Gaffer
SUN · 2 AUG · 26 · VOL. 1

For recruitment teams

Time to turn the page on traditional football data.

A new category of football data

Traditional data stops at the individual. We go beyond that.

Existing data sources

Individual

Top sprint speedAerial duels wonShots /90xG /90Progressive passesPasses /90Tackles wonPass completion Ball recoveries xA /90

Interactional

Offensive impactDefensive attributionFit predictionCompatibilityOff-ball synergy
Gaffer data

What you have today

Your current data stack is dated.

  1. Historic

    It only tells you how a player performed, never how he will perform for you.

  2. Stripped of context

    You are forced to evaluate their performance on the assumption that the context will travel with them.

For example

  • A striker’s goals. Nothing about the service, or the system that fed him.
  • A midfielder’s passing drop. Nothing about the teammate who stopped making runs. The wrong player looks bad.
Advanced metrics too

They measure a player’s actions more precisely. But they still describe him alone, as if his output would travel unchanged to any other squad.

None of this is hidden

Anyone who watches football knows performance depends on teammates, role and system. We just forget it when we recruit.

What you get

Enter interactional data

Players and their actions, put in context. Interactional data measures what happens between players, even for two who have never played together.

  • Which players move together, and which do not.
  • Who covers the space someone else leaves.
  • Which pairs make each other better.

The same two players

  • The striker. Which of your creators feed the runs he actually makes.
  • The midfielder. Which partner keeps him on the ball, and which one leaves him stranded.
Predictive

It works for two players who have never shared a pitch, so it applies to the signing you are deciding on, not just the squad you already have.

Context aware

The answer is different for every club, because the fit a team needs depends on the current squad.

Build winning combinations before they enter the pitch

Measure compatibility

Score how any two players combine, even if they have never shared a pitch. You get their compatibility in play style as a number, with the reasoning behind it.

A. FONTAINE N. HALVORSEN S. OKARO D. PETRIC

Predict behaviour

Remove the uncertainty of whether a player will fit your playing style by predicting their behaviour in your current game plan.

PLAYER A PLAYER B YOUR #10 SIGNATURE RUN PLAYER A EXPECTED RUN PLAYER B EXPECTED RUN

Identify gaps

Reveal weaknesses hidden between players, not just within them. See where compatibility issues reduce team performance and where reinforcements will have the greatest impact.

GK LB CB CB RB xGA AS PLAYED WITH A COVERING LEFT CENTRE BACK

The research

Building on peer-reviewed science.

A pair’s value is what they produce together, beyond the sum of their individual play.

Modelled from how each player acts, so it can be predicted for two players who have never shared a pitch.

The principle

chemistry(A, B) =
value(A & B together)
[ value(A) + value(B) ]
value(A & B together)0.68
value(A) + value(B)0.21
chemistry +0.47 impact /90

Split into an offensive and a defensive component, validated across 361 seasons in 106 competitions.

Tomorrow’s best teams are combinations, not collections.

Bring interactional data into your recruitment and build for how players will play together.

Getting the data

The same data, two ways in.

Use our platform for direct access to our data,
or go custom using our API.

Platform

Purpose-built tools on top of the data.

You gain direct access to our platform upon signup. Your staff log in and the data is ready to use. Nothing to set up.

Fit Impact
N. Halvorsen +0.47
T. Vanhoof +0.34
M. Elzinga +0.23
S. Barreau +0.14
D. Okonjo +0.08
L. Ferreiro +0.02
R. Aaltonen +0.01
E. Marchand −0.02
K. Verlinden −0.04
J. Ostberg −0.07
P. Vasilek −0.09
A. Fontaine −0.12

Example view

API

The data, straight into your own workflow.

An API that is ready for advanced data teams to squeeze every last bit of knowledge out of our data. Custom apps, Jupyter notebooks and AI models, all ready to go.

GET /v1/compatibility?player=1042

{
  "player": 1042,
  "pair": ["N. Halvorsen", "T. Vanhoof"],
  "compatibility": 0.86,
  "confidence": 0.71,
  "drivers": [
    { "factor": "off-ball movement", "effect": 0.21 },
    { "factor": "pressing triggers", "effect": 0.09 },
    { "factor": "occupies same space", "effect": -0.04 },
    { "factor": "build-up tempo", "effect": 0.06 },
    { "factor": "width in possession", "effect": 0.03 }
  ],
  "squad": [
    { "with": "M. Elzinga", "compatibility": 0.74 },
    { "with": "S. Barreau", "compatibility": 0.69 },
    { "with": "D. Okonjo", "compatibility": 0.61 },
    { "with": "L. Ferreiro", "compatibility": 0.54 }
  ]
}

Example response

Built for the whole recruitment room.

Analysts

Give the club a measure of chemistry it has never had.

Scouts

Spend your hours on players who’ll actually fit, not the whole pool.

Heads of recruitment

Fewer mis-buys. Less scouting time lost on the wrong names.

Sporting directors

Put a number on a multi-million call.

The things clubs actually ask.

How can you measure chemistry between players who’ve never played together?

That is the whole point. Gaffer reads how each player plays: style, role, movement, the actions they take, then predicts how the two will combine. It doesn’t need shared minutes, so it works for the signing you’re deciding on, not just the pairs already in your squad.

What do I need to use Gaffer?

Your squad and your scouts. Link the players you already have and the reports your scouts write, and you’re set up. The interactional data comes from us, so there is nothing for your club to collect, build or maintain.

Is this proven, or a black box?

The underlying mechanism is grounded in peer-reviewed research. Gaffer’s own research team improved on this research and is the first to put it in a club’s hands. Every score ships with its confidence.

Which leagues and players are covered?

We cover all professional leagues and then some. We are always in the process of onboarding new leagues.

Field notes on data in football.

Read the Journal →

Measure fit before you sign.

We’re building this with our first clubs. If you’d rather build a team than collect players, let’s talk.