What Can Dallas Keuchel Do for an Encore?

The American League Cy Young winner posted a 2.48 ERA and 2.91 FIP last season, and allowed a ton of soft contact. Can he do it again?

While the 2016 season is already underway, let’s take a second to remember some of the lasting images from 2015: Jose Bautista bat flips, Mets pitchers throwing gas, and Royals hitters slapping timely base hits.

How about another one: Weak ground balls against Dallas Keuchel.

According to Baseball Savant, Keuchel induced more ground balls hit slower than 80 miles per hour than anyone in baseball, and on all balls in play, his average exit velocity allowed of 85.8 mph was third lowest (minimum 200 balls in play).

This may have helped him run a .269 batting average on balls in play, which, in turn, contributed to a 2.48 ERA (62 ERA-), the American League lead in WARP (6.2) and Win Probability Added, and ultimately a Cy Young Award.

Keuchel’s success went beyond good results on balls in play, as he also posted a 71 FIP-, 68 xFIP-, and 77 cFIP ( Contextualized FIP, which is the most predictive run estimator available; FIP-, xFIP- and cFIP are all scaled so 100 is average).

His strikeout rate jumped up to 23.7% after hovering around 18.0% in his first two full big league seasons, while his walk rate of 5.6% marked a continual improvement.

The second-best ground-ball rate (61.7%) among qualified pitchers led to a home run per nine innings rate of just 0.66, despite an unsustainably high home run per fly ball rate (13.6%).

As good as his peripherals were, his run prevention still outpaced them, as his -0.43 ERA-FIP difference was 17th in MLB.

This makes Keuchel a more interesting case than your typical elite starter. While pitchers generally have a limited (at best) control over what happens after a ball is put in play, the extreme level at which Keuchel has suppressed his BABIP is worth a closer look.

A Closer Look At Balls in Play

Keuchel’s 25.2% soft-hit ball rate not only led the Majors last year, but was the lowest in four years. His 21.3% hard-hit ball rate also led MLB, and was the third lowest since 2012, trailing only Garrett Richards’ and his own 2014 seasons (per Baseball Info Solutions, via FanGraphs).

This explains the low BABIP right?

It might not actually be so simple.

While the Hardball Times found there to be a strong relationship between exit velocity and power stats, there was little correlation between exit velocity or hard-hit ball rate and BABIP.

Even among the 22 pitchers with a soft-hit ball rate of 20.0% or better, the median BABIP was just .292 in 2015, hardly better than the MLB average of .296. For this group, BABIP and soft-hit ball rate correlated at just -0.11 (the negative relationship implies that as soft-hit ball rate goes up, BABIP goes down, which we would expect; the relationship is small though).

These findings are counter-intuitive, but do serve as a reminder of how random BABIP can be.

So is all of Keuchel’s ball-in-play success just the product of luck/randomness?

No, almost certainly not, as that is taking things to another extreme. There are probably two questions to answer here:

1. A backwards looking one: Did Keuchel have anything to do his low BABIP?

2. A forwards looking one: If he did, can he repeat it?

Line-drive rate is probably a good place to start, since it correlates moderately well with BABIP (at about 0.37 since 2010; line drives also lead to more hits than grounders or flies, as batters hit about .685 on liners, .239 on grounders and .207 on flies, per FanGraphs).

2015 marked the second straight year Keuchel ranked in the top 20 in line-drive rate allowed, posting a rate of 18.7% last year and 17.2% in 2014. Line drive rate is less consistent than ground-ball or fly ball-rate, however; Baseball Prospectus’ Russell Carleton found the later two stats take about 70 balls in play to hit their stabilization rates, compared to 650 for line drives.

Keuchel has allowed over 1,200 balls in play since 2014, so we have a large enough sample that we can infer he actually possesses some ability to suppress line drives (rather than attributing the low rates to random variation).

But we should probably also look at ground balls, since Keuchel allowed a ground ball on 61.7% of all balls in play last year, and often got praise for inducing soft ones. Plus, Keuchel ran one of the lower BABIPs in baseball, despite the fact that he ranked 92nd out of 113 pitchers in terms of BABIP allowed on non-grounders (minimum 200 non-grounders put in play; for what it’s worth, Keuchel ranked 24th in this group in exit velocity allowed on such plays and was tied for 14th in the majors in infield fly per fly ball rate...maybe his overall BABIP should have been lower!).

Keuchel ran a .203 BABIP on ground balls, was 10th in exit velocity on grounders, and as mentioned, led the league in grounders slower than 80 miles per hour. Does this matter?

It seems so, despite a relatively weak correlation between ground ball exit velocity and ground ball BABIP.

Speed of Grounder Min.Grounders # of Pitchers in Sample Correlation w/ BABIP Median Exit Velo Median BABIP
All Grounders 100 146 0.19 86 0.255
Slower than 80 mph 50 82 -0.1 67.8 0.125
Between 80 & 90 30 58 0 85.3 0.196
80 mph or faster 50 179 0.26 94.5 0.315
90 mph or faster 50 144 0.2 98.5 0.359

The results here are kind of whacky, and a common thread for each “bucket” here is that BABIP doesn’t correlate well with exit velocity. There's still a ton of randomness at play, and since we only have one season’s worth of publicly available batted ball data, the samples are still pretty small (plus, some of the data is missing and flawed).

Refreshingly, though, judging by the median BABIP in each subset here, it does seem that harder hit balls produced higher BABIPs last year.

This brings us back to Keuchel, who, as mentioned, led baseball in terms of grounders allowed that were slower than 80 mph (a bucket which produced a median BABIP of 0.125).

Overall, we have exit velocity data on 322 of Keuchel’s 393 grounders, and these pitches left the bat with an average velocity of 83.3 mph and produced a .213 BABIP.

Keuchel Grounders Slower than 80 mph 81 to 90 mph 91 mph+
# of pitches 127 75 120
BABIP 0.111 0.173 0.35
Approx. MLB Median BABIP 0.125 0.196 0.359

This is an admittedly quick and dirty way of looking at things, but it does support the inference Keuchel’s contact management helped lower his BABIP.

Whether this is sustainable or not is a separate question. The short answer is “who knows?” and the just slightly longer answer is “we can’t say yet.”

What we do know is this is the second consecutive season Keuchel allowed a low BABIP on ground balls. In 2014, despite allowing a .295 overall BABIP, his rate on ground balls was just .210. For his career, on about 1,200 grounders, he has allowed a .219 BABIP.

Still, Russell Carleton found that BABIP (for all batted ball classifications) takes about 2,000 balls in play to stabilize, so it might be best not to jump to conclusions.

In 2015, his overall BABIP rose from .255 to .290 from the first half to the second, and on ground balls only, it increased from .154 to .285. This rise came in spite of the fact that his exit velocity on grounders only rose from 83.0 mph to 83.7.

In general, pitchers seem to have some degree of control over exit velocity, judging by last season anyway. From the first half to the second half of the year, exit velocity allowed in the first half correlated with that of the second half at 0.45 (for pitchers who allowed 150 balls in play in the first and 100 in the second).

On grounders, the correlation was slightly lower between pitchers who allowed 75 grounders in the first half and 50 in the second (though this could easily be a sample size issue).

In any case, my findings here are consistent with previous research that indicates that while pitchers have some degree of control over batted ball velocity, hitters have more control.

At this point, the safest way to approach all this is probably to assume Keuchel’s true-talent BABIP is not .269, but he also has some ability to manage contact which could allow him to again outperform his peripherals.

And, if he doesn’t, these peripherals have still been pretty damn good, so this might not even matter.