Tuesday, May 20, 2014

What even is Dallas Keuchel?

If you've paid any attention at all, you know that Dallas Keuchel is pitching ridiculously well through his first nine starts of the 2014 season. But let's put this into context:

YearIPERA/WHIPK:BBFIP/xFIPGB%
201285.15.27/1.5538K:39BB5.74/5.1252.1%
2013153.25.15/1.54123K:54BB4.25/3.5855.8%
201461.22.92/1.0555K:12BB2.81/2.6867.7%

So right away we can see that things are going really well for Keuchel, but they had started to look better in 2013, even if it didn't exactly show up in his ERA/WHIP (it rarely does, though, amirite? High five.) Keuchel had nine appearances in 2013 in relief which skewed his numbers upward. He had a 4.90 ERA/1.48 WHIP in 22 starts, with a 2.55 K:BB ratio. As a reliever in 2013, that was a 6.75 ERA/1.88 WHIP and a 1.60 K:BB ratio. But as you can see from the table above, Keuchel's FIP and xFIP were trending downward.

What is Keuchel doing differently in 2014? Well, his K/9 rate is up almost a full strikeout per nine innings and he has cut his BB/9 rate almost in half (3.05 in 2013 to 1.75 in 2014). His BABIP is way down, as well - .340 in 2013 to .290 in 2014, and that's reflected in the dive his FIP/xFIP WHIP has taken. (NOTE: Terence's point in the comments below are absolutely correct, and this sentence has been updated as a result.)

Obviously Keuchel is getting far more groundballs this year than in his career. We don't have the hard numbers on Keuchel in the minors regarding groundball rate, but at Keuchel's MiLB page they do list his GO/AO ("groundout / air-out") numbers. So we can combine some data we do have as such:

YearLeagueIPERA/WHIPK:BBGO/AO
2009NYPL56.22.70/1.0844K:9BB1.72
2010CAL120.23.36/1.2897K:25BB2.94
2010TEX53.24.70/1.3036K:11BB1.93
2011TEX127.23.17/1.1276K:27BB1.87
2011PCL36.07.50/1.7815K:12BB1.77
2012PCL92.13.90/1.2150K:20BB2.04

Now this very helpful FanGraphs piece says that you can't just do a direct conversion from GO/AO to GB%, but you can get some general equivalencies. Using the hard math they've already done, we can basically convert the GO/AO lines to GB% like so:

2009 NYPL: 48%
2010 CAL: 57%
2010 TEX: 50%
2011 TEX: 50%
2011 PCL: 48%
2012 PCL: 51%

Those aren't perfect conversions, but it's close enough for us to say with accuracy that Dallas Keuchel has not historically generated groundballs the way he's doing now. What changed? I can't - and more accurately, won't - say it any better than Eno Sarris did in FanGraphs back on April 22, at which point Keuchel had only made four starts:

Keuchel was one of my pre-season deep league favorites, based on pitch-type peripheral analysis. What I noticed last year was that he’d turfed a bad curve for a good slider. He continued that trend this year. Take a look at the swing and ground-ball rates on his curve and slider over the last three years...

...I’ll take a pitch with above-average whiff rates and average ground-ball rates over a pitch with average ground-ball rates and below-average whiff rates. And so will Keuchel. The new slider goes faster than his old curve, and it also breaks away from lefties. 

And there's even a fancy table in between the ellipses which backs up Sarris' point. You started to see Keuchel's GB% start to rise in 2013 and of course it's off the charts in 2014. How off the charts? Heading into games on May 20, Keuchel's 67.7% GB rate is the best GB% any qualified starting pitcher in baseball - 6.0% better than 2nd-place Tim Hudson.

I'm not sure how sustainable a 67.7% groundball rate is. Right now, Keuchel has the highest GB% of any qualified pitcher since 2000. Derek Lowe is currently sitting at #2 (66.9% in 2006) and #3 (66.8% in 2002) behind Keuchel. Of all the qualified starting pitchers from 2002-2013 - the years in which FanGraphs has data - there have only been 26 instances where a pitcher maintained a GB% over 60.0% for the whole season - half of those instances came courtesy of Derek Lowe (seven times) and Brandon Webb (six times). Derek Lowe had an upper-80s fastball with a low-80s slider, mixing in a change and a curve, though the fastball was his most-dominant pitch. Perhaps that sounds familiar.
 
I get it - we're talking about nine starts, which means that Keuchel needs to do what he's been doing for at least 21 more starts. Trying to say that 2014 Dallas Keuchel can sustain his 2014 numbers, which are similar to 2002 Derek Lowe (5.4 WAR), 2006 Brandon Webb (6.3 WAR), or even 2002 Roy Halladay (7.3 WAR), is premature at best. But at least there are some numbers behind Keuchel's surge - you're not looking at an FIP/xFIP over 4.00 and just waiting for the shoe to drop. So let's split the middle and do this: Why don't we just sit back, enjoy what Keuchel has done so far, and hope it continues.

Update: I swear I had not seen Mike Petriello's piece on Keuchel from today before writing this.