Temporal Description of the Stolen Base in High School Softball

Submitted by Robin Lund, Travis Ficklin and Cassie Reilly-Boccia


The purpose of this study is to describe the temporal factors that determine the outcome of a stolen base attempt in high school softball. Two hundred and sixty-eight high school softball players were videotaped using a high-speed video camera to describe the typical steal time of a high school softball player. From the pool of subjects, 29 catchers, 81 pitchers and 2 middle infielders were studied to determine the average catcher pop time (time elapsed for the catcher to deliver the ball to the middle infielder at second base) under three different batter behavior conditions, pitch time and tag time (specific to location of the throw from the catcher). Repeated measures ANOVA indicated a fake bunt from the batter or a swing through by the batter significantly increased the catcher pop times when compared to the batter taking the pitch (p<0.05). A one-way ANOVA indicated that the catcher throws made to the low to inside region of second base resulted in significantly faster tag times (p<0.05). A chi-square analysis showed no effect of batter behavior on catcher accuracy. Coaches may use this evidence-based framework when deciding to attempt to steal second base to maximize run expectancy.


The ability to steal second base provides many significant tactical advantages for the offense. A successful theft of second base removes a force play at second, requiring the defense to throw a batted ground ball across the infield in addition to removing the possibility of a double play. Most importantly, the advancement of the baserunner from first to second without making an out provides the offense with three opportunities to drive the runner in with a base hit. In other words, successfully stealing second base will increase run expectancy.

Run expectancy is the average number of runs a team can expect to score during any specific base/out state. For example, a team may expect to score 0.56 runs any inning with no runners on base and zero outs. This value increases to 0.95 if the lead off runner reaches first base and increases again to 1.19 with a successful attempt of stealing second base. The increase of 0.24 in the run expectancy is the potential reward for attempting to steal second base, determined from run expectancy matrices for Major League Baseball games between 1999-2002 (12). Although the specific probabilities reported apply to professional baseball, trends in softball may be similar since the rules between the two sports are similar.

If the baserunner is thrown out at second base while attempting to steal, the run expectancy drops from 0.95 (runner at first base and zero out) to 0.30 (bases empty and 1 out), resulting in a net loss of 0.65. The potential risk (0.65 decrease) is almost three times greater than the potential reward (0.24 increase). Therefore, a team needs three successful stolen base attempts for every instance a player is caught stealing to break even. Other studies have confirmed that the break even success rate ranges between 70-75% of the time (7).

Regardless of the level of play, softball coaches must use caution and consider several variables such as the risk reward ratio when deciding to attempt to steal second base. On the offensive side, the coach must consider the running time of the baserunner (steal time). On the defensive side, the travel time of the pitched ball (pitch time) and the amount of time elapsed for the catcher to deliver the ball to second base (pop time) should be considered. Finally, the amount of time for the infielder to execute a tag (tag time) should be included. To date, these variables have yet to be reported for the sport of softball. In particular, there is a paucity of data available at the high school level. Therefore, the purpose of this study is to describe the temporal factors that determine the success or failure of a stolen base attempt in high school softball.



All data came from the analysis of 268 high school softball players who participated in two softball camps at the University of Northern Iowa. Of the 268 campers, there were 81 pitchers and 29 catchers who were observed separately. These camps were chosen because participation was limited to players currently playing at an advanced competitive level between the 8-12th grades. Additionally, the camps placed an emphasis on evaluation and feedback of many performance variables, which allowed the researchers to measure the temporal variables that play a role in attempting to steal second base. All procedures were reviewed and were considered exempt by the Internal Review Board at the University of Northern Iowa.


Several of the temporal variables were measured using video analysis. Video cameras (GC-PX1, Victor Corporation, Tokyo, JP) recording at 300 Hz captured baserunners executing steals, catchers making throws from home plate to second base under three different conditions as well as middle infielders making tags. All data were collected during the player evaluation sessions of the camps. All times in seconds were calculated from the recorded videos by multiplying the number of frames recorded for the timed event by 300-1 s. Similar procedures have been used previously (3,4). Additionally, all pitchers were evaluated separately during this same time period. Pitch velocity of each thrown fastball was measured using a radar gun (Jugs Cordless Pro, Jugs, Tualatin, OR).


