This is the analytic segment of Astros Trending. This week we explore the recent decline in Dallas Keuchel. Tomorrow we do the week in review.
Dallas Keuchel: Ups and Downs
In some ways, Dallas Keuchel is baseball’s equivalent to boxing’s Cinderella Man, James Braddock. Braddock, who was a small heavyweight and not a powerful knockout puncher, relied on defense, smarts, tenacity and great counter-punching. Keuchel is likewise the consumate overachiever, his pitches punch over their weight, a fastball that barely touches 90 mph, secondary pitches that don’t have extraordinary movement. Keuchel has always relied on smarts, that is, ingenious sequencing, and great control. All that, and dedication to his craft, because when you can’t throw 95+ in today’s game, then your craft has to be exquisite.
Keuchel, like Braddock, has had to be almost perfect just to be good.
In 2013, slowballing and inexperienced Keuchel got his chance on the only team bad enough to have a place for him, the Houston Astros. He was the fifth starter his first two years on tanking teams that were historically bad, and he barely managed to keep his job, sporting +5 ERA’s both years.
Then he learned his craft. 2014 was a breakthrough for Dallas, winning 12 on another bad team, with an ERA of 2.93. Was this a fluke?
No. The renaissance of the Astros in 2015 coincided with the apotheosis of Dallas Keuchel, who led the league in wins, WHIP, WAR among pitchers, and was second in ERA, at 2.48 just .03 behind league leading David Price. He made the All-Star team and was awarded the Cy Young award.
Who is this guy? Where did he come from? How did he hone his craft?
The Astros were surprise entrants into the playoffs that year, and felt poised, with their new ace and other young talent, to hit it big in 2016.
But a disappointing year it was, no playoffs, in no small part due to the collapse of Dallas Keuchel, who won only 9 games, and whose ERA ballooned 2 runs to 4.55. He was pitching injured, and vowed never to do it again.
In the year of magic, the year of the championship, 2017, Dallas Keuchel rediscovered his own magic, reacquiring his craft, returning to the All Star game, winning 14 games in limited innings, and again posting a sub 3 ERA.
Wait a minute, let’s break that down. 2017 was really a year of contrasts for Keuchel, a microcosm of his career of contrasts. Before the All Star break he was 9-0 with a 1.67 ERA. Then he suffered a nagging neck injury and returned going 5-5 for the slugging Stros of 17, but only had a 4.24 ERA.
Keep this second half 2017 in mind when you consider the 2018 Dallas Keuchel: 3-8, 4.45 ERA, 4.20 FIP, 3.64 xFIP. But first, compare Keuchel 2018 to “bad Keuchel” 2016: 4.55 ERA, 3.87 FIP, 3.53 xFIP.
Now, if you combine second half 2017 Keuchel to first half 2018 Keuchel you get the following results: 8-13, 4.35 ERA, 4.34 FIP, 3.69 xFIP, ERA- 106. Keuchel’s problems this year really began second half, last year. Granted, these numbers are not terrible, merely average, or just slightly below.
(For those Keuchel optimists who want to cite xFIP to project better days for Dallas, his xFIP has been below his FIP in every season of his career and for his career is .3 points lower than his career ERA. FIP is a better predictor of Keuchel’s trajectory as his career FIP, 3.76, is almost identical to his career ERA, 3.71.)
Even 2018 has shown Keuchel’s cyclical tendency. In March/April he had a 4.00 ERA. In early May he seemed on a better trajectory, posting a 1.64 ERA in his first three starts of that month. Since then he is 0-3 in 5 starts, with a 7.33 ERA, 4.36 FIP, and 3.70 xFIP. No doubt a BABIP of .385 and some bad home run luck play a part here, but it is still hard to avoid that we are seeing “bad” Keuchel lately, and through most of 2018.
Will Dallas arise again like a Phoenix?
Well, as I said earlier, for Dallas to be good, he has to be nearly perfect. When he misses location with his below average velocity there is a good chance that major league hitters will make solid contact.
Dallas is Getting Hit this Year
So let’s compare contact rates between the “good Dallas” and the 2018 Dallas. Are hitters hitting Dallas harder, or is he just unlucky lately?
We all know that “good Dallas” gets a lot of ground ball contact. In 2014, 2015, and 1st half 2017 respectively the numbers were 63.5%, 61.7% and 67.4%. In 2018 his ground ball percentage is 54.4%
Good Dallas in 2014, 2015, and 1st half 2017 had very low hard hit rates, 19.7%, 21.3% and 20.6% respectively. 2018 Dallas? 30.8% That’s about 10% more hard hit balls than in the good years of Keuchel’s career. Of course the batted balls are finding holes. They are getting hit in the air, and they are getting hit harder.
