WPA measures the change in a team's win probability attributable to each player action across a season. A player who hits a game-winning three adds high WPA; a player who misses a late free throw subtracts WPA. The sum over a season estimates a player's net contribution to team wins through high-leverage moments.
Explicitly values performance when games are on the line. Captures clutch performance in a measurable way. Zheng & Cao (2011) find significant individual variation in pressure performance, validating the concept that some players are genuinely better than others in high-leverage moments.
Very sensitive to small samples of late-game situations. Regular-season WPA is noisy because not all players get opportunities in late, close games. Highly dependent on opportunity — a team's designated closer will accumulate WPA (positive or negative) at far higher rates than a defensive specialist who rarely touches the ball in late situations.
Not used in APEX v1.7. WPA is identified as a future Clutch Module candidate in APEX's V2 Roadmap, consistent with Zheng & Cao (2011)'s finding of individual clutch variation. However, implementing WPA requires careful sample size controls and opportunity normalization before it provides reliable signal at the season level. Including it in the current model without those controls would add noise rather than signal.