Pixel-Level Field Value Calculation

Rather than computing EPV only at the ball's location, pixel-level field value calculates the EPV of every location on the pitch simultaneously — creating a full-field value map at every moment. This shows not just where the ball is valuable, but where it would be valuable if played there — enabling identification of open spaces, optimal passing targets, and dangerous unoccupied areas. The output is a heatmap where red indicates high-value locations and blue indicates low-value locations, recomputed every fraction of a second.

Pixel-level computation: for each pitch location, estimate EPV under the assumption that a possession is currently there, conditioned on the current defensive and offensive player positions. Key technical requirement: tracking data at high frequency (25Hz or better). The computation is expensive — typically pre-computed for specific possession frames rather than continuously. The resulting maps are most useful for: (1) showing optimal passing targets; (2) identifying off-ball spaces created by player movement; (3) coaching visual aid showing where value exists on the field.

1. Computing field value without conditioning on opponent positions: A high-value location changes as defenders shift; the value map must be dynamic, not static.
2. Using event data instead of tracking data: Pixel-level computation requires continuous position data; event data only gives locations at discrete action moments.