An Introduction to Football Analytics

Cameroon at the 2022 World Cup

Joris Bekkers

Contents

• Introduction
• Career Path


• Football Data & Analytics
• Basics
• Visualizations
• Analytics
• etc.


• Research & Innovation
• Pressing Intensity
• Graph Neural Networks

Introduction

Joris Bekkers 35 y/o Breda, Netherlands Football Analytics Consultant Research Development Deployment PySport Boardmember Unite Sports Analytics Community Grow and Maintain Open-Source

Introduction to Football Analytics

Open-Source Football Analytic Tools!

• kloppy
• floodlight
• socceraction / Footballdata / ETSY
• databallpy
• OpenSTARLab
• UFMG Sports Analytics Lab
• KAIST
• unravelsports
• mplsoccer
• Soccer Analytics Handbook
• penaltyblog
• Friends of Tracking

What is kloppy?

Load, Filter, Transform, Export Standardized Football Data.

Why kloppy?

Standardization!

Opta / StatsPerform

<Event event_id="17" type_id="1" period_id="1" min="0" sec="35" player_id="89076" team_id="497" outcome="1" x="41.4" y="16.4">
      <Q id="3925254967" qualifier_id="56" value="Right" />
      <Q id="3925254973" qualifier_id="212" value="23.1" />
      <Q id="3925254969" qualifier_id="140" value="61.2" />
      <Q id="3925366067" qualifier_id="178" />
      <Q id="3925254971" qualifier_id="141" value="1.7" />
      <Q id="3925254975" qualifier_id="213" value="5.8" />
    </Event>

StatsBomb

{"period" : 1, "minute" : 0, "second" : 3, "type" : {"id" : 30, "name" : "Pass"},
... "possession" : 2, "location" : [ 36.8, 27.3 ], "pass" : { "recipient" : { "id" : 6613}, "length" : 68.33, "angle" : 0.756}

etc.

Supported Providers

Provider	Event	Tracking	Public Data
Hawkeye (2D)
Metrica
PFF
SecondSpectrum	PR #437
Signality
SkillCorner
Sportec
StatsPerform / Opta
Tracab
DataFactory
StatsBomb
WyScout

Implemented Not yet implemented

Event Data

Using Open-Source Football Analytics Tools!

What is Event Data?

• Shots
• Passes
• Defensive Actions (Interceptions, Tackles, Clearances, Fouls etc.)
• Dribbles
• Substitutions
• Cards
  etc.

Load Event Data

from kloppy import opta

event_dataset = opta.load(
    f7_data='data/srml-4-2022-f2277835-matchresults.xml',
    f24_data='data/f24-4-2022-2277835-eventdetails.xml',
)
home_team, away_team = event_dataset.metadata.teams
print(f"{home_team.name} v {away_team.name}")

>>> Cameroon v Brazil

Meta data

Within the metadata we can find a lot of information, for example the player_id of each player.

for p in home_team.players:
    print(f"{p.jersey_no}. {p.full_name} - {p.player_id}")

>>> 20. Bryan Mbeumo - 446008
>>> 8. André-Frank Zambo Anguissa - 203325
>>> 13. Eric Choupo-Moting - 42564
>>> 10. Vincent Aboubakar - 83465

Meta data

event_dataset.metadata.pitch_dimensions

>>> MetricPitchDimensions(x_dim=Dimension(min=-52.5, max=52.5),
 y_dim=Dimension(min=-34.0, max=34.0), 
 standardized=False, unit=<Unit.METERS: 'm'>, 
 ... pitch_length=105.0, pitch_width=68.0)

Anguissa's Events

player = home_team.get_player_by_id("203325")

player_events = event_dataset.filter(
    lambda event: (
        event.player == player if event.player is not None else False
    )
)

>>> <EventDataset record_count=84>

mplsoccer!

