Expected Weighted On-Base Average — wOBA recomputed using batted-ball quality instead of actual outcomes.
Hitters and pitchers are not in full control of their outcomes. A 105 mph line drive directly at the shortstop is worth zero; a 78 mph flare into shallow right field is worth a hit. Over a full season those break roughly even, but in a six-week window, they don't. xwOBA — expected weighted on-base average — strips out the variance.
The construction is straightforward. Statcast measures exit velocity and launch angle on every batted ball. Over years of data, those two inputs map to a probability of single, double, triple, home run, and out. xwOBA assigns each batted ball its league-average wOBA value based on its measured launch profile, ignoring what actually happened on the play. Walks, hit-by-pitches, and strikeouts are added back at their real values.
The output is a number on the wOBA scale — comparable directly to wOBA — that reflects the contact quality the hitter produced regardless of where the fielders happened to be standing. For a pitcher, xwOBA-against is the corresponding measure of how hard he's been hit.
Among public batted-ball metrics, the wOBA-minus-xwOBA gap is the most-cited regression flag. A hitter outperforming his xwOBA by 30 points is statistically very likely to cool off; one underperforming by 30 is, in expectation, due for a hot streak. The metric isn't infallible — speed is undervalued because xwOBA treats all 78 mph grounders the same regardless of who's running — but it's the closest we have to a luck-adjusted hitter rate stat.
Our hitter and pitcher priors regress observed wOBA halfway toward xwOBA when sample size is small. Models that trust raw wOBA in March April get whipsawed; trusting xwOBA gives a stickier estimate of true talent.