HQ NOW: Experts Leagues Winner's Circle, 2023RESEARCH: Ball and strike rates for starting pitchers
@StephenNickrand
July 20th, 2015 11:05pm
There is a statistic printed at the bottom of many box scores to which most readers pay very little attention: the number of balls and strikes thrown by pitchers during a game. It’s easier to focus on a pitcher’s total walks and strikeouts instead of the components that go into them.
But when you’re wondering if a pitcher’s control is legit, it’s important to dig deeper.
Our prior research on swinging strike rate and first-pitch strike rate confirmed that both are useful as predictive tools for a pitcher’s strikeout and walk rates. Those measures also can be found at the bottom of most box scores.
In addition, our detailed research on the stats and skills posted by starting pitchers in different ball-strike counts confirmed that the frequency with which a pitcher falls into certain ball-strike counts greatly impacts the likelihood that an at-bat will end in a walk or strikeout—and even impacts the likelihood that a batted ball will fall for a hit. For example, we found that ball-strike counts that started 0-1 resulted in a walk in just four percent of plate appearances, and the hit rates of starting pitchers elevate dramatically in extremely hitter-friendly counts.
Hypothesis
It seems intuitive that a pitcher who throws more balls as a percentage of total pitches will have a higher walk rate (BB/9) than his counterparts.
Let’s also propose that ball% can be used as a measure to validate a SP’s control and to predict future movements in control rates.
Method
We’ll begin by taking an updated five-year look at the relationship between a pitcher’s control (BB/9) and dominance (K/9) rates with other stats and skills.
Then we’ll examine ball percentages to see how closely they correlate with strikeouts and walks, whether they have predictive value, and whether they are more likely to regress to a SP’s own ball% or to the MLB norm.
As a reminder, correlations can range from -1.0 to +1.0. The strongest correlations are at the extremes; they get weaker as they get closer to zero:
+0.70 to +1.00 Strong positive relationship +0.40 to +0.69 Moderate positive relationship +0.20 to +0.39 Weak positive relationship +0.19 to -0.19 No or negligible relationship -0.20 to -0.39 Weak negative relationship -0.40 to -0.69 Moderate negative relationship -0.70 to -1.00 Strong negative relationship
Results
First, let’s take a look at various stats and skills to see how they are correlated with a pitcher’s rate of walks and strikeouts:
Control Rate Correlations
Indicator Indicator Correlation ========= ========= =========== As indicator goes up, walks go up (and vice-versa): Ctl Ball% +0.70 Ctl WHIP +0.68 Ctl ERA +0.57 Ctl H% +0.21 Ctl HR/9 +0.15 Ctl hr/f +0.12 Ctl S% +0.05 Ctl O-Contact% +0.03 Ctl FAv +0.02 As indicator goes down, walks go up (and vice-versa): Ctl Strike% -0.70 Ctl Swing% -0.53 Ctl O-Swing% -0.49 Ctl FpK% -0.35 Ctl Zone% -0.29 Ctl Win Pct -0.28 Ctl SwK% -0.14 Ctl Z-Contact% -0.12 Ctl Age -0.10 Ctl GB% -0.07 Ctl Z-Swing% -0.04 Ctl Contact% -0.00
It does not come as a surprise to see that WHIP and ERA have strong negative correlations with a SP’s Ctl, since walks directly impact WHIP and make the risk of ERA blowups greater. Nor is it revealing that as strike% declines walks go up significantly.
The above chart does show that ball% and strike% have the strongest correlation to a SP’s control rate out of all indicators tested.
Now let’s take a look at the indicators most closely correlated with a SP’s dominance (K/9) rate:
Dominance Rate Correlations
Indicator Indicator Correlation ========= ========= =========== As indicator goes up, strikeouts go up (and vice-versa): Dom SwK% +0.70 Dom FAv +0.29 Dom H% +0.27 Dom O-Swing% +0.18 Dom Strike% +0.15 Dom Win Pct +0.15 Dom FpK% +0.08 Dom S% +0.06 Dom Swing% +0.06 Dom hr/f +0.02 Dom Zone% +0.01 As indicator goes down, strikeouts go up (and vice-versa): Dom Age -0.05 Dom HR/9 -0.08 Dom GB% -0.12 Dom Z-Swing% -0.13 Dom Ball% -0.15 Dom ERA -0.18 Dom WHIP -0.21 Dom Z-Contact% -0.55 Dom O-Contact% -0.56 Dom Contact% -0.72
As we already know, there is a consistently strong positive correlation between swinging strike rate (SwK%) and strikeouts (Dom). There is also a moderate-to-strong negative correlation between strikeouts and the level of contact made by batters—both inside and outside the strike zone.
