PresentationPDF Available

Quantifying Pitch Control

Authors:
William Spearman at Liverpool Football Club
William Spearman
  • Liverpool Football Club
Download file PDF
Read file
Download citation

Abstract

This presentation introduces a new way to quantify the control that various players exert on different regions of the pitch using spatio-temporal player tracking data from association football.
Content uploaded by William Spearman
Author content
All content in this area was uploaded by William Spearman on Aug 01, 2019
Content may be subject to copyright.
•Specialists in video capture and online streaming. 100% video coverage of over 80 leagues world wide
•Opta data widgets available within Scout 7 player records and fully integrated with Xeatre video.
•Boston based sports and data analysts. Providers of recruitment analysis software to NBA and NFL teams
•Leader in real time player tracking technology
•League wide provision of match analysis software to all 42 Spanish league teams
•Data integration with Red Bee’s Piero software as by broadcasters world wide
•Data integration with SportsCode, leading provider of self coding analysis software
•Widgets integrated to The Sports Office club management system
OptaPro Partnerships
Quantifying Pitch Control
William Spearman
#OptaProForum
Hudl + Replay Analysis + Sportstec
Introduction
Dr. William Spearman
•Ph.D. in High Energy Particle Physics from
Harvard University
•Studied the Higgs Boson
•Works as a data scientist for Hudl
•Not a football expert!
Pitch Control: The Final Product
Referee
Away Player
(with velocity vector)
Home Player
(with velocity vector)
Ball Location
Pitch Control
Field Scale
(β = 2.5)
1= home team control
0 = away team control
Region Controlled
by Away Team
Region Controlled
by Home Team
Neutral Region
Motivation
Physics
The electric potential field quantifies the way
charged particles exert a force on a test charge
in space.
Football
We propose the development of a pitch
control field that can be used to quantify the
way football players control regions on the
pitch.
Definition
We define pitch control field (PCF) for location, x, as the probability that the home team
will end up with possession of the ball if it were at location, x. The PCF predicts the “next
possessor”.
What is Pitch Control?
Get there first with the most men
The PCF will probably depend on the time it takes each player to reach location, x and
this time will depend on the location and velocity of each player.
Using Tracking Data
TRACAB Data
•Player/ball positions at 25 fps
•We smooth it with an S-G filter
Calculating Times Using
•Player position
•Player velocity
•Player acceleration
•Maximum player speed
Choosing a Model
Time for the ith player to
reach the ball
Label for the ith player (1 for Home team
and -1 for away team)
This part will be between -1 and 1
β is a parameter which indicates how much to weight
being the first to the ball (Range = 0 to ∞)
Use a naïve linear scaling to change
output from -1 to 1 è0 to 1.
Understanding the Parameter, β
β= 0: PCF always 0.5
β= ∞: PCF is 1 if closest player is on the home
team, otherwise, it’s 0.
PCF(ti,l
i)="Pilit
i
Pit
i
+1
#/2
Fitting Strategy
1. Sync Opta and Tracab
•Identify when the ball is “in-play”
2. Calculate ball’s possessor
•Contested –both teams are near it
•None –no team is near the ball
•Home/Away –one team has uncontested control of
the ball
3. Calculate ball’s next possessor
•This is done by looking forward in time to see
which team gets the next uncontested possession.
Contested Home Team None Away Team
Current Possessor
1 0
Next Possessor (Home = 1 and Away = 0)
4. Choose Fit Frames
•We focus on frames where there possession is
contested or none.
•We want there to be a clear next possessor.
•The next possessor is truth value for who gets the
ball next
5. Calculate PCF at ball’s location.
•This gives us the model’s prediction for who gets
the ball next.
6. Minimize sum of squared errors
•Calculate residual for each frame:
•r = xpcf-xreal
•Sum of the squared residuals for each frame, i:
•Σr2= Σi(xpcf,i-xreal,i).
Open player
Passing player
Applications: A New Way To Watch Film
Watch Alongside Film
•Watch animated PCF along with video.
•PCF makes it possible to visualize space in-between units and
space behind units.
Defender out of
position
Identify
•Defensive players who are out of position (even if the mistake
doesn’t result in a goal)
•Missed offensive opportunities.
SAMPLE
Open player
Passing player
Applications: A New Way To Watch Film
SAMPLE
Evaluate Passing Decisions
•Passing decisions can be evaluated
by looking at how much open space
is in front of attackers or around
backs.
•Integrate the PCF to determine
open space in passing region.
•Best targets: attackers downfield
with large integrated PCF values.
