Every Manchester City signing starts in the same place: a list of 650,000 footballers. BBC Sport has been given access to the club's data operation, and the picture that comes back is less science-fiction than you might expect. The database is enormous. What City do with it is fussy, slow and heavily cross-checked against people who still watch games with their own eyes.
The department works to one question — what does it take to win? — split into three jobs: find players, judge them, and work out whether City could make them better. Last season their model had Arsenal as title favourites with City somewhere close behind. That is exactly how it finished. This year the club are keeping their projection to themselves.
From 650,000 names to about six
The funnel is brutal. A brief comes in, usually starting with a position, though even that is not simple: City sort full-backs into three separate types, depending on whether they defend, tuck inside, or stay high and wide. Filters cut the 650,000 down to somewhere between 20 and 50 names. That list gets laid over the scouts' own longlist, and there is normally a heavy overlap between the two.
Sharper metrics rank what survives. Scouts then dig further and the list drops to a maximum of around six. Those six get the full treatment — injury history, athletic data, behavioural checks — written up into documents the club calls acquisition packs, which include a radar chart of on-pitch metrics, a utilisation chart covering three years of minutes, a scouting grade, a current impact score, a potential impact score, and an action value score that drops the player into a league table alongside City's own squad and rivals playing the same role.
Juventus full-back Andrea Cambiaso came through that process. The data liked him, particularly his habit of stepping inside to build play and the fact he is comfortable on both feet. City did not move for him in the last window. They are likely to keep watching.
Sporting director Hugo Viana is careful about how much weight he puts on any of it. "Data plays a role in our decision-making, but it is just one part of the overall picture," he told BBC Sport. "We know data cannot tell us everything about a player. It cannot fully show a person's character, mentality or behaviour."
Elliot Anderson did not look obvious on a spreadsheet
Which brings us to the signing that best explains the limits of the whole exercise. Elliot Anderson, 23, cost 116 million pounds, is English, plays for the national team and was clearly one of the best midfielders outside the top six. City beat Manchester United to him. On the face of it, no algorithm required.
Except his numbers did not leap off the page. Anderson made the initial cut from the database but sat low in the group. He did not score much at Nottingham Forest or rack up assists, and because Forest are not a possession side, his readings in City's model were never going to flatter him.
So the analysts adjusted for it. They looked at where on the pitch he was being asked to operate, modelled what he would be doing in a team that plays fifty yards further up, and reconsidered readings like defensive errors in that light — a misplaced pass after winning the ball deep in his own half, forced into a long and risky option, would become a simple short pass at City. "Just because a player doesn't have the opportunity to perform a particular action on a regular basis because of the team he is playing doesn't mean he can't do it," a source said.
His readings improved through the season anyway, scouts started watching him weekly, and by the time the price had climbed past City's own valuation the club felt the risk was low enough to pay it.
Knowing when to sell
The same system works in reverse, and that is arguably where it earns its keep. City say declining trends show up in their metrics two or three years before a player is sold — often before the buying club has noticed anything. Kevin de Bruyne, Kyle Walker, Riyad Mahrez and Ilkay Gundogan were all still performing when they left. The data said the curve had turned.
Julian Alvarez is the cleanest example of the model working. Scouts flagged him first; because the Argentine league caps out in City's settings, the analysts had to dig past the surface numbers. They concluded he was comfortably the best player in that division and would become a starter for a top-half club in one of Europe's top five leagues inside two and a half years. The model valued him between 26 and 40 million pounds. City paid River Plate 14 million in 2022 and sold him to Atletico Madrid two years later for 81 million.
Abdukodir Khusanov did not even make the 50-man longlist at first. A closer look showed his on-ball output needed work but his physical data, particularly covering runs, was extraordinary. He signed from Lens in January 2025 and City say the ball work has since caught up.
Only four players remain from the Treble squad of 2023. The average age is now the third lowest in the Premier League, behind Newcastle and Chelsea, and Enzo Maresca's arrival is not expected to change the brief much — he wants broadly what Pep Guardiola wanted, with a preference for more direct full-backs. Ayyoub Bouaddi, signed from Lille at 18, is not expected to make a real mark for two years. The club's own data says this squad peaks in two or three. It also says they will be in the title race now.
Recruitment facts: Manchester City's player database holds 650,000 names. Elliot Anderson, 23, signed from Nottingham Forest for 116 million pounds. Julian Alvarez signed from River Plate for 14 million in 2022, sold to Atletico Madrid in 2024 for 81 million. Abdukodir Khusanov, 22, signed from Lens in January 2025. City have won five of five in the Premier League this season.
0 Comments