How Opta produces and develops its unique data

Opta’s former Commercial Director, Mike Strong, outlines how the industry’s leading data and stats provider, compile and produce their data

Categories: All Sports, Data, Professional

Opta started out in 1998 as a supplier to the media sector originally focusing on Broadcasters, Newspapers and Digital, which increased to Professional Clubs, Betting (mainly fixtures and results), Mobile, Fantasy and Sponsors.
During this period betting revenues have increased massively due to many factors; general relaxing of gaming restrictions, increased legal betting geographies, introduction of betting exchanges, 24 hour internet betting, in running betting and the general social acceptance of gambling.

The betting industry has always needed quick accurate data but this growth, particularly with the in running, spawned a need for very fast data (< 5secs ) which has been satisfied by the emergence of new suppliers of very fast and accurate basic data via an army of scouts at the grounds all over the world, Running Ball and Bet Radar would be the most prolific suppliers of this data of key events for football, basketball, darts, etc and tennis mainly through IMG.

Secondly as the accuracy and depth of Opta data became better understood new demands for this detailed data came from the betting industry both from the bookmakers and professional punters for modelling. The key components for modelling data are accuracy and depth, please see Rob Esteva’s earlier article on this site. Opta has is an abundance of excellent modelling data on football, cricket and rugby which new customers tested, tried and bought.

What is in the detailed data ? The Opta data is all collected live in house by very skilled analysts (none is outsourced which produces world wide consistency) and then is checked post match, play by play, for errors. The data focuses on every touch on the ball, so in football, where on the pitch, the players involved, pass completions, challengers, tackles, shots, chances created, etc a typical football game can have between 1500-2000 events (although it depends on style of play so Stoke City would average considerably less then Barcelona). There is some subjective analysis but it’s mainly totally objective and you can go to any of the offices for a comprehensive collection demonstration.

How good is Opta data for modelling ? . . . . . On the plus side are accuracy of the data, consistency, depth, coverage and quantity. What it lacks is any real subjectivity?

At Opta our analytics team did some work on assessing striker effectiveness. This was done by analysing all shots with a variety of factors like position on the pitch, goalkeeper position, the defence, etc to form an expected goals total and then comparing this to the actual.

Below is a table produced for both the individuals (top 10 and bottom 10 shown) and teams, this was from early 2013/14 Aug to Mid Oct. In the previous season Gareth Bale was top having scored around 5 goals more than expected.

Player Team TotalShots Non-blockedShots ExpectedGoals Goals AverageChanceQuality Difference
Daniel Sturridge Liverpool

22

20

3.2

6

0.144

2.8

Aaron Ramsey Arsenal

16

12

1.2

4

0.076

2.8

Loïc Remy Newcastle United

16

14

2.4

5

0.152

2.6

Romelu Lukaku Everton

10

8

1.6

4

0.164

2.4

Yaya Touré Manchester City

13

9

1.7

4

0.129

2.3

Lukas Podolski Arsenal

3

2

0.2

2

0.075

1.8

Leighton Baines Everton

8

7

0.4

2

0.045

1.6

Adnan Januzaj Manchester United

9

6

0.4

2

0.046

1.6

Christian Benteke Aston Villa

11

9

2.5

4

0.229

1.5

Ravel Morrison West Ham United

6

5

0.5

2

0.087

1.5

Robert Snodgrass Norwich City

11

7

1.2

0

0.113

-1.2

Kenwyne Jones Stoke City

9

7

1.3

0

0.142

-1.3

Leroy Fer Norwich City

6

4

1.4

0

0.232

-1.4

Nikica Jelavic Everton

9

7

1.5

0

0.166

-1.5

Danny Graham Hull City

9

7

1.5

0

0.167

-1.5

Paulinho Tottenham Hotspur

25

19

2.5

1

0.101

-1.5

Roberto Soldado Tottenham Hotspur

15

11

3.6

2

0.237

-1.6

Samuel Eto’o Chelsea

10

8

1.7

0

0.165

-1.7

Jonathan Walters Stoke City

11

8

2.3

0

0.209

-2.3

Papiss Demba Cissé Newcastle United

18

13

2.4

0

0.135

-2.4

 

 

 

Team TotalShots Non-blockedShots ExpectedGoals Goals AverageChanceQuality Difference
Arsenal

101

73

9.1

14

0.090

4.9

Manchester City

109

87

12.4

16

0.114

3.6

Aston Villa

86

63

7.7

9

0.090

1.3

Fulham

58

44

4.0

5

0.068

1.0

Manchester United

97

76

9.3

10

0.096

0.7

Everton

101

76

9.7

10

0.096

0.3

Cardiff City

71

54

6.8

7

0.095

0.2

Swansea City

96

80

8.0

8

0.083

0.0

Liverpool

91

72

11.1

11

0.122

-0.1

Hull City

68

45

6.2

6

0.092

-0.2

Crystal Palace

69

49

5.4

5

0.078

-0.4

Southampton

87

64

7.8

7

0.089

-0.8

West Bromwich Albion

77

62

7.8

7

0.101

-0.8

Newcastle United

113

86

10.0

9

0.089

-1.0

West Ham United

83

63

8.5

7

0.102

-1.5

Sunderland

93

58

7.0

5

0.076

-2.0

Data availability, there is massive amounts of detailed data available on the internet free, try Who Scored and Squawka, otherwise by subscription which can be used for modelling but for very detailed modelling you tend to need very accurate complete data sets.

Finally, the most amazing thing about data Opta provided was that the strictest demands and requirement for the most accuracy and depth in data didn’t come from the professional clubs to improve performance, it came from modellers who effectively were playing with their own money!

Join the discussion

Home Forums How Opta produces and develops its unique data

This topic contains 4 replies, has 4 voices, and was last updated by   warren zephaniah 1 day, 3 hours ago.

  • Author
    Posts
  • #815
      Sports Trading Network 
    Keymaster

    Opta’s former Commercial Director, Mike Strong, outline show the industry’s leading data and stats provider, compile and produce their data

    [See the full post at: How Opta produces and develops its unique data]

  • #816
     BS Khoo

    Hi,

    I am very interested in developing mathematical models to calculate fair 1×2 odds for a football match. Could you introduce me to some books on this area?

    • #1791
        ®γσ, Lian Hu ENG 
      Participant

      For fixed-odds modelling, you might refer to Dixon&Coles1996.

      I am learning staking model and also rebirth model for soccer live-betting…

  • #817
     eightyape

    are you any good with behavioural tracking algorhythms?

  • #3602
      warren zephaniah 
    Participant

    The key segments for displaying information are exactness and profundity, please observe Rob Esteva’s prior article on this site. Opta has is a wealth of amazing displaying information on football, ReadyDissertationHelp cricket and rugby which new clients tried, attempted and purchased.

You must be logged in to reply to this topic.