Kevin Cash fixes pitchers: the Carlos Carrasco story

When it was announced that Andrew Friedman was leaving, that Joe Maddon was leaving, and that Dave Martinez would not be promoted to replace him, there was one consistent reaction among Rays fans to all three: "I hope this doesn't mean that we'll lose Jim Hickey."

Of course, most of us don't really know what exactly Jim Hickey does, and we can't be sure he's as good at it as he seems, but we do know that during his time as pitching coach, the Rays have had a knack for developing new pitchers and fixing broken ones. It's been one of the strengths of the team.

Well, it turns out that Hickey is not the only man in baseball who can work with pitchers. People who follow the Indians closely give bullpen coach Kevin Cash a significant amount of credit for Cleveland's remarkable pitching staff, which narrowly beat out the Rays to set the single season strikeout record last year. And central in that pitching staff was the abrupt turnaround of Carlos Carrasco from highly-touted disappointment to top-of-the-rotation starter.

So it seems like Kevin Cash, who many consider the leading candidate to be the next Rays manager (read the Kevin Cash profile here), can fix pitchers too. Let's take a closer look at what happened to Carrasco.

The Turnaround

Carrasco was an international signing by the Philadelphia Phillies. Baseball America rated him their best prospect after 2006 and 2007, and their second best prospect after 2008, at which point he was traded to Cleveland as a part of the Cliff Lee deal.

Scouts raved about his changeup and his fastball, but had questions about his ability to handle pressure. From BA:

Opposing managers and scouts use words such as "fold up" or "soft" to describe Carrasco, who's prone to the big inning.

He got his first major league start in 2009, and mostly disappointed on the major league level, before having Tommy John surgery to end his 2011 season.

Year	IP	K%	BB%	ERA	FIP
2009	22.1	9.8%	9.8%	8.87	7.08
2010	44.2	20.2%	7.5%	3.83	4.13
2011	124.2	15.9%	7.5%	4.62	4.28
2013	46.2	13.8%	8.3%	6.75	4.10
2014	134	26.5%	5.5%	2.55	2.44

In 2013, Terry Francona took over as Indians manager and installed Mickey Callaway as pitching coach, and Cash as bullpen coach. While Carrasco continued to struggle for his major league portion of the 2013 season while working with those two, he put it all together in 2014, and pitched better than he ever had at any level.

Pix Mitch

So, what exactly happened? If we're going to give the Cash, we need to identify what he did. Part of it is obvious with a quick look through the excellent resources at Brooks Baseball.

Despite his fastball and changeup getting all the love from scouts, Carrasco's most effective pitch was actually his slider. The above graph shows whiffs per pitch, which is a decent shorthand for pitch effectiveness, but other metrics tell the same story. According to FanGraphs PITCHf/x linear weights, during the 2011 season, Carrasco's four-seam fastball was worth .33 runs per 100 pitches and his two-seam fastball was worth -2.46 (negative is bad for a pitcher in this case, positive is good). His changeup and curve were each worth -.11 runs per 100 pitches. His slider, though, saved 1.86 runs per 100.

There's a basic game theory concept called the "Nash Equilibrium" that applies well to pitching. If a pitcher throws a lot of one pitch, batters will prepare for it, and it will become less effective on a per-pitch basis. At the same time, because they're preparing for that pitch, they'll be unprepared for other pitches and those will be more effective. The point at which the pitcher is overall most effective is when the individual outcomes for each pitch are the same (or at least similar).

That means that if you have a pitch (like the slider in Carrasco's case) that you don't throw very often but that gets better results than everything else when you do, you should probably throw it more. That's exactly what Carrasco did.

When Carrasco threw his slider more, all of his other pitches got better as well. Q.E.D? Well, not quite.

A team doesn't need a Hickey or a Callaway or a Cash to tell their pitchers to use their best pitches. Any analyst worth his salt can figure it out, and it's something that the Rays clearly do just fine (it's a big part of how they fixed Fernando Rodney). However, teams do need experienced coaches to help their pitchers use the game-theory-inspired plan that front office analysts come up with to make those best pitches most effective, because it's not just about what you throw, but when you throw it.

Smart Sequencing

Baseball analysts are pretty good now at describing pitching both on pitch-by-pitch level and on a batter-by-batter level. We are much worse at understanding how those two relate to each other, but it's a relationship that matters. In Carlos Carrasco's case, the pitch-by-pitch level explains most of his improvement but not all of it. Consider the difference between Carrasco's results in 2011 and in 2014 (data from Baseball Savant):

	Called Strike	Swinging Strike	Foul	Ball	In Play
2011	15.6%	8.7%	18.9%	36.1%	20.7%
2014	17.5%	13.5%	18.3%	32.1%	18.3%

This is what happened when he threw his slider more often, and his other pitches played up because of it. He produced more strikes of both types, fewer balls, and fewer balls in play. It's obvious that these new per-pitch results should translate to more strikeouts and fewer walks, but by how much?

To try to answer this question, I dusted off an old project that I affectionately call WHEELS.

It's a probabilistic model of what the results should be on the per-batter level if results on the per-pitch level were distributed regardless of count. Of course, that's not the way results are actually distributed. Count does matter. But my hypothesis is that the difference between a pitcher's actual strikeout and walk rates and what WHEELS thinks he should be capable of is a measure of the effectiveness of his sequencing.

Huge caveats apply. This is a hypothesis, and it's not well-tested. Let's look at Carraso's example anyway.

First in 2011, pre-Cash:

	Acutal	WHEELS	Difference
K%	15.9%	20.5%	-4.6%
BB%	7.5%	9.6%	-2.1%

And in 2014, post--Cash:

	Acutal	WHEELS	Difference
K%	26.5%	30.0%	-3.5%
BB%	5.5%	7.2%	-1.7%

I don't know enough about this stat I've created to make large statements about it, but small changes in the strikeout and walk rates of a pitcher matter. At the same time that Carrasco worked with Kevin Cash to implement a new pitch mix that improved his "stuff" overall, he also got better at turning his "stuff" into results. He still struck out fewer batters than he would in a simple random distribution, but it's closer than it was before.

The bottom line is that Cash helped him radically change the way he he pitches while improving his outcomes on two separate levels.

Mark me down as a person who believes that Kevin Cash understands pitching. He seems like a guy who will be able to understand and add to the preexisting Rays pitching development machine, and in my eyes, that makes him a strong choice for manager.