WAR — What is it good for?
A KPI Case Study — Wins Above Replacement
This article was inspired by Corsair’s Institute’s Baseball Visualization Challenge 2016, #BaseballViz2016 on Twitter. Baseball is a sport rich in visualizations, statistics, and data. It is also rich in KPI, Key Performance Indicators.
Baseball fans have been inventing KPI since the days of Henry Chadwick. Almost from the first game, new statistics like ERA — the Earn Run Average, have been a major part of the game. Nearly a century later, the sport would turn to Sabermetrics and a new enthusiasm for KPI.
Sabermetrics are named for the Society of Baseball Research -SABR. A Sabermetric, specifically OBP — On Base Percentage. OBP was recognized as a superior metric to the Batting Average, one of the early baseball statistical inventions. This stat was core to this century’s love of Moneyball until it too became overemphasized. More on that via this article and hundreds of other offerings across the web.
Wins Above Replacement
Today, we have WAR or Wins Above Replacement and it is not just for Baseball anymore. Other sports are quickly adopting this golden metric for their sports as well. What is WAR? To find out, you could follow one of these numerous links below… but I doubt you will return with anything more than a vague understanding.
As OBP was to Batting Average, which by the way was a simple addition of the other missing ways to get “on base” — like a walk, WAR is to OBP. Well sort of…
First you need to adjust OBP to weight multiple base hits… like Slugging Average. Then you need to account for stolen bases. Next we benchmark this new measure vs the average for other players in the league. Then we need to add a weighting adjustment for the players position… then we need to account for fielding… and none of this takes pitching into account. Are you following…?
The only thing missing here is a reference to the All-Seeing Eye of Mordor, a handful of Hobbits, and several hours of New Zealand landscape. WAR as a metric is an attempt to build one metric “to rule them all”. As a result, it is very complicated.
So what is it good for?
WAR, as it is engineered, is a metric designed to measure the number of extra wins (or fewer) that a team receives when they opt to sign one player over any other average player at that position. It has the added benefit of being a benchmarked measure, so in theory it allows you to compare players from different time periods.
That is one take.
Another view on WAR recognizes that when statisticians get bored, complex models follow. These models, a form of statistical masturbation (yeah, I said it), are designed for the glory and pride of the designer. They are a misguided attempt to oversimplify a complex game, in a complex landscape, into a simple metric with an overly complex equation.
That does not mean they aren’t useful. It is only an observation that having one metric, instead of several, may not actually be any better. What is gained by simplifying the number of metrics you need to calculate is a trade off. Lost in the ‘one true metric’ are varying aspect of the game that can simply not be quantified in this fashion.
Which take do you agree with? Leave your comments below. Thanks for reading! If you enjoyed it, be sure to recommend this article ❤.
Baseball Visualization Challenge is a concept created by Corsair’s Institute. Neither this competition nor Corsair’s Institute is not associated with the MLB in any way. Our visualization expertise is second to none. Our softball team loses in games where the other team forfeits.