To measure steal time, a camera was positioned perpendicularly to the running direction. Video was collected as the subjects ran from first base through a distance of 60 feet to establish the best possible time. The optical axis of the camera was aligned with the 60-ft line marked on the running lane. The camera recorded each trial and allowed to pan on the tripod in order to capture both the start and the crossing of the 60-ft mark. Steal time started when the trail foot left the ground and stopped when the most forward point of the trunk crossed the 60-ft mark. The number of frames between these two events was multiplied by 300-1 s to yield steal time in seconds.

To predict pitch time, pitch velocity was measured during the specific evaluation session. After the subjects warmed up, the researchers stood behind the pitcher and used the radar gun to measure the peak velocity of each pitch. The fastest pitch was recorded (miles per hour, mph). The kinematic analysis was made based upon pitch velocity and pitch distance (40 feet). The official distance between the pitching rubber and the apex of home plate is 43 feet, however, the location of the ball at release from the pitcher’s hand and reception of the ball by the catcher need to be considered. The average pitch is released in 5 feet in front of the rubber (2) and the catcher receives the ball approximately 2 feet behind the apex of home plate (10) resulting in a net pitching distance of 40 feet. Pitch times from several representative pitch velocities are shown in Table 2.

Catcher pop time was calculated from video analysis in a similar manner to steal time. A camera was positioned behind the catcher at a distance and orientation so that the catcher and the middle infielder were both visible. The pop time started when the pitch was received by the catcher and ended when the ball was received by the middle infielder at second base. The number of frames between these two events was multiplied by 300-1 s to yield catcher pop time in seconds. Each catcher made five throws under three different conditions. The first three throws were with a batter standing in the batter’s box. The fourth throw was performed with a batter faking a sacrifice bunt and pulling back the bat as the ball approached the strike zone. The fifth throw was made with the batter swinging through the pitch. The treatment conditions under which the five throws were executed were not randomized because the researchers were not in control of the evaluation portion of the camp procedures.

Tag time started when the ball was received by the middle infielder and ended when the ball-in-glove contacted the ground on the first base side of second base. The number of frames between these two events was multiplied by 300-1 s to yield tag time in seconds. Accuracy of the throw from the catcher was catalogued by the location of the caught ball in one of six regions (see Figure 1).

Statistical Analysis

Descriptive statistics were calculated for all variables. A repeated measures ANOVA was used to determine the effect of the batter behavior (take vs. fake bunt vs. swing through) on catcher pop time. Three throws were made with the “take” (i.e., batter stands motionless) condition, with the first of these throws compared to the other two conditions (fake bunt and swing through). Contrasts using a Bonferonni correction were used as the post hoc analysis (5). A one-way ANOVA was used to determine the effect of throwing accuracy on tag time with the Tukey test for post hoc analysis. Finally, a 3×2 chi-square test was used to test batter behavior with catcher accuracy. Alpha was set at p<0.05 for all tests. SPSS 21 (International Business Machines Corp, USA) was used for all analyses.


All descriptive statistics for each of the temporal variables can be found in Table 1. The Mauchley test of sphericity was rejected (W(2)=0.49, p=0.001) therefore a Greenhouse-Geisser penalty was applied to the repeated measures ANOVA (5). Despite the penalty, a significant batter behavior effect on pop time was observed (F(1.32, 36.9) = 5.03, p=0.02). The post hoc analysis showed that the average pop time was significantly longer when the fake bunt (p=0.045) or the swing through (p=0.008) was used when compared to the batter standing still. There was no difference between the fake bunt and swing through strategy on pop time (p=0.84). The one-way ANOVA indicated that the location of the throw (catcher accuracy) had a significant effect on the tag time (F(5,95)=17.5, p=0.001). Throws made in the “ideal” (below the waist on the 1st base side) region resulted in significantly shorter tag times when compared to all other regions. For the chi-square analysis, throws in the ideal region were considered “accurate” and all other regions were considered “inaccurate.” The chi-square showed that the batter behavior had no effect on catcher accuracy (c2(2)=0.67, p=0.71).


In regards to the baserunning side of the equation, the average steal time was 3.06 s. If the average pop time (2.16 s), pitch time (0.49 s) and tag time (0.33 s) are summed, the total time for the defensive portion of the play takes 2.98 s which gives the advantage to the defense. Interestingly, the success rate of Division I softball players was 79%, well above the break-even mark (9). The success rate in competitive high school players is unknown.

