Each September, the football community gears up for EA Sports' annual release of its FIFA video-game franchise. Such a phenomenon has this title become, its player ratings are now worthy of headlines on their own.
In a game with a staggering 700 clubs, comprising a mesmerizing 18,000 players, one man oversees the 5.4 million data points that bring those players to life, making FIFA 17 one of the most realistic football titles gaming has ever seen. That man is Michael Mueller-Moehring, and he spoke to ESPN FC to explain how he does it.
"There's always a player from the second division in Switzerland who gets bought on transfer deadline day," he tells ESPN FC, "and all you know about this player is his name, date of birth and his position -- and his position may be as precise as, 'Oh, he's a midfielder.'
"And this player has to go into the game."
So how is an unheralded player obscured from the spotlight given a number that represents his skills, abilities, intelligence and character that can be easily compared with the likes of Lionel Messi?
To start with, guesswork.
"We guess a little bit ... until our people have seen the player in action," Mueller-Moehring says. And it doesn't take long until his people see every one of those 18,000 in action.
His people make up a 9,000-member network that EA Sports calls its data reviewers. While this group does contain some professional-level scouts and coaches, it consists heavily of season-ticket holders -- those who can watch many, many matches, and in person, which is ideal.
Those data reviewers then provide feedback on players through a secure EA Sports website. The feedback they're providing is subjective: the eye test. While EA does use advanced stats to search for trends in players' tendencies, the stats don't dictate what makes a player good or bad in FIFA. Why?
"We have many leagues in the game; no stats provider could offer us data for all these leagues, teams and players," Mueller-Moehring says. "This is also the reason why we use this online database, because it's not possible to buy this data some way -- it just doesn't exist."
Not only do advanced stats fail to cover the sheer number of leagues, teams and players available in FIFA, Mueller-Moehring explains that numbers alone can't possibly accurately represent a player's abilities because of the subtle context of how those numbers are created.
"The stats are, in most cases, not taking into account very specific circumstances," he says. "When you look at passing completion, if you play for Bayern Munich or if you play for Manchester City or if you play for Pep Guardiola, if your system is based on possession, you will have more successful passes than other players, but this doesn't necessarily make you a better passer.
"And when you look at attributes like trapping and ball control, there is data, but the data never gives you the specific situations. Same for tackling."
In addition to the blind spots in advanced stats that Mueller-Moehring laid out, EA Sports also utilizes data fields that comprise each player's mental makeup. There are no stats that can showcase how intelligent or resilient a player is, so data reviewers must judge a player's aggression or other intangibles by the style of play on the pitch.
Where a player plies his trade also determines his floor and ceiling when it comes to technical ability. For example, a very good player in the Premier League is likely to have a technical attribute, like passing, that would be rated between 80 and 84, while a player in League Two -- outside of the rare exception -- wouldn't exceed a rating of 70.
"If Messi were playing in the Irish league, his attributes would drop simply because he's not on the highest level anymore," Mueller-Moehring says. "We want to base our ratings on actual performance data."
When it comes to physical attributes, though, no such guidelines exist. "There are fast and strong players in every professional league in the world," Mueller-Moehring says.
With the help of EA's 300 data editors, all that feedback from those 9,000 data reviewers is collated into 300 different data fields and 35 specific attribute categories, eventually yielding each player's overall rating.
The overall rating is simply the result of a formula that weights attributes for each particular position. Of course, there are some instances in which a player may lack the physical and technical traits that make up a strong footballer, but his intangibles make him a force in modern football.
"A case is Thomas Muller, who isn't good at anything, really, apart from his positioning," Mueller-Moehring explains. "He always finds the right spot on the pitch, it's amazing. But he's not a great dribbler and he can't really strike the ball properly -- his finishing is sometimes really, really off. Shot power is not his strength as well.
"So if you rate Thomas Muller properly, he ends up with a rating that we say doesn't make sense. It's too low."
So what does EA Sports do to correct a case like Muller's? They subjectively boost his overall rating to reflect his standing in the game. EA has the power to do the opposite as well, though Mueller-Moehring says it's never happened.
If that seems haphazard, take solace in the fact that the overall rating literally has no meaning in gameplay, which is entirely driven by a player's specific attribute ratings.
Once all this scouting, research and data analysis is completed for every one of the 18,000 players in FIFA 17, it is then used to determine the strength of each of the game's 700 clubs. Each club's 16 best players' overall ratings are taken into account to come to this score, meaning that while Bastian Schweinsteiger isn't playing for Manchester United, his 83 overall rating ensures he contributes to the club's rating.
For all the data used to dictate how good a player can be in FIFA 17, 5.4 million tiny pieces of information, ultimately it's the subjectivity of EA Sports' data reviewers that call the shots.
"Of course, not everyone has the same opinion about the heading ability of this player from Switzerland's second division," Mueller-Moehring says.