Home/Soccer Analytics/Possession Value Decomposition (Action × Success × Outcome)

Possession Value Decomposition (Action × Success × Outcome)

Expected Value ModelsLevel 4 — Expert

What It Is

Instead of computing EPV as a single black-box number, the value of any possession action can be decomposed into three separable components: (1) probability of attempting that action type (pass/carry/shot); (2) probability of success if attempted; (3) expected value of the state after a successful attempt. This decomposition enables attributing value changes to specific causes — a player may choose the right action type but execute poorly (component 1 right, component 2 wrong), or may attempt a high-success action that goes to a low-value location (component 2 right, component 3 wrong).

Correct Execution

For any action: compute P(action type) from the distribution of historical action choices in that state; compute P(success | action type) from historical completion rates in that state; compute E[value | success] from the historical EPV of destinations for that action type. The product gives expected action value; comparing to what actually happened reveals whether value was added or lost in each component. This is the foundation of action-level attribution — who made good decisions, who executed well, and whose choices led to good outcomes.

Progression Levels

Diagnostic Tree

Coaching Cues

  • "Three things have to go right: choose the right action, execute it, and have it go somewhere valuable. Decompose which failed."
  • "Bad decision, good execution, good outcome — that's not a compliment. That's luck." — Javier Fernandez, 2019

Common Errors

  1. Using outcome quality as the only performance signal: Confounds all three components; a player who makes consistently poor decisions but gets lucky will look good in outcomes.
  2. Treating all three components as equally important: Decision quality (component 1) is most trainable; execution quality (component 2) is partly physical; outcome quality (component 3) has the most variance.

Edges

💎 Elite-Only Behavior

Outcome-Masked Bad Decisions Are Regression Time Bombs

A player making consistently poor decisions but getting lucky outcomes will be rated well by any outcome-based metric. When luck regresses, performance collapses "suddenly" — but decision quality was always poor. Decomposing value into decision, execution, and outcome quality catches this before regression.

What most people do
Evaluate on outcomes. Good xG contribution = good player. Decision quality only questioned after bad results.
What the best do
Decompose action value into decision, execution, and outcome quality separately. Flag players where outcomes >> decisions as regression candidates.
Why it's an edge: Identifying "luck-masked bad decisions" before regression gives 3-6 months lead time. Avoid buying in lucky phases; coach decisions proactively.
How to exploit: Build "decision quality gap" = outcome rank - decision rank. Large positive gaps = regression candidates. Large negative gaps = breakout candidates.
Javier Fernandez, FC Barcelona, 2019-10-22
🔑 Hidden Causal Lever

The Most Valuable Possession Actions Are Often Not the Final Pass or the Shot

Possession value decomposition reveals that the highest-EPV-delta action in a goal-scoring possession is often the 3rd or 4th action from the end, not the assist or the shot. A press-breaking carry that advances 30 yards often has a higher EPV delta than the final through ball, because it shifted the entire possession from low-value to high-value territory. Credit assignment that weights only the terminal actions (goal, assist, key pass) misses the player who actually created the scoring opportunity.

What most people do
Assign credit using goals, assists, and key passes — the terminal actions in the sequence.
What the best do
Decompose EPV across the entire possession chain. Identify the action with the highest EPV delta (the "value creation point") regardless of where it falls in the sequence. Credit the player who created the value, not just the player who finished it.
Why it's an edge: The transfer market prices goals and assists because they're visible. The player who consistently provides the 3rd-from-last action with the highest EPV delta is creating the goals but getting none of the credit. These players are systematically underpriced.
How to exploit: Compute per-player EPV delta rankings. Cross-reference with assists and key passes. Players who rank high on EPV delta but low on assists are the underpriced value creators — their contribution is in the setup, not the finish.
Javier Fernandez, FC Barcelona, 2019-10-22. Possession value decomposition showing highest-EPV actions often occur mid-sequence.

Sources

  • Javier Fernandez, FC Barcelona, StatsBomb Innovation in Football Conference 2019, YouTube, 2019-10-22 — presented the EPV decomposition formula (action probability × success probability × expected post-action value) as a way to attribute value changes to specific causal components