WAR Misconceptions
Wins Above Replacement (WAR) is currently the best statistic available at capturing how much value a player contributes to his team, but that doesn’t necessarily mean that it’s perfect or that it doesn’t have limitations. Like with any other statistic, it is possible to misuse WAR and to draw faulty conclusions from it.
Here are some of the most common misconceptions and limitations of WAR:
1) Replacement Level. The most common misunderstandings about WAR stem from people not fully understanding what “Wins Above Replacement” means. For a full explanation of this concept, see our page on replacement level. But for a quick explanation, if a team’s starting shortstop gets injured, they will have to replace that player with a generic bench / Triple-A player. WAR attempts to measure how much value players contribute above that level of production.
So if you look at a team’s roster and WAR total, you’ll notice that it won’t line up with their win total in the standings. That’s because a team of “replacement players” would theoretically win somewhere around 43 games over the course of a season (varies depending on the year), so you need to add that baseline to a team’s WAR total in order to estimate their win total.
2) Measuring the past, not “true talent level”. WAR totals measure past performance, and shouldn’t necessarily be used as predictive tools. If a player was worth +5 wins last year, it’s entirely possible that he’ll be worth +5 wins again next year…but then again, depending on the rest of that player’s history, he could also dramatically under- or over-perform that level. If you want to get an idea about a player’s true talent win level, look at their past three seasons of WAR totals and average them out, weighing the most recent seasons more heavily (5/3/1 weighting is standard).
3) Defensive data. To account for defensive value, WAR does a few different things. For non-catchers, it uses UZR values for every year from 2002 on, and it uses TZ to account for defense in earlier years. For catchers, it includes both rSB and CPP, as those two stats measure a catcher’s effectiveness at preventing stolen bases and passed balls. Catcher WAR does not include pitch framing (yet).
But as much as we love UZR, it is important to note that all public defensive metrics are still imperfect. The data is collected by BIS scouts and is manually input, so there are likely some small human errors included. BIS is the best publicly available data source, though, and UZR goes to lengths to ensure that it is accurately representing defensive value. Due to the uncertainties in all defensive stats these days, you should not take WAR as perfectly precise right down to the decimal point — under the right circumstances, a +3.8 win player could be more valuable to their team than a +4.1 win player.
4) Environment is important. WAR totals are adjusted in order to make them park neutral, but that still doesn’t necessarily mean it removes all environmental biases from them. For instance, if an extreme ground ball pitcher gets traded from a superb defensive team to one of the worst in the league, it’s entirely possible that their performance will get worse and their win value will drop. A player’s environment and skill set interact with one another, so even with park adjustments, this is something to keep in mind.
5) Context neutral. For both pitchers and position players, WAR is entirely context neutral. This means that it doesn’t take into account that some hits are more important in games than others; sometimes, a single can be more valuable than a triple, depending on the time and context. And it doesn’t take into account that some relievers pitch in key, high leverage situations, making their results more important than, say, the appearances by the team’s long reliever.
This should not be viewed as a flaw in WAR; it’s merely one of its features. You can find a player’s contextual “win” total by looking at WPA, so WAR gives the other side of the picture — and in many ways, it’s the more important side. If we want the most accurate measure of how much value players add to their team, we should use WAR.
Links for Further Reading
Piper was the editor-in-chief of DRaysBay and the keeper of the FanGraphs Library.