0.8
0.2
0.3
0.9
0.7
Applications: A New Metric for Player Performance
Identify Controlling Players
•Certain players are more capable of contesting for the
football than others.
•This should appear if we calculate their “player-specific”
beta values.
•In other words, how much does their presence improve
their team’s chance of gaining possession of the ball above
the average.
How?
•This is done by fitting βiseparately for each player, i.
•Need more statistics to make inferences about specific
player trends.
•βonly matters w.r.t. other players. No common sense
interpretation.
PCF(ti,l
i)="Piliti
i
Piti
i
+1
#/2
Results
•Above we see the fitted beta value for specific midfielders
when compared to the league average.
•Some are more controlling while others exert less impact.
•Error bars show the standard deviation among four games.
Player Pitch Control
Applications: Player Positioning
Quantify the Effect of Player Positioning
•How much impact does a player’s position have on his team’s control
of the pitch?
•In other words, how different would the PCF be if a certain player
weren’t on the pitch?
PCF(j, ti,l
i)="Piliti
i
Piti
i
Pi6=jliti
i
Pi6=jti
i#/2
•Specialists in video capture and online streaming. 100% video coverage of over 80 leagues world wide
•Opta data widgets available within Scout 7 player records and fully integrated with Xeatre video.
•Boston based sports and data analysts. Providers of recruitment analysis software to NBA and NFL teams
•Leader in real time player tracking technology
•League wide provision of match analysis software to all 42 Spanish league teams
•Data integration with Red Bee’s Piero software as by broadcasters world wide
•Data integration with SportsCode, leading provider of self coding analysis software
•Widgets integrated to The Sports Office club management system
OptaPro PartnershipsThank you
William Spearman
#OptaProForum
Special Thanks:
Armen Badeer
Austin Basye
Nathan DeMaria
Keenan Hawekotte
Sam Lloyd
John McGuigan
Paul Pop
Markus Woodson
... There are also the g+ [15] and VAEP [8] models, and the work of [9]. With the arrival of tracking (spatial-temporal) data and applications of physics principles, Spearman [5] introduced "pitch control," a model quantifying the probability that a given team will have possession of the ball if it were at a location x. [7] provides a version adapted to individual player attributes. More recent papers have looked at team motion, including that [16] studies the motion of a team as a whole, with a focus on team synchronization. ...
... As a first step toward defining "safe configurations" (Definition 2.3), we build on Spearman's notion of pitch control ( [6,5]) to define "passable regions" where a particular team is most likely to control the ball upon arrival (please see §2.1). For a configuration of players on an attacking team to be considered "safe," we require that a suitable combination of passing lanes (please see §2.3) is within the attacking team's passable region. ...
... The region player B controls is only useful for aerial passes. This kind of situation is taken into consideration by Spearman et al in [6], but not included in the application of the Pitch Control Function [6,5]. See Definition 2.1 for the full definition of a team's passable region. ...
View
... With the arrival of tracking (spatial-temporal) data and applications of physics principles, William Spearman [5] introduced "pitch control," a model quantifying the probability that a given team will have possession of the ball if it were at a location x. ...
... As a first step toward defining what we will call "safe configurations," (Definition 3.3) we build on Spearman's notion of pitch control, as defined in [6,5], to define "passable regions" where a particular team is most likely to control the ball upon arrival (please see §3.1). For a configuration of players on an attacking team to be considered "safe," we will require that a suitable combination of passing lanes (please see §3.2) is within the attacking team's passable region. ...
... To illustrate the significance, one can imagine a situation where it looks like there is a clear passing lane from player A to player B and player B controls a region for "receiving" the ball, but the ball is intercepted because an opposition player could reach that location of the passing lane before the ball could, so the region player B controls is only useful for aerial passes. We note that this kind of situation is taken into consideration by Spearman et al in [6], but was not included in the application of the Pitch Control Function [6,5]. See Definition 3.1 for the full definition of a team's passable region. ...
