Introducing pWAR: A Predictive Wins Measurement for Pitchers

When WAR was first introduced, it attempted to answer a baseball question that has existed as long as teams have played professionally: how much can one player contribute to a team’s record? It is an ongoing debate to this day, but WAR is widely recognized as an excellent measure of overall performance. WAR numbers for major league players are regularly cited and recognized by both Major League Baseball and the Elias Sports Bureau. However, there is another question behind WAR, also long predating the existence of the statistic, and it’s a question that has yet to be answered by it. When a front office is deciding whether or not (or how aggressively) to pursue a player, they’re ultimately searching for the answer to one question — how many more games will we win if we get this guy?

Enter pWAR.

What is pWAR? 

pWAR, or Predictive Wins Above Replacement, is a statistic attempting to estimate, in WAR, how much value a pitcher will bring to a team in the future.

How is it calculated?

The formula is based on FanGraphs’ fWAR formula for pitchers, but uses Connor Kurcon’s pCRA (Predictive Classified Run Average) as a runs allowed estimator in place of xFIP. The initial formula can be seen below:

pWAR = {[(pRAAP9/dRPW) + RL] * IP/9}

The individual formulas that make up pWAR are as follows:

Dynamic Runs Per Win (dRPW) = {[([(18 – IP/G)*(AL or NL pCRA)] + [(IP/G)*pCRA]) / 18] + 2 } * 1.5

Predictive Runs Above Average Per 9 (pRAAP9) = AL or NL pCRA – pCRA

Replacement Level (RL) = 0.03*(1 – GS/G) + 0.12*(GS/G)

Predictive Wins Per Game Above Average (pWPGAA) = pRAAP9/dRPW

Predictive Wins Per Game Above Replacement (pWPGAR) = pWPGAA + RL

Predictive Wins Above Replacement (pWAR) = pWPGAR * (IP/9)

Doesn’t this already exist?

Short answer: no. Long answer: CRA, the mother stat of pCRA, does have a WAR component, labeled cWAR, but no such stat exists for pCRA. CRA, an ERA estimator utilizing batted ball data, is described by Kurcon as “a descriptive stat, due to the lack of predictability of some inputs such as xBA, xSLG, and xwOBA.” pCRA utilizes a similar formula but incorporates Barrels as a form of batted ball data, which correlate more with future performance, making pCRA a much more predictive statistic, and thus a much better base for pWAR.

How can/should this stat be used?

It’s an important distinction to make that any pitcher’s pWAR for a season is not meant to indicate that season’s performance, but rather what to expect from that player going forward. If you wanted to take the “projection” piece a step further, you could replace innings pitched — or any — numbers with projected numbers or even your own guesses for a future year to calculate potential future pWAR. You would just need to make sure to adjust all other numbers accordingly so that you don’t wind up mixing current and future numbers and averages. This is essentially what the pitcher’s results “should have” been or how they “will” perform, all things considered. It is also important to note that this formula can (and will) be updated and improved as more tests are run and more data is collected.

Why not just use pCRA? How is this better?

This isn’t meant as a replacement for pCRA or any other ERA estimator, as those are excellent stats to use! The point of creating pWAR was not to create a “better” or more accurate stat, but simply to present another angle from which to view a pitcher’s abilities.

What’s next for pWAR?

As mentioned above, pWAR is not set in stone. As it begins to be utilized and more data is collected, I will continue to adjust and update it as needed.

Tal Schulmiller is a 16-year-old high school junior. He pitches for a club team and is looking to play and study data science and analytics in college while pursuing a career in sabermetrics. He can be contacted via email or on Twitter.