Soccer Analytics, or how to see beyond 300 Million eyes

Soccer analyticsOr how to see beyond 350 million eyes

Methodology proposal & Expected results

January 2014

Dir. Jaume Sués Caula

Managing Director

[email protected]

Avda. Doctor Pouplana 18-20 | 08950 Esplugues del Llobregat (Barcelona) | Fix. +34 934 732 482 | Mob. +34 610 525 034

Advanced statistical analysis applied to

Top sports, to support decision making process to

their sport directors1

What Sports

analytics is

We propose 4 areas where soccer

analytics will be specially useful, by analyzing

the data and obtaining “so what” conclusions3

Collaboration

opportunities

We show an example based on a Premier

League game, and identify some questions that

simple game watching is not able to answer2

How to be

applied to soccer

Content of this document

What Sports analytics is

Sport analytics started in 2002 with baseball in US, which nowadays is a sports

discipline fully extended in elite clubs

A true story of success

� Oakland Athletics general manager Billy Beane (Brad

Pitt) met Peter Brand (Jonah Hill), a young Yale

economics graduate with radical ideas about how to

assess baseball players' value, and together start using

data analytics to plan their season and each play of

each game

� For 2002 Season, Beane and Brand selected players

based almost exclusively on their on base percentage

(OBP). By finding players with a high OBP but with

characteristics that lead scouts to dismiss them, Brand

assembles a team of undervalued players with far more

potential than the A's hamstrung finances would

otherwise allow

� The Athletics go on to win 19 consecutive games, tying

for the longest winning streak in American League

history.

� Thereafter, Beane passes up the opportunity to become

the general manager of the Boston Red Sox, despite an

offer of a $12.5 million salary, which would have made

him the highest-paid general manager in sports history

� He returns to Oakland to continue managing the

Athletics. In 2004, two years after adopting the

sabermetric model, the Boston Red Sox win their first

World Series since 1918.2

What Sports analytics is

It’s only after consistently analyzing player’s detailed data that you obtain

useful information to understand what’s happening on the field

3

Lebron James spatial “Points per attempt” evolution

2011 NBA Finals 2012 NBA Finals

The maps summarizes the density of all field goal attempts during the study period: Larger squares indicate areas where many field goals

were attempted; smaller squares indicate fewer attempts.

The color of the squares is determined by a spectral color scheme and indicates the average points per attempt for each location.

Source: “Court Vision (MIT Sloan Sports Analytics Conference 2012)

“The sexiest job in the next ten

years will be statisticians… The

ability to take data—to be able

to understand it, to process it,

to extract value from it, to

visualize it, to communicate it

that’s going to be a hugely

important skill.”

Hal Varian

Chief Economist at Google

Source: How the Web Challenges Managers (McKinsey Quarterly)4

Sports

analytics is

already is

key in NBA

decision

making

process

5

6

And what about soccer?




What Sports

analytics is




Collaboration

opportunities




How to be

applied to soccer


Quick data analytics on 2013

Southampton 3 – Manchester City 2

(First half)

How to be applied to soccer

During first half, only 8 of the 88 plays in attack of Manchester United ended in

a shot (including one penalty) or a goal

Outcome of Manchester City first half plays

Goal

1%Shot

8%

Foul

8%

Loss

(passing)

76%

Loss

(conducting)

7% � 83% of plays in attack of Manchester City

ended with possession loss

— Where is concentrated the possession loss?

— In between which players?

— In which field areas?

— When defended by who?

� 9% of the plays end in a goal or shot

— How does this plays better start?



� And what about the opponent’s attack plays?

— Why opponent’s plays end in goal or shot?

— Why does not?

9


During this period, only 5% of the 2.000 plays occurred really added value to

the game, while 11% might have risked Manchester City goal

Quality of Manchester City first time plays

0%

1%

4%

15%

65%

2%

3%

11%

4/4 (Excelent)

3/4 (Very Good)

2/4 (Good)

1/4 (Quite Good)

0 (Neutral)

-1/4 (Quite Bad)

2/4 (Bad)

3/4 (Very bad)� 11% of plays of Manchester City where “Very

badly” qualified

— When created a risk in Manchester City goal?


— Done by who?


� 5% of plays of Manchester City where

“Good”, “Very Good” or “Excelently”

qualified

— When increased Manchester City probability to

score?


— Done by who?


