‘For years we’ve heard stories of real-life managers and scouts using our data to help with the recruitment process,’ says Miles Jacobson, Studio Director at Sports Interactive.
Jacobson’s comments were made in 2014 after Sports Interactive signed an agreement to supply data from Football Manager to Prozone. This data would be used in Prozone Recruiter, a database designed for scouts and managers looking to find and assess potential signings.
In recent years, as football clubs have started to embrace data analysis, their approach to the game has in some ways started to resemble Football Manager. The simulated world of data within Football Manager has become entangled with the real data analytics of the professional game. Scouts search databases of potential signings. Managers consult detailed match statistics when developing tactics and evaluating the performances of players. And as real life has become more like the video game, the video game has adapted to reflect some of the changes that the data revolution has made to football. Players of Football Manager can now see the expected goals stat after each match and hire their own data analysts.
The Football Manager series, developed by Sports Interactive, is the one of the world’s most popular management simulation video games. It allows players to take control of a club of their choice and try to emulate the success of real-life managers such as Jürgen Klopp and Pep Guardiola.
Whereas other popular football-based video games, such as the FIFA series, make use of state-of-the-art graphics to make the game feel realistic, Football Manager has always been focused on data and statistics.
When deciding which player to sign, the player of the game can look over a detailed list of attributes, each given a numerical rating between 1 and 20. When looking over a previous match to work out what happened, the player can access detailed statistics about how many tackles each player made, how many passes, how many shots, and so on.
The game began life in 1992 as Championship Manager, a name the series would keep until 2004, when the game’s developers split from its publisher. Back in the early 1990s there was little attention paid to data and statistics in football. Clubs did not hire data analysts. Sports analytics companies such as Opta and Prozone had yet to be founded.
Since then, things have changed drastically. Data analytics is now a major part of professional football and continues to grow in importance.
Over the years, the Football Manager series has become more detailed, expanding its data and improving its statistical models in order to more accurately simulate the real world of football. At the same time, football clubs, pundits, and data companies have also expanded and improved their use of data to understand the game more thoroughly and make better decisions. This has led to overlaps between the video game’s data and the use of data by real-life clubs.
With the use of data in football expanding, it was unsurprising that Football Manager’s database would prove of interest to clubs. In 2008, Sports Interactive signed a deal with Everton which allowed the club to use the game developer’s database to search for new players and staff. The deal with Prozone made six years later meant that many more clubs would have access to this data.
‘Prozone Recruiter has been built to supplement the intuition of scouts and coaches by delivering detailed performance information on over 80,000 players worldwide,’ Prozone’s CEO, Thomas Schmider, explained at the time the service was launched. ‘The Sports Interactive database is a highly accurate and valuable resource that will further enhance the recruitment services that we provide.’
Prozone were one of the first companies to provide sports analytics to Premier League clubs. This was not a case of a small upstart trying to piggy back on the success of a popular video game, but a serious and established analytics company seeing the potential of the Football Manager data. The collaboration with Prozone went two ways. In exchange for access to the Football Manager database, Prozone provided Sports Interactive with data about player movement and fitness to help improve the game’s realism.
One of the reasons why the game’s data was so attractive is that the game’s researchers far outnumber the scouts employed by any major club. They gather detailed data from leagues all over the world, including countries and leagues which did not normally receive much attention. When combined with the historical data which Prozone had itself gathered, including archived video footage, this data from FM meant that Prozone Recruiter offered a comprehensive and innovative scouting platform.
How Football Manager Gathers its Data
The latest version of the game, Football Manager 2021, officially abbreviated as FM21, contains over 300,000 players and 60,000 clubs. Since 1992, around 900,000 different players have featured in the game. In a couple of years’ time that number is likely to hit one million.
In the game, each player is given ratings for a range of attributes. There are fourteen technical attributes, such as passing and tacking, fourteen mental attributes, such as determination and composure, and eight physical attributes, such as pace and balance. There is also a set of goalkeeping attributes for goalkeepers and every player has six hidden attributes, including consistency and injury proneness, which cannot be seen within the game, but which can be referenced by coaches and scouts in their reports.
All these attributes are rated from 1 to 20. A player with a score of 20 for his pace will be the quickest, or joint-quickest, within the game and a player with a rating of 1 for the same stat will be the slowest.
The ratings are based on work by a network of over 1,300 researchers who help to construct the database. They provide information on thousands of players around the world from first teams to youth teams. There are 86 full-time researchers, with major footballing countries, such as Germany, Italy and Spain, having two head researchers. In England, there is one for the Premier League and one for the lower divisions. Each of these researchers has a team of assistant researchers underneath them, who are often football fans keen to give detailed information about their local clubs, watching reserve and youth team games as well as first-team games to gather information.
These researchers are carefully selected. ‘Not all FM players would make good researchers and not all obsessive fans would make good researchers as they may be too biased,’ says Mark Woodger, Head of Research for Sports Interactive. ‘So it does need an analytical mind, attention to detail and to be able to take a step back from your own biases as a fan of a particular club.’
This army of researchers has slowly built up over the years. When the game was first being developed by Paul and Ov Collyer in their spare time, there were no reserachers on hand to scout players and help them get the information right. In the first version of the game, players could only manage teams from the top four English divisions. The game did not include real players and players in foreign teams did not even have names, but were labelled with a number from 1 to 11 based on their position. Real names were included in the second version.
At first, data about players was taken from Rothmans Football Yearbook, an annual book of statistical information about the previous season in English football. Soon the developers started to contact fanzine writers to get first-hand information from fans. As the game grew, the number of researchers grew and became more organised and professional.