from mplsoccer.pitch import Pitch
pitch = Pitch(pitch_color='grass', line_color='white', stripe=True)
fig, ax = pitch.draw()

mplsoccer Documentation

from mplsoccer import Pitch

def heatmap(xs, ys):
    pitch = Pitch(
        line_zorder=2,
        pitch_type="secondspectrum",
        pitch_length=transformed_events.metadata.pitch_dimensions.pitch_length,
        pitch_width=transformed_events.metadata.pitch_dimensions.pitch_width,
    )
    fig, ax = pitch.draw()
    
    ax.set_title(f"#{player.jersey_no} - {player.last_name} - {player.team.name}")
    pitch.kdeplot(xs, ys, ax=ax, cmap="YlOrRd", fill=True, levels=100)


transformed_events = player_events.transform(
    to_orientation="ACTION_EXECUTING_TEAM"
)

xs = [event.coordinates.x for event in transformed_events if event.coordinates is not None]
ys = [event.coordinates.y for event in transformed_events if event.coordinates is not None]

heatmap(xs, ys)

Shotmap

from kloppy.domain import EventType

team_shots = (
    event_dataset
    .filter(
        lambda event: (
            event.event_type == EventType.SHOT
        )
    ).transform(
        to_orientation="STATIC_HOME_AWAY"
    )
)

pitch = Pitch(
    line_zorder=2,
    pitch_type="secondspectrum",
    half=False,
    pitch_length=events.metadata.pitch_dimensions.pitch_length,
    pitch_width=events.metadata.pitch_dimensions.pitch_width,
)
fig, ax = pitch.draw()

ax.set_title(f"Brazil & Cameroon Shots")

for event in events:
    if event.coordinates is not None and event.result_coordinates is not None:
        color = "#479A50" if event.team == home_team else "#FFDC02"
        x = event.coordinates.x
        y = event.coordinates.y
        x_end = event.result_coordinates.x
        y_end = event.result_coordinates.y
        
        pitch.scatter(x, y, ax=ax, s=100, color=color, edgecolors="black")
        
        pitch.arrows(x, y, x_end, y_end, ax=ax, width=2, headwidth=5, 
                        headlength=5, color=color, alpha=0.6)

Momentum

Positional Tracking Data

PFF Open World Cup Tracking + Event Data

• 64 Full Matches of World Cup 2022 Tracking + Event Data

from kloppy import pff

match_id = "3840"

dataset = pff.load_tracking(
    raw_data=f"{match_id}.jsonl.bz2",
    meta_data=f"{match_id}.json",
    roster_meta_data=f"{match_id}.json",
    coordinates="secondspectrum",
)
home_team, away_team = dataset.metadata.teams
print(f"{home_team.name} v {away_team.name}")

>>> Cameroon v Serbia

len(dataset.records)

>>> 105709

dataset.records[0]

>>> <Frame>

pitch = Pitch(
    line_zorder=2,
    pitch_type="secondspectrum",
    half=False,
    pitch_length=dataset.metadata.pitch_dimensions.pitch_length,
    pitch_width=dataset.metadata.pitch_dimensions.pitch_width,
)
fig, ax = pitch.draw()

plt.title(f"Period {frame.period.id} - {frame.timestamp}")

for player, coordinates in frame.players_coordinates.items():
    if coordinates is None:
        continue
    color = "#479A50" if player.team.name == "Cameroon" else "#ffffff"
    pitch.scatter(coordinates.x, coordinates.y, ax=ax, s=100, color=color, edgecolors="black")
    
if frame.ball_coordinates is not None:
    pitch.scatter(frame.ball_coordinates.x, frame.ball_coordinates.y, ax=ax, s=10, color="white", edgecolors="black")

Aboubakar v Serbia '63

Advanced Football Analytics

Advanced Football Analytics

Pitch Control

Spearman (2018)
Fernández & Bornn (2017)

Advanced Football Analytics

Pressing Intensity

Pressing Intensity

Requirements

Use positional tracking data Multipurpose / Extendable No ML / prediction model Not too complex Not too simple Measurable at every moment Summable Measure pressing in the whole "system”

Pressing Intensity

Requirements

Use positional tracking data Multipurpose / Extendable No ML / prediction model Not too complex Not too simple Measurable at every moment Summable Measure pressing in the whole "system”

Total Pressure = 1 - [(1 – 86%) * (1 – 19%)] = 89%

What is the chance 🏃‍♂️^defense will reach 🏃‍♀️^attack or ⚽️ within
some amount of time given their current direction and speed?"