Controlling ball progression in soccer
Preprint
Full-text available
  • Oct 2022
In this paper, we examine how soccer players can use their spatial relationships to control parts of the field and safely move play up the field via chains of ``safe configurations,'' i.e. configurations of players on a team ensuring the possessor of the ball has a collection of open passing options all connected by open passing lanes. An underlying philosophy behind our work is that it is most difficult to disrupt an attacking team's progression forward (with the ball) when this attacking team has multiple ``good'' options of how to proceed at each moment in time. We provide some evidence of this. Our main construction is a directed weighted graph where the nodes encode the configurations of players, the directed edges encode transformations between these configurations, and the weights encode the relative frequencies of the transformations. We conclude with a few applications and proposed further investigations. We believe that our work can serve as a launching platform for significant further investigation into how teams can ``safely progress'' the ball up the field, strategy development, and sophisticated decision making metrics. For coaches and players, we aim to streamline the process of moving safely up the field. In particular, we aim to construct a framework for creating new successful patterns of movement and aim that our framework allows for the development of new strategy. At the same time, our work could help identify configurations on the field which often result in turnovers or that allow for a lot of strategic flexibility (also impeding defensive containment of the attacking team). *Because the contributions of the female authors to this paper were by no means less than those of the male authors, we have chosen to reverse convention and to list the authors in reverse alphabetical order to ensure that the female authors were not listed only after the male authors.
View
... A key input to the model in Fernandez et al 2021 [18] was the output of a pitch control model. Pitch control models are tracking data analyses that use spatial control as the target variable instead of goal probability [19][20][21]. Similar to the approach of Decroos et al 2019 [17], but extended to tracking data, Spencer et al 2019 [6] calculated a feature set on each individual frame of AF tracking data and made expected point predictions on each frame. ...
... Team pitch control. The amount of surface area "controlled" by each team taking into account player locations and movement vectors [19]. In this case this feature is analogous to the total density of each team. ...
Assigning goal-probability value to high intensity runs in football
Article
Full-text available
High intensity run counts—defined as the number of runs where a player reaches and maintains a speed above a certain threshold—are a popular football running statistic in sport science research. While the high intensity run number gives an insight into the volume or intensity of a player’s work rate it does not give any indication about the effectiveness of their runs or whether or not they provided value to the team. To provide the missing context of value this research borrows the concept of value models from sports analytics which assign continuous values to each frame of optical tracking data. In this research the value model takes the form of goal-probability for the in-possession team. By aligning the value model with high intensity runs this research identifies positive correlations between speed and acceleration with high value runs, as well as a negative correlation between tortuosity (a measure of path curvature) and high value runs. There is also a correlation between the number of players making high intensity runs concurrently and the value generated by the team, suggesting a form of movement coordination. Finally positional differences are explored demonstrating that attacking players make more in-possession high intensity runs when goal probability is high, whereas defensive players make more out-of-possession high intensity runs while goal probability is high. By assigning value to high-intensity runs practitioners are able to add new layers of context to traditional sport science metrics and answer more nuanced questions.
View
... William Spearman has developed pitch control models for the Liverpool Football Club where ideas are sketched out in a conference presentation (Spearman 2016) and also in the YouTube video https://www.youtube.com/watch?v=X9PrwPyolyU. The approach is also probabilistic and is based on the estimated time t i that it takes the ith player to reach a given location. ...
... Our approach also emphasizes how long it takes a player to reach a certain location. In a Friends of Tracking video (Shaw 2020), Laurie Shaw provides details on implementing a pitch control model based on some of the ideas developed by Spearman (2016). A slightly different pitch control idea based on a minimal arrival time was developed by Narizuka et al. (2021). ...
A new metric for pitch control based on an intuitive motion model
Article
Full-text available
  • Jun 2024
  • COMPUTATION STAT
With the availability of tracking data, the determination of pitch control (field ownership) is an increasingly important topic in sports analytics. This paper reviews various approaches for the determination of pitch control and introduces a new field ownership metric that takes into account associated sporting dynamics. The methods that are proposed utilize the movement of the ball and players. Specifically, physical characteristics such as current velocity, acceleration and maximum velocity are considered. The determination of pitch control is based on the time that it takes the ball and the players to reach a given location. The main result of our investigation concerns the validation of the resultant pitch control diagram. Based on a sample of 5887 passes, the team identified as having pitch control was the observed recipient of the pass with 91% accuracy. The approach is generally applicable to invasion sports and is illustrated in the context of soccer. Various parameters are introduced that allow a user to modify the methods to alternative sports and to introduce player-specific maximum velocities and player-specific accelerations.
View
... Since classifying an image that only has some blue, red, and black points is a difficult task for a CNN model, I decided to enrich each image with the space each team controls known as "pitch control" models (Spearman, 2016). A simple Voronoi diagram was used to show the controlled space by the blue team, the red team, and the passer. ...
Potential Penetrative Pass (P3)
Preprint
Full-text available
  • Jan 2023
To score goals in football, a team needs to move forward on the pitch and there are various ways to do so. Depending on the game plan & philosophy; some teams prefer to play long balls from either wings or defense. Others, prefer to penetrate in depth with passes and outplay the opponent players. To objectively & in an automated way evaluate how teams play penetrative passes compared to the number of times they had the potential to do so, the "Potential Penetrative Pass (P3)" concept is presented here.
