A contextual pass completion model predicts the probability that a given pass attempt will be completed, controlling for the difficulty of that specific attempt. Key features include: originating pitch zone, destination zone, pass distance, pass direction (forward/sideways/backward), whether the receiver was under pressure, and whether the passer was under pressure. The model's output — expected completion (xC) — is used to compute "completion above expectation," which is a more reliable signal of a player's passing skill than raw completion rate.
Correct model building: treat pass completion as a binary outcome (completed vs. not); use logistic regression or gradient boosted classifier; include zone-pair interactions (from_zone × to_zone), distance as a continuous feature, and under_pressure as one binary among many features (not the primary driver). The model should explain most variance via distance and zone, with pressure adding incremental signal. Player-level skill is estimated as the mean residual (actual − expected) over a sufficient sample (~500+ pass attempts).
When building a contextual pass completion model, the pressure coefficient as a main effect is tiny — less than 1% raw completion difference. The model is correct: pressure alone barely changes completion rate. But pressure INTERACTS with distance and direction: pressure on a long forward pass degrades completion far more than pressure on a short lateral pass. The main effect is nearly zero while the conditional effects are substantial.
Pass completion rate is inversely correlated with pass ambition. The very best passers attempt harder passes — longer, more progressive, under more pressure — which mechanically lowers their raw completion rate. A midfielder with 78% completion who is +6% above expected on every pass is objectively better than a midfielder with 91% completion who is +1% above expected but only attempts safe passes. The market rewards the 91% player because the number looks better.