Unlike with a SP’s control rate, strong correlations do not exist between ball/strike% and dominance rate.
***
Let’s establish some ball% and strike% baselines so that we know the marks typically shown by starting pitchers.
From 2010 to mid-2015, the ball and strike percentages of starting pitchers have stayed extremely steady, both from season-to-season and by league:
Historical Ball and Strike Percentages—Starting Pitchers
Season League Ball% Strike% ====== ====== ===== ======= 2010 MLB 37% 63% 2011 MLB 37% 63% 2012 MLB 36% 64% 2013 MLB 36% 64% 2014 MLB 36% 64% 2015* MLB 36% 64% *through July 5, 2015
Just under two-thirds of pitches thrown by starting pitchers are strikes. A little more than one-third are balls. There are no significant differences by league.
Sixty percent of SP produce a ball% between 34% and 37%:
Ball% Stability
Does ball% regress to league norms or do players set their own baselines?
Using our data pool from 2010 to mid-2015, there were 1,114 instances of SP with at least 40 IP during a season.
Their average ball% vs. MLB ball% variance was +/- 1.91%. By comparison, the average variance between a SP’s ball% and his own career ball% was smaller, at +/- 1.71%.
That indicates that SP tend to set their own ball% baselines more often than regressing to an overall norm. This make sense, as we typically view balls, strikes, and walks as areas that are under a SP’s control.
Predicting Control Rate from Ball% and Strike%
Using data from 2010 to mid-2015, we can establish a SP’s expected control rate (xCtl) using his ball%. The following table confirms the strong correlation between ball% and Ctl:
Expected Control Rate by Percentile
Ball% 10th 25th 50th 75th 90th
===== ==== ==== ==== ==== ====
<30% 0.8 0.8 1.1 1.3 1.4
31% 1.3 1.4 1.5 2.0 2.1
32% 1.2 1.5 1.7 2.0 2.3
33% 1.5 1.6 1.9 2.2 2.5
34% 1.7 2.0 2.3 2.6 2.9
35% 1.8 2.1 2.3 2.6 2.9
36% 2.0 2.3 2.7 3.1 3.4
37% 2.3 2.7 3.0 3.3 3.7
38% 2.6 2.8 3.2 3.6 4.1
39% 2.9 3.2 3.5 4.0 4.5
40% 3.1 3.5 3.9 4.5 4.9
41% 3.5 3.9 4.6 5.1 5.5
>42% 4.2 4.7 5.1 5.6 6.6If you flip the above data and use strike% instead of ball%, the same ranges of values apply:
Expected Control Rate by Percentile
Strike% 10th 25th 50th 75th 90th
======= ==== ==== ==== ==== ====
>70% 0.8 0.8 1.1 1.3 1.4
69% 1.3 1.4 1.5 2.0 2.1
68% 1.2 1.5 1.7 2.0 2.3
67% 1.5 1.6 1.9 2.2 2.5
66% 1.7 2.0 2.3 2.6 2.9
65% 1.8 2.1 2.3 2.6 2.9
64% 2.0 2.3 2.7 3.0 3.4
63% 2.3 2.7 3.0 3.3 3.6
62% 2.6 2.8 3.2 3.6 4.1
61% 2.9 3.2 3.5 4.0 4.5
60% 3.1 3.5 3.9 4.5 4.9
59% 3.5 3.9 4.6 5.1 5.5
<58% 4.2 4.7 5.1 5.6 6.6Ball% Outliers
From 2010 to 2014, 83 SP posted a control rate that was +/- 1.0 of their expected control rate (xCtl) based on their ball%, using the 50th percentile of the above table as xCtl.
