Seen this way, Manchester City’s summer transfer campaign looks simple enough. The midfield needed fixing and in came Bouaddi from Lille for €95 million, Anderson from Nottingham Forest for €135 million and Enzo Fernández from Chelsea for €145 million. That is €375 million, rising to €526 million once the arrivals of Ndiaye from Everton, Rulli from Olympique Marseille, Detourbet from Troyes, Monga from Leicester and Charles from Sheffield Wednesday are added — the last three were sent out on loan to Monaco, Swansea and QPR. City did sell €337 million worth of players, but there is no question they spent.
Having that much money available, though, does not mean it has to be spent. City’s scouting department can draw on an enormous database containing reports on 650,000 players. The BBC has looked at how the club’s statistical analysis works. In Manchester, one of the main aims is to reduce the risk attached to every transfer, whether buying or selling a player. City assesses players according to a range of factors: quality, overall characteristics, the system they play in at their current club, and the way their manager asks them, and the team, to position themselves and read the game.
Once those elements have been assessed, the aim is to understand not only how good a player is, but how good he might become at the Etihad. Since winning the Treble in 2023, Manchester City have reshaped their squad extensively according to this approach. Only four players remain from that historic team: Haaland, Aké, Foden and Rúben Dias. It is worth stressing, though, that while data has significantly changed the way City operate in the transfer market, traditional scouting remains the decisive part of the assessment when deciding whether to sign a player.
The starting point for any new arrival is City’s database, which contains 650,000 players from around the world. The next step is to define the profile: what kind of player does the club want? Usually it starts with the position, although the process is more complicated than that suggests. City, for example, divides full-backs into three categories: those who are more defensive, those who move inside and those who play high and wide. Once the requested profile has been established, the data departments apply a series of filters to assess the player’s overall level, including his ability to use both feet. The enormous initial database is then reduced to a list of between 20 and 50 players.
That list is compared with the one produced by the scouts. Usually there is considerable overlap between the players identified through the data and those already on the scouts’ radar. More specific parameters are then applied to order the names and establish priorities within the list. The scouts carry out further checks, reducing the group to a maximum of around six players. The club then goes deeper still, looking at injury history, athletic performance and behavioural habits.
All the information is brought together in individual reports known as acquisition packs, which contain the main indicators used to make a decision on the player. Among them is a radar chart, which scores a series of specific measures relating to his performances on the pitch. There is also a utilisation chart, which records how often a player has been used over a three-year period, alongside an assessment from the traditional scouts indicating how highly his profile is regarded. Every player also receives a current impact score, intended to predict what his immediate impact might be after joining the club, and a potential impact score estimating the contribution he could make in the future.
Perhaps the most interesting figure in the acquisition packs, though, is the action value score. It effectively ranks the player alongside City players and those at rival clubs who perform a similar role. Once the reports have been analysed, the list is reduced again to two or three players, taking into account the financial value City has assigned to each profile. The data contributes significantly to establishing the price range considered appropriate for a player, but the valuation is never expressed through a single number. At that point, the priority target is identified.
All the names on the shortlist continue to be followed at the same time, however, in case circumstances change — if the first choice becomes unavailable, for example, or proves impossible to sign. The ability to adapt quickly to developments in the market is one of the areas in which the data department believes it has grown most. Elliot Anderson is a good example. He looked like a natural signing for a major Premier League club. At 23, he had become one of Nottingham Forest’s key players, was English and had established himself with England. On the surface, signing him looked like a fairly simple decision.
But the way City analysed the deal shows how meticulous its process is, even when, at least apparently, there would have been little need to go so far given Anderson’s obvious qualities. The fact that the club still ran its algorithm and followed all the usual steps says a lot about elite football now, but also about the limits of data when taken in isolation. Because, perhaps surprisingly, Anderson’s numbers had not initially attracted much attention.