In order to apply the results of this study, each coach should consider the steal time of each player prior to game play. The player with the best maximal running speed may not have the best steal time. For example, the distance between first and second base is relatively short, where players may not reach top absolute speed, indicating an acceleration factor. To determine steal time, a coach may instruct the players to perform a steal attempt of second base and use a stopwatch to approximate the time. Coaches may also consider using the video feature on a smart phone or tablet for a more accurate steal time. The frame rate used with most products ranges between 30, 60 and 120 frames per second. This will provide an error range of +/- 0.033, +/- 0.017 and +/- 0.008 seconds respectively.

Once a coach knows the steal time of each player, the next step is to determine the pitch time of the opposing pitcher fastball. Pitch time is the elapsed time of the pitched ball to leave the hand of the pitcher and be received by the catcher. The variability of pitcher velocity plays a relatively small role in the big picture. For example, a pitcher throwing 50 mph will have a pitch time of approximately 0.55 s. Conversely, a pitcher with a velocity of 60 mph will have a pitch time of approximately 0.45 s, assuming that the ball is traveling 40 feet. This variance of 0.1 s accounts for 3-4% of the entire duration of the play. In other words, the amount of error occurring determining pitch time is negligible. If a radar gun is accessible, Table 2 can be used to determine the pitch time. Otherwise, a stopwatch may be used to determine the pitch time.

After the last warm-up pitch for each inning, the catcher will typically throw to second base allowing the coach to use a stopwatch to measure pop time. Pop time is the time elapsed for the catcher to deliver the ball to the middle infielder at second base. The stopwatch would be started when the catcher receives the ball and stopped when the thrown ball is received by the middle infielder. Although the scenario seems to be in favor of the defense, throwing out a runner attempting to steal is challenging. The results of this study suggest that the batter may engage in certain behaviors to disrupt the ability of the catcher to deliver the ball to second base. The fake bunt and the swing through in the current study added 0.12 and 0.13 s, respectively, to the overall pop time. Closer inspection suggests that the batter behavior did not affect the quality of the throw (i.e., no effect on accuracy) versus the ability to catch the ball cleanly. Large data ranges from the fake bunt and swing through trials indicate inconsistent catcher transitions resulting in double or “triple-clutching” throws. In other words, the catcher often failed to receive the ball cleanly, resulting in multiple attempts to retrieve the ball cleanly from the glove.

The last variable that a coach may consider is tag time. Tag time is the time elapsed for the middle infielder to swing the glove down in front of second base after receiving the ball from the catcher. This has received little attention but may have a significant influence on the outcome. Data from this study indicate that tag time is a function of catcher accuracy. For example, a perfect throw from the catcher is low and on the first base side of second base (Figure 1). Under these conditions, an infielder can make the tag in 0.21 s. If the catcher has a tendency to throw the ball higher, the coach can use a tag time of 0.40 s for decision-making purposes (average tag time from the two regions above the waist). If the catcher has a tendency to throw the ball to the 3rd base side of second base, the coach can assume a tag time of 0.50 s. If the catcher throws the ball consistently in the dirt, a tag time of 0.45 s may be used. If the catcher is accurate, a tag time of approximately 0.2 s may be assumed.

There are many other factors that a coach may consider before deciding to give the baserunner the steal sign. The field of Sabremetrics may provide evidence to support these considerations. For example, the coach may be more aggressive as the number of outs increase because the risk reward ratio becomes more favorable. In other words, the run expectancy is very low, increasing the difficulty to score from first base with two outs. If a coach attempts to steal second base and the player is caught stealing to end the inning, the run expectancy was low anyway therefore very little risk (1).

Another factor to consider is the quality of the catcher. According to Rosciam (11), at the Major League level, the top defensive catchers were able to throw out even the best base stealers at a rate greater than the league average. In summary, there may be catchers rarely tested based on superior pop-to-pop times and accuracy.

