Possible Side Impacts of Base Stealers

Having grown up playing catcher from Little League through college, I always recognized the temptation and situational changes that occurred in terms of strategy and pitch selection with runners on, particularly base stealers, versus with no runners on base.  As a catcher, my thought process with a base stealer on, is always to try and have my pitcher get the ball to me as quickly as possible.  An earlier study I read dealt with the correlation between pitchers’ times to home, and that being a much stronger factor in throwing out a base-stealer than catcher pop times.  Logically, in thinking of pitch selection as a way of controlling the run game, the quickest way to get the catcher the ball is with one’s fastest pitch.

To evaluate the impact of base-stealers I defined a base stealer as a player who swiped 20 plus bags in 2013.  Using Baseball Reference, I slotted 6 pairs of base stealers and their following hitters.  The criteria for those hitters being 400 plus plate appearances in the same slot in the batting order.  Nick Swisher however is an exception because he had 250 plus appearances behind both Michael Bourn and Jason Kipnis, but I decided to include him.  I should also note that all the statistics in this study are from 2013.  Using Baseball Savant’s Pitch f/x database I defined a fastball as a 4 seam, 2 seam, sinker, splitfinger, and cutter and every other pitch as a breaking ball.  I then compared the fastball and breaking ball rates with each hitter with a runner on 1^st or nobody on.

It is taken from granted that for a hitter the best pitch to hit is a fastball.  While there are many different approaches, one of the most common is “fastball adjust,” meaning the hitter always looks, or anticipates, a fastball as you get in the box.  However, if you recognize something different out of the pitcher’s hand, you should have more time to adjust.  Hitters are always fastball hunters first, that’s why we call 2-0, 3-1 counts “hitter’s counts” because they will most likely get a fastball and at the same time are sitting fastball.  As proof we used the probability of scoring a run per 100 pitches of a certain pitch above the prototypical average players.  The league average probability of scoring runs against what I defined as a fastball type pitch for every 100 pitches in 2013 was 0.0167 and for every 100 off speed pitches was -0.07.  That is over an 8/100ths difference in the likelihood of scoring a run above average, which added up over the thousands of pitches a player can see a year can make an impact.  Below are the 6 hitters I used for this study and their run probability rates against different pitches:

 

Name	Team	wFB/C	wSL/C	wCT/C	wCB/C	wCH/C	wSF/C	wKN/C
David Wright	Mets	1.74	-0.13	2.75	1.95	2.01	-4.82
Shane Victorino	Red Sox	1.53	1.29	-1.28	-0.52	-0.33	1.16	0.11
Dustin Pedroia	Red Sox	0.11	-0.72	3.87	1.86	1.47	9.6	-2.77
Nick Swisher	Indians	1.02	0.23	0.97	0.37	-0.55	-0.77	-4.47
Jean Segura	Brewers	0.19	0.45	0.82	-0.18	2.7	-5.61
Manny Machado	Orioles	0.17	0.23	1.15	-1.73	1.2	2.31	-1.34

 

As the data above supports, the best pitch to hit, the pitch a hitter is most likely to score more runs from, is a fastball.

So that being said, if a reputed, or habitual, base stealer is on base, then will the hitter at bat see an unusually high rate of fastball-like pitches?  With a higher rate of fastballs the hitter should therefore have a greater chance of success.  The theory being that an offense built more on speed and base stealing should see a higher rate of fastballs which then gives that team a greater probability of scoring more runs.

Now the total overall fastball rate for the league as a whole for the 2013 season was 57.8%.  The total fastball rates I arrived at were derived from simply taking the situational fastball rate and dividing it by the total pitch percentage or fastball percentage plus breaking ball percentage: fastball% / (fastball% + breaking ball%).

 

Base Stealer:	Following Hitter:	Runners on Fastball%:	Runners on Breaking Ball%:	Nobody on Fastball%:	Nobody on Breaking Ball%:	Total Fastball% with runner on:	Total Fastball% with Nobody on:
Norichika Aoki	Jean Segura	20.3001%	9.5322%	37.5552%	20.4325%	68.05%	64.76%
Jacoby Ellsbury	Shane Victorino	16.8302%	9.5191%	38.2237%	22.8165%	63.87%	62.62%
Daniel Murphy	David Wright	21.0498%	9.534%	33.5833%	18.3717%	68.83%	64.64%
Nate McLouth	Manny Machado	18.1782%	11.9856%	36.5961%	21.8138%	60.26%	62.65%
Shane Victorino	Dustin Pedroia	22.1729%	11.0694%	34.1647%	17.2532%	66.70%	66.45%
Michael Bourn/Jason Kipnis	Nick Swisher	19.8731%	12.0587%	31.4954%	21.4597%	62.24%	59.48%

 

Looking at the results, in particular the totals, there is no significant difference in percentages of fastballs vs off speed seen with a runner on first or not.  The biggest difference is a 4.46% difference with David Wright.  And David Wright scores 21.1 runs above average against fastball type pitches (wFB).  While maybe an extra 4.46% increase does not make a world of difference it still contributes to overall run production and as we know in baseball 1 run can decide a game and 1 game can decide a season.  However, it appears that my hypothesis is false and there is no significant difference in situational pitch selection with a base stealer on 1^st.

Now I will be the first to admit that there are definitely ways to improve upon the accuracy of my theory.  The biggest problem being that I could not find a database on the internet that allowed me the option of isolating at bats with only specific runners on, so the next best thing was Baseball Savant’s option of isolating at bats with the option of runners on certain bases or a combination thereof.  So all these plate appearances measured are just with a generalized runner on 1^st who could be anybody or nobody on at all.  This study is assuming that the runner on 1^st, for a majority of the time, is the base stealer who hits 1 spot in front of the selected hitter.  BIG assumptions I realize.  Also this is only covering 6 hitters in their 2013 season, which is a small sample size considering.  Unfortunately I did not have all the resources necessary for the most accurate representation for this study as a whole and on that note I hope many of you who perhaps have more available to you, can dig deeper and build on my theory.

This is my first time posting something like this so if you have any helpful questions/comments/criticism/advice please feel free to comment.  And if you have a way to more thoroughly complete this study please do so!  Thanks and I hope you enjoyed.