RESEARCH: Starting pitcher ball-strike counts, part 1

The toolbox we use to analyze and forecast the performance of starting pitchers has evolved significantly over the years.

To assemble a really good starting rotation, we know not to target guys based on surface stats like wins and ERA. Instead, we use measures like Ctl (BB/9), Dom (K/9), and groundball rate (GB%).

We also know to discount pitchers with low hit rates (H%) and low home runs per flyball rates (hr/f) because research has shown that those marks tend to regress to consistent levels. As those low marks increase, their ERAs and WHIPs will rise.

One area that has received little attention involves the ball-strike counts that pitchers encounter during a plate appearance and how those counts can impact pitching results.

Here, we will show the importance of using ball-strike counts as an analytical tool for starting pitchers and uncover some significant findings in the process.

It seems logical that SPs who can avoid hitter-friendly 3-0 or 3-1 counts would have better results than those who are consistently behind in the count. We'll test that assumption.

We also wondered if regressive indicators like H% and hr/f actually vary based on ball-strike count, which could help us explain why some pitchers have high marks that are well above the 30% H% and 10-11% hr/f baselines we use.

Ball-strike count frequencies

To find out, we took a look at the ball-strike counts that starting pitchers (minimum 40 IP) found themselves in during 2013. We excluded 0-0 counts, since every plate appearance starts with that count.

Here were the frequency of their ball-strike counts as a percentage of total counts (excluding 0-0 counts):

As a percentage of all pitch counts—excluding 0-0 counts—it should not come as a surprise that 0-1, 1-1, and 1-0 counts are the most frequent.

This data also gives us baselines that we can use to analyze ball-strike count frequency on an individual level, an exercise we will undertake in part two of this series.

Stat/skill variance by ball-strike count

Next, let’s look at how stats and skills vary based on the ball-strike counts that a pitcher encounters during a plate appearance.

We’ll start by looking at strikeout and walk rates in plate appearances that reach the following ball-strike counts:

  • Starting Pitcher BB% and K% by Ball-Strike Count*
    Count         As % of Total  BB%  K% 
    =====        ==============  ===  ===
    Through 0-1        19%        4%  26%
    Through 1-1        15%        8%  23%
    Through 1-0        15%       13%  15%
    Through 1-2        11%        5%  39%
    Through 2-2         9%       11%  34%
    Through 2-1         8%       17%  19%
    Through 0-2         8%        3%  42%
    Through 2-0         5%       27%  11%
    Through 3-2         5%       29%  24%
    Through 3-1         3%       40%  11%
    Through 3-0         2%       57%   7%
    Total             100%        7%  19%
    *starting pitchers 2013, min 40 IP

To illustrate the data, the next two graphs represent the strikeout (K%) and walk (BB%) rates that starting pitchers owned in 2013 during plate appearances in which they faced the ball-strike counts shown. The percentages reflect the portion of plate appearances that resulted in the outcomes shown.

The above data validates the perceived strong link between getting ahead of hitters and strikeouts. It also shows us that a SP’s strikeout rate increases by more than 10 percentage points when he starts the count 0-1 versus 1-0.

Similarly, our analysis of walk rates also yields some interesting takeaways:

Here, we can see the importance of first-pitch strikes, as ball-strike counts that started 0-1 resulted in a walk in just four percent of plate appearances. 

Excluding intentional walks (IBB) from these calculations altered our findings in only two ball-strike counts: 3-0 and 2-0. Doing so in plate appearances that had 3-0 counts lowered the BB% in those plate appearances from 57% to 54%. Removing IBB in plate appearances that had 2-0 counts dropped the BB% from 27% to 25%.


Let’s move on to balls in play data so that we can test the assertion that hit rates (BABIP, or H%) and home runs per flyball (hr/f) rates do not regress to their established norms in all ball-strike counts.

Here is a table summarizing balls-in-play data at the ball-strike counts listed:

  • Starting Pitchers Stats & Skills by Ball-Strike Count*
    Count   As % of Total  ERA   H%   HR%   GB%  LD%  FB%  hr/f
    ======  =============  ====  ===  ====  ===  ===  ===  ====
    At 0-1        19%      3.07  30%  2.0%  47%  21%  31%   10%
    At 1-1        15%      3.47  31%  2.2%  46%  21%  33%   11%
    At 1-0        15%      4.55  31%  2.8%  43%  21%  35%   12%
    At 1-2        11%      2.38  31%  1.6%  49%  21%  30%   10%
    At 2-2         9%      2.76  32%  1.8%  47%  21%  32%   10%
    At 2-1         8%      4.14  32%  2.5%  43%  22%  35%   10%
    At 0-2         8%      2.18  31%  1.4%  49%  21%  30%    9%
    At 2-0         5%      5.64  32%  2.8%  41%  22%  37%   13%
    At 3-2         5%      3.66  34%  1.7%  43%  21%  35%   10%
    At 3-1         3%      5.39  32%  2.5%  40%  21%  39%   14%
    At 3-0         2%      7.73  37%  1.9%  40%  23%  37%   13%
    Total        100%      3.90  30%  2.6%  45%  21%  34%   11%        
    *starting pitchers 2013, min 40 IP

Let’s illustrate this data for each indicator, starting with H%:

This tells us that H% spikes in 3-0 counts and increases significantly in 3-2 counts, which could help explain why some starting pitchers can’t seem to regress to the overall 30% H% norm.

Home runs allowed as a percentage of total batters faced also showed some significant variance at certain ball-strike counts:

1-0, 2-0, 2-1, and 3-1 counts are especially deadly for home runs allowed. The modest HR% in 3-0 counts is the result of many of those plate appearances ending in walks.

The groundball rates of starting pitchers also fluctuate based on ball-strike counts. Pitchers can induce groundballs at a much higher rate when they are ahead in counts:

Line drives jump a bit in 3-0, 2-0, and 2-1 counts, but are otherwise pretty consistent across counts:

Since groundball rates increase when pitchers are ahead in counts and line drives remain consistent in most counts, it seems logical that a SP’s flyball rate would spike when he is behind in counts. Our research confirms that assertion:

Finally, as we did with H%, let’s test the regression of hr/f to see if it really holds true in all ball-strike counts:

This suggests that pitchers who are in a lot of 3-1 and 2-0 counts will have a higher percentage of their flyballs end up as HR than those who are able to avoid being in those counts more than usual.



Given all these outcomes, we can see several general findings that could and should influence how we think about SPs in our roster management. In particular:

  • First-pitch strikes have a profound impact on the stats and skills posted by starting pitchers
  • Starting pitchers who get ahead of hitters see their walks go down, their strikeouts go up, their home run rates dip, their groundball rate rise and their home runs per flyball rates decline
  • Starting pitchers that fall behind hitters see their walks go up, their strikeouts go down, their home run rates rise, their groundball rates decline, and their home runs per flyball rates soar
  • 3-0 and 3-1 counts are especially deadly to the above indicators
  • Starting pitchers see their hit rates progressively increase when reaching two-ball and three-ball counts, and they soar during plate appearances that reach 3-0 counts

In part two of this research, we will take a look at these findings on a multi-year basis to validate our research.  We will also look at this data on an individual level to identify starting pitchers whose ball-strike count frequencies differ significantly from the norms established here, to see if that variation might explain their under- or over-performance.  We’ll also determine if ball-strike count frequencies tend to be consistent on an individual basis from year-to-year.

Click here to subscribe

  For more information about the terms used in this article, see our Glossary Primer.