At the start of last season, despite the uncertainty surrounding Guardiola’s future, City knew that in the summer it would need at least one central midfielder, because Bernardo Silva was expected to leave and the futures of Rodri and Nico González appeared uncertain — in the end, both were sold. The club knew that replacing Bernardo and Rodri with two players of the same type would be practically impossible. It was therefore looking for a midfielder who could be equally effective in the first phase of build-up and in the more advanced central areas. As the season went on, doubts also increased over Tijjani Reijnders’ future, while Enzo Fernández was regarded as the preferred candidate to replace the Dutchman.
Anderson, at least initially, was not particularly high on the list. His numbers did not immediately suggest what he would later become: an almost inevitable signing. Looked at more closely, though, City realised that Anderson’s case also exposed some of the limits of data analysis. The midfielder did not score many goals for Forest and did not provide assists consistently; Nottingham are not generally a possession-oriented team in the way City are and, as a result, some of his broader numbers in the algorithm did not stand out. The analysts therefore had to consider something fundamental: Anderson usually played closer to his own goal than he would at Manchester and, as a result, had fewer opportunities to show his qualities in the final third.
The club studied his role at Forest in detail and modelled how he might operate in a team such as City, which generally plays closer to the opposition goal, dominates possession and has a higher level of quality around him. Among the measures included in his radar chart were defensive recoveries, duel success rate, line-breaking passes, aerial duel success rate and defensive errors. That last figure, in particular, required a different reading.
Because Forest base so much of their game on transitions, some of Anderson’s errors were linked to the area of the pitch in which he was being asked to operate. He might, for example, win the ball very close to his own penalty area and then lose it immediately because he was forced to attempt a long and risky pass, with the team needing to move forward quickly. In a similar situation at City, according to the club’s projections, Anderson would have more chances to play a short pass, giving him a better chance of completing it.
Anderson’s numbers improved significantly as the season progressed, prompting City’s scouts to watch him more frequently and more closely. The midfielder’s figures compared very favourably with those of the players already in the squad and with several potential midfield targets. The checks carried out by the data department, the scouts and the other specialists led City to view the deal as low-risk, even after Anderson’s price rose beyond the valuation initially assigned to him by the club.
At City, this is how things have worked for years. Before the Treble season, the club’s internal indicators were already suggesting that the squad would soon need a major restructuring. Selling players such as Kevin De Bruyne, Kyle Walker, Riyad Mahrez and Ilkay Gundogan was not easy, given what they had achieved and the fact that they were still performing at the highest level. The data, however, allows the club to identify negative trends well in advance, often two or three years before a player is actually sold.
These parameters help City identify the right moment to sell a player, perhaps even before the club interested in signing him has noticed a decline in his performances. When those trends emerge, the club can plan well ahead. It can decide whether to start introducing a player from the academy or turn to the market to replace the departing player. The arrival of Bouaddi from Lille this summer is a good example. At 18, City knows that the midfielder will not necessarily make a significant impact this season, but believes that within two years he could become much more important.
After the Treble season, the brief given to City’s scouting and recruitment departments was clear: lower the average age of the squad and build a group of players capable of sustaining success. Doing that in Manchester, though, comes with a condition: City must continue to compete for trophies. Lowering the average age of a squad is one thing; doing it while continuing to win is much more difficult.
In previous years, for example, City’s systems had identified three Brighton players, Moisés Caicedo, Marc Cucurella and Carlos Baleba, who were later held up as examples of data-led scouting. In Manchester, though, there was a feeling that developing players of that kind would take too long for them to become immediately useful to a team whose objective was to win titles. At the same time, City wanted to develop players who could later be sold.
City have nevertheless managed to bring down the average age of the squad significantly, which is now the third-youngest in the Premier League, behind Newcastle and Chelsea. With the arrivals of goalkeeper Gianluigi Donnarumma, defender Marc Guehi and midfielders Fernández and Anderson, the club has also added players who can grow alongside the new generation while already possessing extensive experience at the highest level. According to City’s data, the current squad will reach its peak in two or three years. The same projections indicate, though, that the team should already be among the leading contenders for the title this season. Perhaps that is no coincidence.