So much of what makes pitchers effective at the major league level is their ability to keep hitters off-balance. Sure, a 95 mph fastball with movement and a Lord Charles curveball help, but even these physical tools are only as effective as a pitcher’s ability to create uncertainly in the hitters mind from pitch to pitch.
One — admittedly crude — way of looking at this is whether a pitcher throws the type of pitch that’s expected in a given count. Does a pitcher throw fastballs in “fastball counts”, or do they throw off-speed pitches? Pitchers that throw counter to expectations are often said to “pitch backwards”. The Rays’ James Shields is someone that has been referenced as such a pitcher over the past few years.
But exactly how backwards does Shields pitch? And who are some other pitchers that fit into this category?
To tease this out I first need to define what are “fastball counts”. I used PITCHf/x data from 2010 through 2012 to calculate the percent of fastballs (i.e. four-seam, two-seam, cutters, sinkers, etc.) and off-speed pitches thrown in each of the 12 possible counts. Here’s how the counts shook out, sorted by descending fastball frequency:
|Count||League Average: Fastballs|
Not surprisingly, pitchers tend to go to a fastball of some sort in three-ball counts at least 70% of the time. Facing a 3-0 count, a pitcher wants to avoid walking the hitter and generally goes to the pitch they can control the best–most often, their fastball (94%). 2-0 counts draw a fastball 81% of the time, likely due to pitchers wanting to avoid a 3-0 count and all that comes with it.
Pitchers league-wide generally move away from fastballs as the count moves in their favor (i.e. two-strike counts, excluding 3-2), throwing no more than 56% of fastballs in these counts on average. This makes sense, as pitchers have the advantage as hitters will generally expand their zone to protect against a called third strike, making them more likely to swing at breaking balls and other off-speed pitches.
Now that we’ve got a baseline to work off of, let’s see who the most backwards pitchers are in the league — meaning, those pitchers that vary the most by count in terms of the frequency of their fastball usage. For this article, I am focusing just on starting pitchers*:
|Count||Lowest Fastball Percentage|
|3-0||Bronson Arroyo (38%)|
|3-1||Bronson Arroyo (34%)|
|2-0||Bronson Arroyo (32%)|
|3-2||Francisco Liriano (30%)|
|2-1||Bruce Chen (34%)|
|1-0||Bronson Arroyo (44%)|
|0-0||Brett Anderson (48%)|
|1-1||Felix Hernandez (48%)|
|0-1||James Shields (24%)|
|2-2||Francisco Liriano (25%)|
|0-2||James Shields (23%)|
|1-2||James Shields (26%)|
Some analysis I’ve seen defines pitching backwards primarily off of first pitches, or 0-0 counts. In that instance, the Athletics’ Brett Anderson takes the title, throwing a fastball in just 48% of 0-0 counts since 2010 (league average = 69%). James Shields does throw fewer fastballs in 0-0 counts than league average, but his 57% is only the 17th fewest in the league.
Where Shields really throws counter to expectations is in two strike counts as well as 0-1. Shields throws the fewest fastballs in the league in 0-1, 0-2, and 1-2 counts and throws the second fewest fastballs in 2-2 counts (31%). However, Shields pretty much pitches to expectations in those counts with the highest fastball frequencies league-wide: 3-0, 3-1, and 2-0. In fact, Shields throws more fastballs in each of those counts than league average.
So who throws the fewest fastballs in those counts?
The Reds’ Bronson Arroyo:
The chart plots the fastball frequencies of Arroyo, Shields, and Francisco Liriano against league average for each count. These were the three pitchers that showed up most often in the top-three for each count.
Liriano and Shields are very similar, consistently throwing more fastballs in “fastball” counts and throwing less in less frequent fastball counts. Arroyo, however, appears to be the mirror image of both Liriano and Shields, throwing near or above league average in less frequent fastball counts and throwing fewer (and, in many cases, the fewest) fastballs in the highest frequency fastball counts.
Now, this is the most basic way to define and examine pitching backwards. More advanced analysis could take into account not just the count, but the base-out state as well. Furthermore, you could also take into account the batter being faced. 3-0 counts may result in a fastball 94% of the time, but that number will likely drop when facing the eighth hitter in the National League.
For now, we can name Bronson Arroyo as the most backwards starters in the league.
*For obvious reasons, I excluded knuckleball pitchers from the analysis.
Commenters have convinced me that Arroyo is more the prototypical “backwards” starter, here, with Shields more typical in heavy fastball counts and more likely to throw off-speed in less-frequent fastball counts. I can live with that.
Bill leads Predictive Modeling and Data Science consulting at Gallup. In his free time, he writes for The Hardball Times, speaks about baseball research and analytics, has consulted for a Major League Baseball team, and has appeared on MLB Network's Clubhouse Confidential as well as several MLB-produced documentaries. He is also the creator of the baseballr package for the R programming language. Along with Jeff Zimmerman, he won the 2013 SABR Analytics Research Award for Contemporary Analysis. Follow him on Twitter @BillPetti.