Coaches may also consider the accuracy of the pitcher. In general, coaches may feel more comfortable stealing when the pitcher is consistently throwing low pitches. At the Major League level, using the data from the PITCHf/x system, the location of the pitch has been demonstrated to have an effect on success rate of catchers (6). During the 2008-09 season, when the pitch was low (one foot or below), catchers had very low caught stealing percentages. Pitches that were above the belt and higher increased success dramatically with pitchouts giving the catchers a 50% chance of throwing out the runner. An analysis confirming these results was performed with the PITCHf/x data from the 2008-2013 seasons (13).

The relative productivity of the offense is another factor that may be considered. According to the risk reward ratio, the amount of risk increases as a team is more productive with runs scored. For example, if the team tends to have offensive innings where multiple runs are scored, then there is less inclination to steal. If the offense struggles to score runs, the risk reward ratio indicates a lower risk to steal as demonstrated at the Major League level. As run production decreases, the number of stolen bases per stolen base opportunity is increased (8). The inning of play also affects the risk reward ratio. A lower risk is to attempt to steal during one run games (or tie games) in the later innings. A higher risk, for example, is indicated in an attempt to steal by a team behind by two or more runs in the later innings (12).


In conclusion, the results of this study indicate an evidence-based framework providing information for coaching decisions about the stolen base. The temporal variables and techniques described provide the tools to predict the likelihood of success. Furthermore, previous work identified various scenarios which affect the risk reward ratio. Taken collectively, the informed coach should attempt the stolen base under optimal conditions to maximize run expectancy.


Coaches may create a system for quickly determining the defensive total time and compare to steal time. This would allow the coach to determine which players are able to steal second base with a lower risk of thrown out. The coach should consider using the fake bunt or the swing through to improve the probability of success for slower players. Finally, for the attempt to steal second base, the coach should consider the number of outs, the quality of the catcher, pitcher accuracy, offensive productivity, the inning of play and the score.

lund table


  1. Fox, D. (2006, August). Schrodinger’s bat: Valuing the running game. Retrieved from http://www.baseballprospectus.com/article.php?articleid=5455.
  2. Ficklin, T. (2012). The mechanics of ball velocity production in windmill-style pitching. (Unpublished doctoral dissertation). Indiana University, Bloomington, IN.
  3. Ficklin, T., Dapena, J., & Brunfeldt, A. (2009). A comparison of base running techniques in baseball. Paper presented at the 33rd Annual Meeting of the American Society of Biomechanics. Abstract retrieved from http://www.asbweb.org/conferences/2009/asb_schedule.htm
  4. Ficklin, T., Lund, R., & Schipper, M. (2014). A comparison of jump height, takeoff velocities, and blocking coverage in the swing and traditional volleyball blocking techniques. Journal of Sports Science and Medicine, 13, 73-78.
  5. Field, A. (2009). Discovering statistics using SPSS (3rd ed.). Los Angeles, CA: Sage.
  6. Greenhouse, J. (2010, July). Stolen bases and PITCHf/x. Retrieved from http://baseballanalysts.com/archives/2010/07/stolen_bases_an.php.
  7. Lederer, R. (2006, October). Net stolen bases: Leaders and laggards. Retrieved from http://baseballanalysts.com/archives/2006/10.
  8. Moore, J. (2012, November). The stolen base matters more now. Retrieved from http://www.fangraphs.com/blogs/the-stolen-base-matters-more-now/.
  9. National Collegiate Athletic Association (n.d.). Stolen bases per game. Retrieved from http://www.ncaa.com/stats/softball/d1/current/team/348.
  10. Reilly-Boccia, C., Ficklin, T., Lund, R., Hervas, J., & Beatty, K. (2013). Bat quickness and bat velocity in Division I softball players. Paper presented at the 37th Annual Meeting of the American Society of Biomechanics. Abstract retrieved from http://www.asbweb.org/conferences/2013/abstracts/447.pdf
  11. Rosciam, C. (2004). Professional thieves vs. the constabulary. The Baseball Research Journal, 33, 82-84.
  12. Tango, T., Lichtman, M. & Dolphin, M. (2007). The book: Playing the percentages in baseball. Dulles, VA: Potomac Books Inc.
  13. Weinstein, M. (2014, March). How pitch location affects caught stealing percentage. Retrieved from http://www.hardballtimes.com.




Figure 1. Tag times based on region. * Denotes larger tag time than the ideal region (p<0.05).