View
... Moreover, one may seek to first define an effective 'action space' for players (i.e., more granular or structured actions than 'move left' or 'move right'), before solving the RL problem. Finally, the combination of the previous learnt models with pitch control (Spearman, 2016), a technique to determine which player/team has control over a specific area of the pitch, will provide additional information on open space, passing and scoring opportunities, yielding a powerful tool to enable in-match tailored coaching tools. ...
Game Plan: What AI can do for Football, and What Football can do for AI
Article
  • May 2021
  • JAIR
The rapid progress in artificial intelligence (AI) and machine learning has opened unprecedented analytics possibilities in various team and individual sports, including baseball, basketball, and tennis. More recently, AI techniques have been applied to football, due to a huge increase in data collection by professional teams, increased computational power, and advances in machine learning, with the goal of better addressing new scientific challenges involved in the analysis of both individual players’ and coordinated teams’ behaviors. The research challenges associated with predictive and prescriptive football analytics require new developments and progress at the intersection of statistical learning, game theory, and computer vision. In this paper, we provide an overarching perspective highlighting how the combination of these fields, in particular, forms a unique microcosm for AI research, while offering mutual benefits for professional teams, spectators, and broadcasters in the years to come. We illustrate that this duality makes football analytics a game changer of tremendous value, in terms of not only changing the game of football itself, but also in terms of what this domain can mean for the field of AI. We review the state-of-the-art and exemplify the types of analysis enabled by combining the aforementioned fields, including illustrative examples of counterfactual analysis using predictive models, and the combination of game-theoretic analysis of penalty kicks with statistical learning of player attributes. We conclude by highlighting envisioned downstream impacts, including possibilities for extensions to other sports (real and virtual).
View
... the combination of the previous learnt models with pitch control (Spearman, 2016), a technique to determine which player/team has control over a specific area of the pitch, will provide additional information on open space, passing and scoring opportunities, yielding a powerful tool to enable in-match tailored coaching tools. ...
Game Plan: What AI can do for Football, and What Football can do for AI
Preprint
  • Nov 2020
The rapid progress in artificial intelligence (AI) and machine learning has opened unprecedented analytics possibilities in various team and individual sports, including baseball, basketball, and tennis. More recently, AI techniques have been applied to football, due to a huge increase in data collection by professional teams, increased computational power, and advances in machine learning, with the goal of better addressing new scientific challenges involved in the analysis of both individual players' and coordinated teams' behaviors. The research challenges associated with predictive and prescriptive football analytics require new developments and progress at the intersection of statistical learning, game theory, and computer vision. In this paper, we provide an overarching perspective highlighting how the combination of these fields, in particular, forms a unique microcosm for AI research, while offering mutual benefits for professional teams, spectators, and broadcasters in the years to come. We illustrate that this duality makes football analytics a game changer of tremendous value, in terms of not only changing the game of football itself, but also in terms of what this domain can mean for the field of AI. We review the state-of-the-art and exemplify the types of analysis enabled by combining the aforementioned fields, including illustrative examples of counterfactual analysis using predictive models, and the combination of game-theoretic analysis of penalty kicks with statistical learning of player attributes. We conclude by highlighting envisioned downstream impacts, including possibilities for extensions to other sports (real and virtual).
View
View
Pass Evaluation in Women's Olympic Hockey
Preprint
Full-text available
  • May 2022
Passing during power plays in hockey is a crucial component to move one's team closer to scoring a goal. With the use of women's ice hockey event and tracking data from the elimination round games during the 2022 Winter Olympics, we evaluate passing and assess players' risk-reward behaviours in these high intensity moments. We develop a model for probabilistic passing that accounts for order of arrival to a desired location and potential interceptions along the way. This model is based on a player-specific motion model and a puck motion model that determines how far each player can reach in the time it takes the puck to reach a target. In addition, we model the rink control for each team and the scoring probability of the offensive team. These models are then combined into new metrics for quantifying where a pass should be made such that it would result in a high scoring opportunity or result in a high chance of maintaining possession. Finally, we use various metrics the evaluate passes made throughout the available power plays and compare them to the optimal options at that time. This can be used to identify the risk-reward tendencies of various players and can be used by coaches when selecting which players are best suited for a power play given the circumstances of the game.
View
ResearchGate has not been able to resolve any references for this publication.