Estimating the probability distribution of pass destinations across the full pitch for a given game state, identifying the "typical pass" as a baseline against which to measure unexpectedness. The model predicts where most players would pass from this exact situation β deviation from that baseline = originality.
Train a model to predict pass destination probability surface from game state features (passer location, teammate/opponent positions, pressure, game dynamics). For each passing moment, the surface shows where "most players" would pass. The actual pass destination's probability in this surface = how typical it was. Low probability = original.
By computing how surprising each pass is relative to the expected pass distribution at that moment, you get a pass originality score. Players who consistently make high-originality passes that succeed β passes that the model wouldn't predict but that work β are demonstrating creative vision that no standard metric captures. This is distinct from pass completion, progressive passes, or xA, all of which can be generated by volume.