Of those, 35 SP had at least 40 IP as a SP during the following season:
YEAR ONE YEAR TWO
Name Ball% Ctl xCtl Diff Ball% Ctl
==================== ===== === ==== ==== ===== ====
---------------Control Rate Underperformers--------------
Harden, Rich 39% 6.0 3.5 +2.5 38% 3.4*
Sanchez, Jonathan 40% 5.9 3.9 +2.0 44% 7.4
Jimenez, Ubaldo 40% 5.5 3.9 +1.6 37% 2.8*
Volquez, Edinson 40% 5.4 3.9 +1.5 40% 5.2*
May, Trevor 36% 4.2 2.7 +1.5 35% 2.0*
Deduno, Samuel 41% 6.0 4.6 +1.4 39% 3.4*
Volquez, Edinson 40% 5.2 3.9 +1.3 39% 4.1*
Ramirez, Erasmo 38% 4.4 3.2 +1.2 36% 3.2*
Billingsley, Chad 37% 4.0 3.0 +1.0 37% 2.7*
Norris, Bud 39% 4.5 3.5 +1.0 37% 3.4*
Bedard, Erik 38% 4.4 3.2 +1.2 37% 3.5*
Petit, Yusmeiro 30% 2.3 1.1 +1.2 32% 1.5*
Santiago, Hector 38% 4.3 3.2 +1.1 38% 3.8*
Moore, Matt 37% 4.1 3.0 +1.1 40% 4.6
Bedard, Erik 37% 4.0 3.0 +1.0 38% 4.4
Paulino, Felipe 39% 4.5 3.5 +1.0 37% 3.6*
Doubront, Felix 37% 4.0 3.0 +1.0 39% 3.8*
Chacin, Jhoulys 38% 4.2 3.2 +1.0 36% 2.8*
Darvish, Yu 38% 4.2 3.2 +1.0 38% 3.4*
---------------Control Rate Overperformers---------------
Litsch, Jesse 41% 2.9 4.6 -1.7 40% 3.5*
Millwood, Kevin 37% 1.3 3.0 -1.7 37% 3.1*
Anderson, Brett 38% 1.8 3.2 -1.4 36% 2.7*
Francis, Jeff 37% 1.8 3.0 -1.3 37% 3.0*
Alvarez, Henderson 34% 1.1 2.3 -1.2 36% 2.6*
Gomez, Jeanmar 39% 2.3 3.5 -1.2 40% 3.5*
Haren, Dan 34% 1.3 2.3 -1.1 35% 1.9*
Kuroda, Hiroki 37% 1.9 3.0 -1.1 36% 2.7*
Guthrie, Jeremy 38% 2.2 3.2 -1.1 37% 2.8*
Gibson, Kyle 41% 3.5 4.6 -1.1 39% 2.9
Shoemaker, Matt 36% 1.6 2.7 -1.1 36% 2.0*
Cahill, Trevor 41% 3.6 4.6 -1.1 39% 3.3
Villanueva, Carlos 37% 2.0 3.0 -1.0 36% 2.5*
Capuano, Chris 37% 2.1 3.0 -1.0 36% 2.6*
Haren, Dan 36% 1.7 2.7 -1.0 35% 1.6
Cahill, Trevor 40% 2.9 3.9 -1.0 41% 3.6*
*year two Ctl moved in direction of year one xCtlIn 29 of the 35 cases (83%), the SP’s control rate in year two moved in the direction of his year one xCtl.
Those results show that SP with wide variances between Ctl and xCtl will overwhelmingly experience a correction in the direction of xCtl—as calculated by using ball%—the following season.
Conclusions
There are strong correlations between a SP’s ball% and strike% and the number of walks he allows.
- Ball% provides the closest link to a pitcher’s Ctl out of all indicators studied.
- SwK% provides the closest link to a pitcher’s Dom out of all indicators studied.
- There are weak correlations between a SP’s ball% and strike% and the number of strikeouts he gets.
- Ball% more often regresses to a SP’s career norm than it regresses to an MLB norm.
- SP with wide variances between Ctl and xCtl will overwhelmingly experience a correction in the direction of xCtl—as calculated by using ball%—during the following season.
For more information about the terms used in this article, see our Glossary Primer.
