Event data

Impact Lab

I wanted a player-impact page that starts with the ball, not the spreadsheet.

Most public football ratings ask a narrow question: who shot, who assisted, who ended up near the goal? That is useful, but it misses the long middle of a possession where players tilt the pitch before anyone gets credit. This lab is my first attempt at showing that middle layer in public.

View stats

The problem

Shots are the cleanest events to value, but football is not basketball. Most possessions never become shots. If a full-back pins a winger back, if a midfielder keeps receiving under pressure, if a centre-back keeps breaking the first line, the old box score barely notices.

The first lens

Territory Value is deliberately simple. It asks: did the action move the team forward, did it enter the final third, did it reach the box-zone, and was the player involved often enough to matter over real minutes?

The caveat

This is still a lab. The next version needs possession direction correction, event-type weighting, possession chains and pressure context. For now, the value is in making the assumptions visible instead of hiding them behind a fake final rating.

What goes into the score?

Territory Value per 90 currently combines four ingredients. The weights are intentionally readable so we can argue with them later.

Forward territory gain
How much positive x-axis ground a player’s actions move.
Final-third entries
Actions that carry the possession into the attacking third.
Box-zone entries
Actions ending in the central high-value zone around the penalty area.
Successful involvement
A small stabilizer for players who keep actions alive repeatedly.

Event explorer

View players through the lab lens

Loading event-data competitions.

ModelTerritory Value
Fixtures--
Sample-- players
Statusresearch

Average action map

Each marker is a top player’s average action start location. Larger markers mean more box-zone entries per 90.

Score leaders

Territory Value per 90, with the underlying split shown on hover.

    Evidence table

    Use this like a notebook: score first, then check the ingredients before trusting the name.

    RankPlayerTeamPositionMinutesTerritory ValueTerritory GainFinal-third entriesBox-zone entriesSuccess