Run Expectancy

The following article appeared in the National Fastpitch Coaches Association’s (NFCA) publication Fastpitch Delivery in September, 2012.

By Jon Nachtigal

Sabermetrics, the statistical approach to professional baseball that revolutionized the sport, is also available for women’s college softball.

Sabermetrics is the term coined in 1980 by writer and statistician Bill James for his approach to baseball analysis. The term combines SABR, which stands for the Society for American Baseball Research, with metrics. Sabermetricians, as they would become called, searched for new and better ways to analyze baseball. After years of struggle, sabermetricians were eventually accepted into baseball’s mainstream. Their story was documented in the book and movie Moneyball.

Play-by-play data is the key to much of the analysis provided in sabermetrics. Play-by-play data documents each event that occurs on the field and allows for the generation of a run expectancy chart. Making this analysis possible for softball is the play-by-play data that the NCAA began publishing with the 2011 playoffs.

Compiling this data is a painstaking process, involving the analysis of 130 Division I playoff games and 7,636 individual plays. The analysis presented here by this author resulted in a first-ever run expectancy chart for women’s college softball.

Run Expectancy – 2011 Softball Playoffs

Base Runners

0 outs

1 outs

2 outs

Bases Empty

0.518

0.287

0.100

Runner at 1st

0.959

0.556

0.286

Runner at 2nd

1.032

0.890

0.356

Runners at 1st and 2nd

1.733

1.026

0.379

Runner at 3rd

1.000

0.915

0.545

Runners at 1st and 3rd

1.720

1.678

0.611

Runners at 2nd and 3rd

1.615

1.589

0.744

Bases Loaded

3.054

1.930

0.658

Run expectancy charts document the base and out situation, and the number of runs produced from the situation until the end of the inning. There are potentially eight base situations (bases empty, runner at first base, runner at second base, etc.) and three out situations (0 outs, 1 outs, 2 outs), resulting in a total of 24 base and out situations.

In the run expectancy chart for women’s softball from the 2011 playoffs, at the start of an inning where the bases are empty and there are no outs, a team can expect to score 0.518 runs, or just over half a run. Get the leadoff hitter on first base and the average number of runs scored jumps to 0.959, or almost one run on average. In other words, a team is much more likely to score if they get the leadoff runner on base.

A number of analyses can be done using a run expectancy chart. For example, teams in Major League Baseball have used run expectancy charts to demonstrate the costs and benefits of base stealing. This is why many teams utilize the stolen base less in professional baseball than in the past, because run expectancy analysis has shown that it’s a poor strategy for generating runs.

In women’s softball the necessary success rate for stealing with a runner on first base and no outs is 90%. While this analysis was performed through the use of the run expectancy chart and probability theory, it is fairly evident through the run expectancy chart alone that attempting a stolen base in this situation is a bad move strategically. With a runner on first base and no outs, a successful steal of second base increases the expected run total from only 0.959 to 1.032. That’s not much of an increase considering the risk. The risk involves getting the base runner thrown out, which would result in a situation of no runners on base and one out, where the run expectancy drops all the way down to 0.287.

The situation where attempting a stolen base becomes strategically viable is the steal of second base with one out. In this instance, a success rate of stealing second base at just a 60 percent rate makes the move a better risk.

The necessary success rate for stealing second base with two outs is 80 percent, meaning it’s likely best for most base runners to stay at first in this situation.

A weakness of this first-ever run expectancy chart for women’s softball is that only 130 games were available for analysis. While there were 1,781 instances of a runner at first base with no outs, leading to a high level of confidence for this situation, there were only 13 instances of a runner at third base with no outs, meaning the confidence level in the latter situation is considerably lower. More games are required for analysis in order to improve the accuracy of some of the numbers in the chart.