Pitching (and hitting) backwards in 2013

As crafty veteran arms battle for that last rotation spot, broadcasters looking to fill spring training air time will espouse the virtues of "pitching backwards." It's hard to quantify exactly what it means to pitch backwards; in general, it's used for pitchers who throw anything other than a fastball in the traditional "hitter's counts." R.J. Anderson wrote a good primer on the subject for Baseball Prospectus back in 2011, and Bill Petti looked at the most "backwards" pitchers after the 2012 season.

In this article, I'll be looking at which pitchers were the most backwards in 2013. I also want to determine which hitters saw the most backwards pitching last season, and how they performed in those at bats.

I'll define pitching backwards by looking at the raw percentage of non-fastballs thrown on the first pitch, and the usage ratio Anderson defines as:

(First-pitch non-fastballs/first pitches)/(Total non-fastballs/total pitches)

This ratio allows us to identify those pitchers who threw more first-pitch breaking balls or offspeed pitches relative to their typical offerings.

Before we can do this, though, we have to agree on what is a fastball and what isn't. I divided 13 of the PITCHf/x labels into one of three larger groups -- hard, breaking, and offspeed* -- according to the grouping used on BrooksBaseball.net. The 13 labels of interest, their general grouping, and their roughly equivalent Brooks Baseball label are given in the table below.

* - Anderson's article uses "breaking ball" and "offspeed" somewhat interchangeably, but the two are usually distinct.

Keep in mind, though, that Dan Brooks and Harry Pavilidis reclassify every pitch by hand, often with insight from the pitchers themselves, so they may disagree with the labels assigned by the PITCHf/x algorithm to specific pitches*. Any pitch types not listed above were excluded from this analysis; note that this deliberately excludes knuckleballers who often rely heavily on the knuckler in hitter's counts. Since I'm more interested in starters than relievers, I'm also excluding any players with fewer than 350 PA last season.

* - I don't know if the information is out there, but I would love to learn how the PITCHf/x classification algorithm was designed. I'd also be interested in a confusion matrix between the Brooks Baseball classifications and the automated classifications.

So let's start looking at pitchers: first, by looking at those pitchers with the lowest rates of "hard" first pitches. All of these pitchers threw significantly more first-pitch offspeed and breaking balls than the league average of 31 percent.

Already we see that the most "backwards" pitchers aren't who we would expect. Sure, Bronson "Crafty" Arroyo is right up at the top, but he's followed by Ubaldo Jimenez, whose mean fastball velocity sits right around the league average of 92 mph. And sixth on that list is Justin Verlander! So clearly this list is more than a "Who's Who in Non-Roster Invitees."

You'll notice I've also included each pitchers' wOBA allowed, as well as their wOBA allowed when starting an at-bat with fastballs and with softer offerings. Here, too, no clear pattern emerges: Madison Bumgarner's wOBA is 52 points lower in plate appearances where he opened with a fastball in 2013, but Arroyo's is 23 points better, and most of the other pitchers atop this list see little appreciable difference.

This table gives us a different perspective: these pitchers are more likely to throw a breaking ball or offspeed pitch on the first pitch than later in the same plate appearance. But this table also makes pitching backwards seem more like a preference than something forced on a pitcher with a weak arsenal, since we again see little correlation between higher ratios and fastball velocity, or between higher ratios and performance (as measured by wOBA allowed).

If pitching backwards is in fact a matter of feel, teams might find it useful to compile these same numbers for the catchers who call these games*. Maybe certain catchers are more likely to call for offspeed pitches early in the count. Armed with this knowledge, opposing hitters would thus be less likely to be surprised by a first-pitch curveball or slider.

* - And I'm sure they've already been doing this for years, both by automated methods like PITCHf/x and manually by advanced scouts.

Performing this same analysis for hitters, on the other hand, does reveal some interesting patterns. Here are the top ten hitters sorted by percentage of first-ball offspeed or breaking pitches.

It might not be universal, but most of the names on the top of this list fit a similar profile. Hamilton, Alvarez, and Rasmus all have reputations as power hitters with a willingness to chase pitches outside the strike zone and post high strikeout numbers as a result. Let's take Alvarez as an example. When the first pitch he saw was a fastball, his wOBA was .371, which was Bryce Harper's wOBA last season. But give him a first pitch breaking ball and that wOBA drops to .268. That's Juan Pierre territory. If you had a pitch that turned Bryce Harper into Juan Pierre, wouldn't you throw it more often than not?

It's interesting to see Wil Myers on this list, too. Myers struck out 91 times in a little over a half a season; give him 650 PAs and he's probably among the league leaders. There's a dropoff in his performance, too, from among the league leaders to league-average. If Myers struggles this spring, it could simply be regression, but it could also be caused by pitchers throwing him more breaking balls and changeups until he adjusts.

. . .

PITCHf/x data courtesy MLB Advanced Media. All other statistics courtesy of FanGraphs and Baseball-Reference.

Bryan Cole is a featured writer at Beyond the Box Score. You can get him to chase a high fastball, and you can follow him on Twitter at @Doctor_Bryan.