UnravelSports Blog

The unravelsports package aims to aid researchers, analysts and enthusiasts by providing intermediary steps in the complex process of converting raw sports data into meaningful information and actionable insights.

• Polars DataFrame Conversion
• Graph Neural Network Training, Conversion and Prediction
  [Bekkers & Sahasrabudhe (2023)]
• Pressing Intensity
  [Bekkers (2024)]

1. Load Kloppy & Convert

from kloppy import sportec

from unravel.soccer import KloppyPolarsDataset

kloppy_dataset = sportec.load_open_tracking_data()

kloppy_polars_dataset = KloppyPolarsDataset(
    kloppy_dataset=kloppy_dataset
)

KloppyPolarsDataset

2. Pressing Intensity!

from unravel.Football import PressingIntensity

import polars as pl

model = PressingIntensity(
    dataset=kloppy_polars_dataset
)
model.fit(
    start_time=pl.duration(minutes=1, seconds=53),
    end_time=pl.duration(minutes=2, seconds=32),
    period_id=1,
    method="teams",
    ball_method="max",
    orient="home_away",
    speed_threshold=2.0,
) 

>>> PressingIntensity(n_frames=143)

Probability to Intercept

[[0.07 0.   0.   0.   0.   0.03 0.01 0.   0.   0.   0.23]
 [0.28 0.   0.   0.   0.   0.02 0.   0.   0.   0.   0.  ]
 [0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.  ]
 [0.09 0.   0.   0.   0.   0.1  0.06 0.   0.   0.   0.05]
 [0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.  ]
 [0.   0.   0.   0.   0.   0.09 0.6  0.   0.   0.   0.  ]
 [0.   0.   0.   0.   0.   0.83 0.08 0.   0.   0.   0.  ]
 [0.02 0.   0.   0.   0.   0.02 0.   0.   0.   0.01 0.  ]
 [0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.  ]
 [0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.71]
 [0.3  0.   0.   0.   0.   0.06 0.   0.   0.   0.   0.03]]

3. Video

Full Example

Advanced Football Analytics

Graph Neural Networks

Graph Neural Network

Counter Attack Predictions

GNN Architecture

• Quick Start Guide
• From Kloppy to GNN with Polars
• BigDataBowl Example
• What are GNNs (FAQ)

from unravel.soccer import SoccerGraphConverterPolars

kloppy_polars_dataset.add_dummy_labels()
kloppy_polars_dataset.add_graph_ids(by=["frame_id"])

converter = SoccerGraphConverterPolars(dataset=kloppy_polars_dataset)

dataset = CustomSpektralDataset(graphs=converter.to_spektral_graphs())

from spektral.data import DisjointLoader

train, test, val = dataset.split_test_train_validation(
    split_train=4, split_test=1, split_validation=1, random_seed=43
)

from unravel.classifiers import CrystalGraphClassifier
model = CrystalGraphClassifier()

loader_tr = DisjointLoader(train, batch_size=batch_size)
loader_va = DisjointLoader(val, epochs=1, shuffle=False, batch_size=batch_size)

model.fit(
    loader_tr.load(),
    epochs=epochs,
    steps_per_epoch=loader_tr.steps_per_epoch,
    use_multiprocessing=True,
    validation_data=loader_va.load(),
    callbacks=[EarlyStopping(monitor="loss", patience=5, restore_best_weights=True)],
)

Thanks!