PLAYER-IMPACTMethodology

What is Player Impact (PI)? Measuring a Player's Effect on Team Results

Author: JPick Editorial Team📅 Published: 2026-04-25

What is Player Impact (PI)? Measuring a Player's Effect on Team Results

TL;DR

Player Impact (PI) measures how much a team's results change when a specific player starts versus when they do not. It is not a measure of individual skill — it is an in-team influence score on a −100 to +100 scale.

Basic Definition

Player Impact (PI) is JPick's proprietary in-team influence metric. The score ranges from −100 to +100. The further above zero, the better the team has performed in matches the player started.

PI does not measure how good a player is. It only measures how much the team's outcome shifts based on whether that player is on the field. The two are not the same. A player with a high PI is not necessarily one of the league's best individual performers, and vice versa. Tactical roles and chemistry with teammates can push a quiet player's PI higher than the box score would suggest.

The same idea exists in other sports. Basketball's plus-minus measures how a team's score margin moves while a given player is on the court. Basketball lends itself to that calculation because of the volume of scoring events and the constant flow of possessions. Football is the opposite — fewer goals, longer phases, more tactical variance — so JPick's PI works at match level (the player's start versus non-start) rather than at the possession level.

How It Is Calculated (the Concept)

PI's core question is: how do team results differ between matches where the player started (On games) and matches where they did not (Off games)? The football-specific application of this idea is sometimes called WOWY (With Or Without You), as introduced by American Soccer Analysis.

JPick uses two main inputs in this comparison:

  • PPG (points per game) difference — the gap between the team's average league points in On versus Off matches. This is the primary signal.
  • xGD (xG difference) gap — the gap in expected goal difference between On and Off matches. This serves as a supporting signal on the underlying performance.

PPG captures actual results, but football is low-scoring and highly susceptible to single-match variance. A team can dominate a match and still draw or lose because of one finishing error or a deflected shot. xGD smooths over those swings by reflecting the underlying chance creation and concession, providing a truer reading of how the player's presence shifts performance — not just outcomes.

Weights for these inputs were derived by backtesting historical data to determine which combination best predicts future match outcomes. The PPG signal carries the bulk of the score, with xGD acting as a corrective on top of it.

To avoid the small-sample trap, PI pulls scores back toward zero when On or Off counts are low. Deriving a PI of +80 from a one- or two-match sample risks mistaking statistical variance for a genuine signal. This follows the same logic as Tom Tango's Marcel projection system in baseball, which regresses player performance toward the mean — a standard technique to dampen statistical noise from small samples. The more games a player has, the closer the score reflects the raw On/Off gap; the fewer games, the closer it sits to zero.

The final value is clipped to a ±100 scale. Players who start every match in the season have no Off games to compare; for them, JPick falls back to a separate estimate based on their average match rating.

PI also carries a confidence flag — High, Medium, or Low — based on how many games sit on each side of the comparison. Low-confidence scores should be read with caution.

Data Source

JPick computes PI from match results, lineup data, and xG figures pulled via API-Football. The calculation spans the current and previous seasons, refreshing via a daily batch process.

No additional API calls are required — the work is purely a database aggregation. New scores appear the day after a fixture is recorded as Final Time (FT).

How to Read It

Specific numbers help ground the metric. The following cases illustrate how PI should be interpreted.

Case A: PI = +40, High confidence The team has earned significantly more points when this player starts and stalled without them. The team relies heavily on this player, or the manager has built the squad's tactical identity around them.

Case B: PI = ±5, High confidence Team results barely change with or without this player. They are interchangeable rotation pieces, or their contribution shows up in metrics other than win-loss outcomes.

Case C: PI = −20, Low confidence The team has done better in Off matches. With a small sample, this is short-term noise rather than a real signal. The number will likely move significantly once more matches accumulate later in the season.

Case D: PI = +30, fallback value A player who has started every match in the season. With no Off games to compare, JPick estimates the score from the player's average match rating. This value is not directly comparable to other PI scores.

Time horizon matters. Scores from a handful of early-season matches can be moved by the luck of which opponents fell on which side of the split. Past 10 matches, On and Off samples both grow, and the score becomes more reliable.

Limitations

PI is a useful in-team influence metric, but several factors sit outside what the data can show.

It is not an individual skill rating. A high PI does not certify that a player is among the league's best, and the opposite is also true. Defensive midfielders and players in supporting tactical roles can post high PI without registering goals or assists. Conflating PI with "skill" is a misreading.

Co-occurrence effects are not adjusted. If a player happens to miss matches against top-tier opponents or enjoys a streak of home fixtures during their "On" stretches, PI cannot correct for that schedule luck. Basketball's Adjusted Plus-Minus (APM) decomposes the influence of teammates and opponents statistically; JPick's PI does not go that far.

Tactical context is not encoded. Contextual variables such as opponent strength, venue (home/away), tactical formation, and substitution timing are not encoded into the PI calculation.

Injury and managerial choice are conflated. PI does not distinguish between an Off match caused by a long-term injury and one caused by a coaching rotation. A long-term absentee and a player rested for a single matchday are both treated as "Off."

Low-confidence scores are advisory only. Low confidence appears often early in the season or for players with limited starting time. Conclusions should wait until the underlying samples grow.

Related Metrics

PI sits alongside several other metrics. The differences are worth keeping in mind.

Match rating evaluates an individual's performance within a single match. While PI measures the team-level outcome gap, match rating isolates individual technical contribution. Reading them together can surface players who stand out individually but fail to move the team-level needle.

xG (expected goals) captures shot quality and feeds into PI as one of the inputs. xG on its own describes a team's attacking efficiency; PI translates that lens into a per-player view of in-team influence.

JPick Edge is a separate proprietary metric that estimates a player's potential to break out this season. PI looks backward at past influence; Edge looks forward at upside.

How JPick Uses It

PI surfaces in several places across the JPick app.

Lineup-announcement insights. When the day's starting eleven is published, JPick checks whether any core players have been left out. If so, the influence is estimated from PI, and an alert insight is generated automatically.

Lineup Strength. The mean PI of the eleven starters becomes a team-level Lineup Strength score, which is shown alongside the rolling five-match average and the change since the previous matchday.

Player detail pages. Every player's page shows their PI as a score bar so readers can see in-team influence at a glance.

PI is not meant to be read as a single number in isolation. Combined with confidence, sample size, and recent match context, it surfaces parts of a team's reality that the league table alone misses. That is the JPick Lab analytical stance.

References

← Methodology Back to listExplore more data →