According to Statcast, In 2015 Keuchel was elite in various contact categories, being very low in barrel % (3.4), exit velocity (85.4), and by their calculation, hard hit % (28.3) The average launch angle was 0.9. In 2018 the barrel % is 5.4, the exit velocity is 88.2, and the launch angle is 3.9. The hard hit % this year is 35.5 by Statcast measures.
Another contact difference, often associated with more home runs, is the pull rate. In 2014, 2015 and 1st half 2017 Dallas allowed 41%, 36% and 39.2% of the balls hit to be pulled. In 2018 that rate is 45.7%
Is Keuchel allowing more home runs? In 2014 his HR/9 rate was .5 and in 2015 it was .66. 1st half 2017 it was .71. This year he is giving up 1.27 home runs /9 innings.
One more measure of reduced effectiveness is Keuchel’s decreased swing and miss rate. In 2014 it was 9%, 2015 10.3%, and in first half 2017 it was 10.5% . This year he is missing bats only 8.7% of the time.
More contact, harder contact, higher contact, more pulled contact means more hits, more home runs, higher ERA.
I believe we have fairly established that Dallas Keuchel is not pitching as well as usual this year, both by performance standards and sabremetric standards.
Enquiring minds want to know; why?
Keuchel’s Demise: Is it the Slider?
I am going to try to prove that Keuchel’s problems this year are due to the loss of effectiveness of a number of his pitches, but most importantly, the slider, and that is mostly due to a loss of slider command. I will use data from Fangraphs, Brooks Baseball and Baseball Savant to make this case.
To simplify this discussion I am going to isolate and compare two years, “good Keuchel, 2015, and current Keuchel, 2018. First, let’s establish his mix and pitches, 2015, and 2018, which is somewhat different.
Keuchel pitch usage 2015, 2018
Three of his five pitches were used slightly less in 2018, the four-seam slightly more, but the cutter has seen a doubling in usage. We’ll look at that later.
According to Fangraphs there has been a slight drop-off in velocity for Keuchel, which might be a small factor in this analysis. Sinker velocity, 2015: 90.5. 2018: 89.5. Four seam velocity, 2015: 90.5, 2018: 90. Slider velocity, 2015: 80.2, 2018: 79.2. Change velocity, 2015: 79.9, 2018: 80.3. Cutter velocity, 2015: 87.4, 2018, 86.2.
Before you say end of discussion, he’s lost velocity, these numbers are slightly better than his mostly successful 2017 year.
Next, and this is a very important table, is Fangraph’s runs above average measure/100 pitches. Zero is average, positive numbers mean above average pitches.
Keuchel Runs Above average/100 pitches, each pitch, 2015, 2018
This data shows that three of Keuchel’s 2018 pitches have lost effectiveness compared to the same pitch in 2015, his bread and butter sinker, the change, and most severely, the slider. This has gone from being his most effective pitch, to his least effective. The four seam is positive for the first time in Keuchel’s career, and the cutter is the best it has ever been since Keuchel emerged as a star. Not coincidentally, Keuchel is using these two pitches more than he ever has in the past. They rate as his two most effective pitches this year.
The next chart will show changes in the trajectory and movement in Keuchel’s pitches between 2015 and 2018. This data comes from Brooks Baseball and was compiled before last Sunday’s game.
Keuchel Trajectory, each pitch, 2015, 2018
|pitch||Hor. mvt||Vert. mvt|
|pitch||Hor. mvt||Vert. mvt|
|4 seam 2015||2.94||9.52|
|4 seam 2018||-0.76||7.4|
The four seam looks like a new pitch, instead of breaking away from the right hander it moves slightly in. This transformation actually began in September, October of 2017. It is a more effective pitch as we have seen, which he uses almost exclusively against right handers.
The sinker is not much different, but it moves a little less and is a mph slower. This combination may account for the somewhat reduced effectiveness of this pitch.
The change has slightly less drop.
The slider has very similar movement, but is about one mph slower than in 2015. As we shall see, what is very different about the slider is the command and location.
The cutter, like the 4 seam, is a very different pitch in 2018. Instead of coming in basically straight, as in 2015, it now has break towards the right handed batter. This is the pitch which Keuchel now relies on 20% of the time with increased effectiveness mostly against right handers.
The following table shows performance and contact data for each pitch, 2015, 2018.
Keuchel Performance and Contact Data, each pitch, 2015, 2018
|4 seam 2015||3||0.31||0.31||0.385||17.57||25.93||229|
|4 seam 2018||0||0.27||0.324||0.357||15.87||44.44||128|
According to the table only the 4 seam and the cutter have lower batting averages in 2018 than in 2015. Every pitch produced a higher slugging percent in 2018 than 2015, but the most dramatic increase by far is with the slider, which opposing teams are slugging at a .707 clip. That’s Ruthian. The BABIP is .444. Is the high slugging just luck, or is the pitch being scorched? Probably both. But look, Keuchel has already surrendered more home runs on the slider in 2018 in about a third as many pitches as he did in 2015. When we look at spray maps I think you will see just how hard the slider is being hit.
