# Introducing Reliever WAR

As a huge math nerd and an even bigger baseball fan, I have fallen in love with the new statistics. I used to spend hours on end looking up players on Fangraphs.com trying to find trends in an individual player’s stats. That is part of what has led me to writing for this site as I get to use this massive amount of information to write my articles. While there are a number of different metrics out there one place where I think the genius minds around the game have dropped the ball is when it comes to relief pitching.

This will not be an article ranting about why baseball needs to lose the closer job, I’ll save that for another day.

I have tried for a while to look for the best way to grade relievers because as a college reliever myself I know that ERA and many other stats do not tell the full story. Often times someone will come into the game with runners on and give up those runs. He did not do his job but his ERA did not suffer. I have always felt this was a major shortcoming of how we look at the game. With that thought in mind I have tried to develop a metric, which follows a WAR-esque scale to look at the most valuable relievers in baseball.

The role of every relief pitcher is different as they all enter the game in different situations of runners on base, hitters they are facing, and the score of the game. My belief is, in order to accurately quantify the value of a relief pitcher we need to take all of these factors into consideration. Enter RE24 (Run Expectancy Based On The 24 Base-Out States) and LI. For those of you who do not know, RE24 is based off of the different base state situations. Each situation (like runner on first no outs) is given a value based on the number of runs that are expected from that situation. The player is then credited or dinged for how he changes that situation.

For example, if a pitcher strikes a batter out with no one on or out, his RE24 will go up .218 (.461-.243) while conversely the hitter’s will go down the same amount. The change is related to the differnce in the base state from no one on, no one out to no one on one out, and so forth. If you would like to read more check out this article by Fangraphs, which explains it. LI stands for Leverage Index and what this attempts to do is quantify the importance of the situation a particular pitcher comes in to deal with. This looks at aspects like the inning, the score, and the hitters that he is facing. An LI of 1.0 means a pitcher pitched is mostly neutral in leverage situations while anything above 1.0 means more difficult situations occurred and less than one is obviously easier. Yet again here is a Fangraphs link explaining it further.

I think that RE24 does the best job of handling the situation a pitcher comes into, as if he comes into that same situation with runners on and gives up the runs his RE24 will be hurt. The leverage index I felt was the perfect way to deal with the importance of the situation a pitcher was coming in to. In order to combine these I think the best way was to multiply them together, which created what I call ADRE or adjusted RE24.

Using these two metrics we could create a new way to evaluate relievers. In order to do this I looked at a few other metrics that have been used to evaluate other types of players. The main one that I wanted to model this after was wRAA, which takes a players’ wOBA-league wOBA and multiplies it by PAs. So with that idea in mind, I calculated the league average ADRE per inning pitched by totaling the ADRE values for every player who pitched out of the bullpen in a given season and dividing that by all of the RP (Relief Pitcher/Pitched) innings in baseball that season. Then for each pitcher I calculated their ADRE/IP and subtracted the league average value I got. Then by multiplying by innings pitched it gave me their runs above average. This runs above average number could be calculated into a “wins above average” metric by using the standard 10 runs equals one win calculation that is used in typical WAR calculations.