory, when there are runners on, there are probably runners on first, and Gibson’s going to be looking for a groundball. Sure enough, his rate is highest with a man on first. Balfour has decreasingly featured his fastball over the years as his fastball has lost some life — and his off-speed stuff is better suited to get grounders.

Out of the pool, 24 pitchers have decreased their fastball rates by at least 10 percentage points with runners on base. Those 24 names:

Pitcher	FB%, empty	FB%, on
Grant Balfour	69%	53%
Jordan Walden	70%	56%
Tommy Hanson	54%	41%
Greg Holland	61%	49%
Paul Clemens	71%	58%
Yovani Gallardo	58%	46%
James Russell	52%	40%
Wandy Rodriguez	62%	51%
Tyler Skaggs	67%	56%
Joe Blanton	61%	50%
John Lannan	65%	53%
Jeff Locke	69%	58%
Jose Fernandez	60%	49%
Craig Kimbrel	73%	62%
John Danks	68%	57%
Tom Gorzelanny	60%	49%
Chad Qualls	65%	55%
Josh Outman	56%	46%
Dan Jennings	63%	52%
Roy Halladay	62%	51%
Bryan Morris	57%	47%
Francisco Liriano	44%	34%
Roenis Elias	53%	43%
Marco Estrada	60%	50%

Meanwhile, for the sake of consistency, here are the 24 guys with the biggest fastball-rate increases:

Pitcher	FB%, empty	FB%, on
Kyle Gibson	63%	73%
Pedro Hernandez	60%	69%
Ryan Pressly	56%	65%
Johnny Cueto	66%	75%
David Price	68%	77%
Vance Worley	67%	76%
Chris Withrow	62%	71%
T.J. McFarland	62%	71%
Aroldis Chapman	78%	87%
Anthony Swarzak	62%	71%
Andrew Cashner	62%	70%
Troy Patton	47%	55%
Chris Tillman	58%	66%
Manny Parra	43%	50%
Dale Thayer	73%	80%
Phil Coke	56%	63%
Josh Roenicke	57%	63%
Shelby Miller	73%	80%
Bronson Arroyo	41%	48%
Homer Bailey	58%	64%
Jeremy Hellickson	52%	58%
Will Harris	66%	71%
Mike Pelfrey	71%	77%
Mark Melancon	79%	85%

As a quick experiment, I compared their split performances. Overall, in the league, hitters perform a few points of wOBA better with runners on base. FIP goes up .13 points, and xFIP goes up .25 points. With runners on, the decreased-fastball group averaged a wOBA increase of seven points, with an FIP increase of .20 points and an xFIP increase of .22 points. With runners on, the increased-fastball group averaged a wOBA increase of 12 points, with an FIP increase of .15 points and an xFIP increase of .33 points. Because the samples are small and the errors are pretty big, there’s not enough evidence here to say one group is doing things better than the other. Each group has been a little worse with runners on base, as you’d expect. Realistically, these are just guys pitching to different strengths, because the population of pitchers includes a bunch of different strengths.

You might be interested in the same data on a team-by-team basis. So here’s that table, again starting in 2013. These are just overall numbers, so a team might have different pitcher weights between the two groups (for example, Craig Kimbrel throws more pitches with no one on than with a man or men on, because he’s awesome), but that shouldn’t cause too much of an effect. Things are most interesting at either extreme.

Team	FB%, empty	FB%, on	Difference
BAL	61%	64%	3%
CIN	63%	66%	3%
MIN	62%	64%	2%
LAD	59%	60%	1%
STL	69%	70%	1%
DET	57%	58%	1%
TB	60%	60%	0%
NYM	65%	65%	0%
SD	57%	57%	0%
COL	63%	62%	0%
TEX	62%	61%	-1%
BOS	70%	68%	-2%
ARI	66%	65%	-2%
SEA	61%	59%	-2%
PHI	67%	65%	-2%
WSH	68%	65%	-3%
SF	54%	51%	-3%
KC	62%	59%	-3%
CLE	66%	63%	-3%
CWS	62%	58%	-4%
TOR	59%	55%	-4%
OAK	66%	62%	-4%
ATL	66%	61%	-5%
CHC	66%	61%	-5%
PIT	65%	60%	-5%
MIA	65%	60%	-5%
HOU	64%	59%	-5%
NYY	59%	52%	-6%
LAA	67%	61%	-6%
MIL	62%	54%	-7%

Just by difference, the Brewers are the most extreme team, with a fastball-rate drop of seven percentage points. But the Orioles’ increase is actually 1.9 standard deviations from the mean, while the Brewers’ decrease is 1.8 away. The Orioles have been pretty consistent in this regard in both 2013 and 2014, separately. The Twins had a bigger positive difference in 2013, but they’ve since increased their overall fastball rate with the bases empty. That could be about having different pitchers, that could be about a coaching-staff strategic adjustment or that could be about having Kurt Suzuki. Or it could be a blend of things. As far as the Brewers are concerned, I wouldn’t be surprised if this were conscious, on Jonathan Lucroy and Martin Maldonado’s parts. I don’t think anything in this post is an accident.

There are no sweeping conclusions, only observations. Pitchers pitch differently with nobody on base. Then, some of those same pitchers pitch differently differently with runners on base. Each of them has his reasons, and they’re all just trying to figure out how to be maximally successful. With no one on, Kimbrel’s posted an FIP north of 2.00; in a little more danger, he’s posted an FIP south of 1.00. With no one on, Shelby Miller’s posted an FIP in the mid-4.00s; in a little more danger, he’s posted an FIP in the mid-3.00s. Kimbrel’s thrown fewer fastballs. Miller’s thrown more fastballs. It’s a complicated game that we watch.