10


The 2.000 plays of a single game must be linked to understand how the goal

opportunities are created or destroyed

Linked plays for the first Manchester City goal

Jack Rodwell

Yaya Toure

Samir Nasri

Carlos Tévez

PassPlayer pass accuracy?Player quality of plays in that game?� On which area field� During which match moment� When defended by who� When linked with Toure Yaya

PassPlayer pass accuracy?Player quality of plays in that game?� On which area field� During which match moment� When defended by who� When linked with Samir Nasri

PassPlayer pass accuracy?Player quality of plays in that game?� On which area field� During which match moment� When defended by who� When linked with Carlos Tévez

GoalPlayer shot accuracy?Player quality of plays in that game?� On which area field� During which match moment� When defended by who� When linked with Samir Nasri

11


Then, interaction between players and how the flow of the plays is generated

arise …

Manchester City players interaction during first half

12

When closer, players have had more interactions during games. Bubble size proportional to the number of plays per player

Source: 2-Factorial analysis of Manchester City plays during first time

Yaya Toure

Jack Rodwell

Samir Nasri

David Silva

Pablo Zabaleta

Vincent Kompany

Gael Clichy

Joleon Lescott

Carlos Tevez

Edin Dzeko

Joe Hart

Which players interactions most value created to M City?




… and thus, the real quality and influence of each player can be stated, for his

solely actions but also in the concatenated plays created

Manchester City players evaluation

13

Player Plays Quality of the play Quality of the next linked

plays

Influence of the play Influence of the next

linked plays

Yaya Toure 74 0.07 0.28 1.34 6.74

Jack Rodwell 56 0.02 0.18 1.38 7.17

Samir Nasri 52 0.27 0.53 2.06 8.19

David Silva 51 0.08 -0.01 2.22 6.88

Pablo Zabaleta 49 -0.37 -0.50 0.55 3.58

Vincent Kompany 43 0.02 0.14 0.64 5.01

Gael Clichy 40 0.05 0.17 1.56 6.52

Carlos Tevez 34 0.03 -0.25 2.91 6.65

Joleon Lescott 32 -0.03 0.12 0.56 5.26

Edin Dzeko 19 -0.11 0.00 2.03 8.71

Joe Hart 8 0.25 0.30 0.88 4.36

Were Samir Nasri the one that most influenced the game?


— Done by who?


So what this can be useful for?




What Sports

analytics is




Collaboration

opportunities




How to be

applied to soccer


Collaboration opportunities

We propose 4 successful collaboration opportunities to help improve a Soccer

team results

Collaboration Opportunities

16

� Identify best line ups and substitutions for a

given team opponent, moment of the season

and / or competition

Match

preparation

� Identify a player progression along the season

by field position, defenders and competitionPlayer

follow up

� Identify the best trade off between players

follow up and incoming opponents for mid term

gains maximization

Season

planning

� Set cost-effective multiple statistical and

consistent follow ups through a huge number

of young promises

Talent

identification

I

II

III

IV


Imagine having more info than the rival, and be better prepared in the “battle

game” analysis, both in attacking and defensive mode

Match Preparation

17

I

� Identify best attacking line up

— When does develop in a goal opportunity?


— When defended / not defended by who?

� Identify best defensive line up

— Where is concentrated opponents possession loss?


— When does develop in a goal opportunity?

— When defended / not defended by who?

� React to unexpected opponents line up or

substitutions

— From an offensive point of view

— From a defensive point of view


Imagine having a detailed follow up of a given player, and even a projection of

how he’ll perform in following games

Player follow up

18

II

� Identify player “static” performance


— When linked to which other players?


� Identify player “dynamic” performance

— In which moment of the season?

— In which competition?

— With how many minutes played already?

� Foresee player performance

— Given next opponents

— Given next competitions

— Given its individual performance cicle

11%

17%

24% 23%

32%

36%37%

45%

38%

35%

Sep

tem

be

r

Oct

ob

er

No

vem

be

r

De

cem

be

r

Jan

ua

ry

Feb

rua

ry

Ma

rch

Ap

ril

Ma

y

Jun

e


Imagine concatenating following matches and player analysis to be able to plan

in advance and react to inconveniences

Season planning

19

III

� Maximize your team results

— Given players projected performance

— Given incoming opponents

— Trading off (“Montecarlo simulation”) between:

— Players effort accumulation

— Probability to win incoming games

— Impact on whole season of next games

� React to unexpected inconveniences

— By player injury

— By better / worse position than expected

— In a competition

— Against an opponent

— With a given player


Imagine having thousands of eyes exhaustively following your desired youth

players, with the best cost-effective methodology

Talent identification

20

IV

� Identify the youth talent elsewhere


— Under which tactic schemes?

— With which specialization?

— At what age?

� Calculate how a new player would integrate

in your team


— Under which tactic schemes?

— With which specialization?

� Obtain thousands of reports at best cost-

effective methodology

Thank you very much for your Thank you very much for your Thank you very much for your Thank you very much for your

attention!attention!attention!attention!

September 2013

Dir. Jaume Sués Caula

Managing Director

[email protected]

Avda. Doctor Pouplana 18-20 | 08950 Esplugues del Llobregat (Barcelona) | Fix. +34 934 732 482 | Mob. +34 610 525 034

Soccer Analytics, or how to see beyond 300 Million eyes

Sports

soccer analytics

plays of manchester

half plays goal

sport analytics

attack of manchester

field areas

plays end

manchester city probability