# Does Velocity Impact Reliever Success?

All throughout baseball, the trend has been that pitchers are throwing harder and harder. The average fastball according to Pitch f/x in 2015 was 92.4 MPH, the highest value of the Pitch f/x era (2007-present). Not surprisingly, average reliever velocity was higher than that of starting pitchers 93.2 vs 91.9. While a little over one MPH is not a major difference, it is notable because it proves that guys who throw exceptionally hard are better suited for success in the bullpen.

When looking at the Pitch f/x velocity leaderboard sorted for the hardest throwers in baseball one finds only two starting pitchers, Nathan Eovaldi and Noah Syndergaard, in the top 20. Often times when projecting relievers, the common narrative is “He throws hard, therefore he profiles to have success late in games.” While, there are many different bullpen roles, it is hard to argue against the success of fire ballers like Aroldis Chapman,Â Wade Davis, and Dellin Betances late in games. Using my formula for RWAR, which applies a value to relievers, I tried to see if there is a correlation between average velocity and relief pitcher success.

RWAR vs Velocity

The data set for this graph was all relievers in 2015 with the x-axis being velocity and the y-axis being their RWAR data. As the graph shows, there was a very slight positive correlation between the two variables, as evidenced by the red line.

As the table shows there is very little correlation between the success of a reliever and his average fastball velocity, Â asless than 1% of the variation can be explained by a change in velocity. However, as expected, there is some support for the belief that high velocity pitchers perform better out of the bullpen.

However, the above data set contains all players who have pitched in relief in 2015, including a number of position players and players who have only thrown a few innings, which may affect the results of the correlation. In an attempt to correct this, I limited the sample to pitchers who threw more than 30 innings out of the pen.

The sample was restricted from over 500 players down to 214 and by doing this, it showed a slightly stronger correlation between the two variables.

As it can be seen by the values in the table, when I restricted the sample, the slope doubled and the amount of the variation explained by the correlation also doubled from 0.6% up to 1.3%. While this supports the argument that relievers are more likely to succeed, it proves that this is just one of many factors of the reliever valuation.

Using the same idea as the previous experiment, I replaced the dependent variable of RWARÂ with ADREIP. For those who do not know what ADREIP means it is:

Simply stated what this value shows is the number of runs a particular pitcher saves his team on a per inning basis, adjusted for the leverage of the situations he has been in. Using the constrained sample size of pitchers who threw more than 30 innings out of the pen I came up with the following correlation.

Once again looking at the chart and the resulting trend-line there is a slight positive correlation. Let’s once again take a look at the numbers associated with this chart.