Wins Above Replacement (WAR) attempts to measure player value by estimating the number of wins a player contributes relative to a hypothetical "replacement-level" player. As an aggregate statistic it's a great proxy for value and an easy way to compare and rank players across the league; But WAR on its own doesn't tell us how a player provides value. For instance, two players may both have the same career WAR, but accrue it through drastically different ways. Player A may have a defense-first skillset whereas Player B could be valuable based on their hitting ability alone. If we look only at total WAR, we miss this nuance. Here's a visualization of what I mean. We'll break it down in the following paragraph.
The equation for WAR is comprised of three ingredients: Batting, Fielding, and Baserunning. Each of these ingredients allow us to break down a player's value into distinct skill areas and create visualizations like the one shown above. WAR then becomes the sum of indepently modeled statistics like batting runs, defensive runs saved, and baserunning runs. Fangraphs and Baseball Reference each have their own ways of calculating WAR (fWAR and bWAR respectively). Shown below is the position-player fWAR equation:
You'll notice some additional terms in the WAR equation that I didn't mention. These are adjustment terms for position and league that allow us to compare players across positions, scoring environments, and eras.
Since WAR is a counting stat, it favors players with longevity. To compare raw ability rather than accumulation, we divide total WAR by games played, yielding a rate stat: WAR per G (WAR/G). Breaking this down further into Batting WAR/G, Fielding WAR/G, and Baserunning WAR/G we can generate a three dimensional player "skill profile".
When we compare these rate stats against league-average values, we can see whether a player contributes above or below expectations in each area. The visualization above shows this breakdown by computing percentiles for each skill area vs league averages for MLB players from 2005-2024.
WAR is far from a perfect measure of player value. Defensive metrics in particular are notoriously difficult to model, noisy, and more uncertain when compared to offensive metrics. Readers should interpret the visualization above as a rough picture of skill profile, rather than a precise measurement.