The whiff/swing rate gives a good indication of the deceptiveness of each pitch. The swing and miss is the most preferred outcome for the pitcher. The whiff/swing rate is down on every pitch in 2018 except the cutter. The difference is most pronounced with the slider.
Ground balls are preferred outcomes to most other hit balls because they lead to outs more often as well as double plays. It has always been the key to Keuchel’s success. The ground ball rate is down in 2018 on every pitch except the 4 seam and the cutter. The biggest drop in ground ball rate was on the slider.
In summary, according to both Fangraphs and Brooks Baseball data the slider, which used to be a relative strength, is by far Keuchel’s worse pitch this year.
A look at some heat maps and spray maps should confirm this and perhaps shed some insight as to why.
Here is a spray map of hits against Keuchel’s slider in 2015.
Compare this to the spray map from 2018.
I could show you the spray maps for the other four pitches for both years and of course they will show changes from year to year, but trust me, none are near as dramatic as this. Obviously, the slider is being driven this year, including to right field, by left handers to left, and right handers to right.
And I think the following heat maps explain why. The first map will show the placement of Keuchel’s slider in 2015.
The map for 2017 is similar, except a little more distribution directly below the strike zone.
Below is the heatmap for the slider from 2018
The first heat map, although it represents almost three times as many pitches, is smaller and tighter. It shows greater command and control. It also shows a much greater proportion of pitches in the left handers’ swing and miss zone, whereas the 2018 heat map shows too many pitches in the strike zone, indeed, too many in the sweet part of the zone.
Another indication that the slider is the main culprit is the reverse splits Keuchel has this year lefty versus righty.
The WOBA against on right handers for Keuchel in 2018 is .311. On lefties its .348. Compare those numbers to those of 2015: .268, .201.
The following chart illustrates the difference between the effectiveness of Keuchel’s slider against left handers in 2015 compared to 2018.
Keuchel’s slider against left handers
In 2018 Keuchel has thrown 225 sinkers, 98 sliders and only 17 other pitches to left handers. Against lefties Keuchel is a sinker, slider pitcher. The slider is supposed to be the put away pitch to left handers which he throws to them about 30% of the time, twice the rate he uses it against righties. In 2015 it was working. In 2018 it is getting crushed.
Following is the spray map of Keuchel’s slider against lefties in 2015 and then a map of his slider against lefties in 2018.
This is 2018.
This has been a long essay, and I must mention that to a less dramatic degree the sinker has been problematic for Keuchel this year as well. Since this is still his most oft used pitch it has hurt his performance significantly as well compared to earlier years. But the difference is not nearly as pronounced as with the slider.
Before concluding I want to discuss a recent article in The Athletic by Eno Sarris entitled “The One Thing Dallas Keuchel is Missing This Year.”
According to Sarris, the culprit is the four seam fastball. Using advanced research methods Sarris claims that what resurrected Keuchel’s career in 2017 was the use of the high four seamer. Because it tunneled well with the secondary pitches, especially the slider and change up, it made the other pitches more likely to achieve swings and misses.
According to Sarris’ research what is missing in 2018 is the four seam command, which makes the pitch less effective in setting up the other pitches.
I have a problem with his hypothesis. According to Sarris’ own data Keuchel only threw the four seamer 116 times in 2017, 4.28% of pitches according to Brooks, less than any other offering in Keuchel’s arsenal. That is well less than once per inning.
I just don’t see how a pitch which in the majority of at bats a hitter wouldn’t see even once could make all the other pitches more effective as Sarris claims.
Sarris’ hypothesis also doesn’t explain why left handers are doing so much better this year, since the four seamer is used almost exclusively against right handed hitters.
He is using the four seamer twice as much this year, and according to Fangraphs it is a much improved pitch. But that is not Sarris’ point. He is saying its tunneling isn’t being used as effectively this year to set up the other pitches, and that explains the reduced effectiveness of the other pitches.
Well, even if Sarris is right about that, it doesn’t contradict the evidence here but perhaps adds another explanation as to the reduced effectiveness of the slider, and to a lesser extent, the sinker.
I still prefer to rely on the simpler explanation; that Keuchel is throwing his slider more promiscuously this year, and finding the fat part of the strike zone with it too often, and that his sinker has lost a little movement.
How to fix those issues will take the expertise of someone way more knowledgeable than me to figure out. Perhaps this is all just another one of Dallas Keuchel’s natural cycles and all will be good again.
He is, after all, the Cinderella Man of baseball. The Cinderella man made a great comeback. Likewise, the Cinderella Man of baseball never quits, and just when you think you’ve got him down, he comes at you with a mighty counter-punch.