As well as analysing current ability, the researchers also estimate a player’s potential ability, so that a talented youngster will see his attributes increase over time as the game progresses. Each player is given a potential ability score between 1 and 200. This element of prediction is hard to get right. Just as scouts in real life do not always get it right when recommending the next big talent, so the youngsters identified by the researchers do not always go on to become stars outside the game.
Cherno Samba is the classic example of a young player whose potential never reached the heights predicted by the game.
In the 2001-02 edition of Championship Manager, Samba starts the game as a 15-year-old at Millwall. As the seasons go by, he develops into one of the best strikers in the world. Word quickly spread about his potential and for many players of the game Samba became one of their key signings. In real life he never made it to the top of the game. After a journeyman’s career which included spells at Haka in Finland, Panetolikos in Greece and Plymouth Argyle in England, he retired at the age of just 29 due to injury.
Sometimes, though, they have got it right. One successful prediction was Lionel Messi. When he was still a teenager in the Barcelona youth team, he was identified by the game’s researchers and given great potential stats. The Scottish manager Alex McLeish recalls a time when he was in charge of Glasgow Rangers and his son, who was a keen player of Football Manager, recommended the unheard-of youngster to his dad as a potential signing: ‘He told me that this guy was going to be the best player in the world. I said “ok son” and gave him a pat on the head.’
The idea that Rangers may have been able to sign Messi if only McLeish had listened to his son makes for a great anecdote, although in reality it is unlikely that the Spanish club would have sold such a talented youngster.
Because of the difficulty of getting predictions right, the game has introduced a system called ‘minus rating.’ This varies a player’s potential ability in each new save started in the game. The more uncertain the researchers are about whether a player is likely to one day be a success, the more the algorithm will be allowed to vary the player’s potential within the game. The game’s developers also make last minute changes based on feedback from fans who have played the beta version before the official release. For the most recent version, they made 671,753 changes based on such feedback.
‘We will get it wrong sometimes; real-life managers or scouts get it wrong all the time—saying they’re not good enough or saying they’re too good,’ says Woodger, ‘so we have those checks and balances. Usually, we come to a consensus that someone is over or under-rated.’
Getting the data about player ability correct is important for the realism of the game. Players generally want the in-game experience of managing a football team to feel as close as it can be to real life. It is this realism that makes the database attractive to clubs.
For Matt Neil, working as a researcher for Football Manager was the starting point for his career as a football analyst.
He started off at the age of just 15, providing information for Sports Interactive about Truro City, a Non-League club based in Cornwall. He was only the second person under the age of 16 to work as a researcher on the game. The developers were
impressed by his work and he soon moved on to providing information about another of his local clubs, Plymouth Argyle, who were higher up the football pyramid.
Scouting the Plymouth squad for Football Manager meant that Neil was around the club on a regular basis and got to know some of the players and the staff. One day, when he was talking to the club secretary, he recommended a young player who was scoring lots of goals in Non-League football. The secretary passed on the recommendation to the club’s manager, Carl Fletcher, who asked Neil for a statistical and video presentation on the player.
Fletcher was so impressed with the work that he offered Neil a job as the club’s Lead First Team Football Analyst when the position became vacant at the end of the season. Neil spent seven years at the club before moving to Rotherham United. He is currently Recruitment Analyst for Salford City, the League Two club owned by ex-Manchester United stars including Ryan Giggs, David Beckham and the Neville brothers.
He still works as a researcher for Sports Interactive, but he does not rely on information from the game when analysing players: ‘When I watch players I don’t ever judge them on their Football Manager ratings. If someone has got a passing attribute of 17, I don’t immediately think they’re going to be an unbelievable passer, because that might be after someone’s watched them for three games.’
The level of analysis and data in football has grown so much that the Football Manager database is being superseded in depth and detail. Scouts from football clubs now regularly search through databases of potential recruits provided by companies such as WyScout and the Twenty First Club. Some clubs, such as Liverpool, even have their own private databases. As Neil’s comments make clear, data from Football Manager alone is no longer enough, although it can still be a useful starting point and working on it certainly helped his career.
The example of Matt Neil shows that the skills needed to gather data for Football Manager can be developed into those required for a job at a real football club.
Just as clubs’ player recruitment has been influenced by Football Manager, so the growing use of match analysis data by clubs and the media is having an impact on the video game.
The series has always relied on match data. In early versions of the game, there was no match engine to allow players to watch footage of matches. All the player had was statistics and a written commentary. This was seen as a negative when the game’s original developers were first trying to find a publisher. Electronic Arts rejected the game because it did not include enough ‘live action.’
The use of statistics meant that players of the game were looking closely at numbers such as the number of passes and the number of key tackles made by each player long before data analysts were a regular part of the background staff at football clubs. A 2D match engine was introduced in 2003 and in 2008 a 3D match engine was included for the first time, but data continues to be important to how people play the game.
The game now creates an abundance of match data that rivals the amount of data collected in real-life football matches. The game now even includes an expected goals metric, designed by the developers to work with their match engine. The game also allows players to see data presented as graphs and pitch maps, mirroring the ways in which data analysts present their work to managers in real life.
Mostly, the data is realistic and strives for accuracy, but a few things have been sacrificed in order to make the game more enjoyable. ‘One of the biggest complaints we have from users is the amount of injuries their players sustain in the game,’ Miles Jacobson has revealed. ‘In reality, we have only 70% the frequency of injuries that occur in real professional football.’
As football has become more data driven and started to resemble aspects of the Football Manager series, real-life data analysts have started to feature within the game. They were first included in 2017 and their importance has increased in more recent versions. One of the data analysts included in the game’s database is Salford City’s Matt Neil, who has gone from a researcher for the game to a football analyst to a character within the game.
Share this article
Our team provides news and insights from the cutting edge of football analysis.