How data and analytics work at Chelsea from top to bottom

Posted by Patria Henriques on Wednesday, May 8, 2024

(Other contributors: Tom Worville and Simon Johnson)

On a normal day at Cobham, the majority of Chelsea’s first-team squad will arrive ‪between 10am and half-past‬. Some will head to the canteen for breakfast, while others will have eaten theirs at home. When they gather in the dressing room to get changed, part of the routine includes putting on the Catapult vests, fitted with GPS units, that have been left in their boot holes.

Advertisement

As they put their kit on, the eyes of some will be drawn to either of the two TVs in the room. One will be showing an edited highlights package of each player’s best physical actions from the previous match, the other displays a more basic table of their performance metrics: total distance covered, top speed, number of high-intensity sprints and more. The thinking is that appealing to the players’ competitive nature is the best way to get them to engage with the data, and the screens are frequently a source of banter as individual numbers are compared.

Out on the training pitch — which is lined with Catapult sensors — head coach Frank Lampard and his staff will be accompanied by an analyst carrying a laptop or tablet. Chelsea have refined their system to provide live data reports that can be viewed as the session progresses and every player has his own individual targets to hit for each performance metric every day.

Within an hour of the end of a session, the fitness coaches will be provided with a report; the first page containing an easily digestible statistical summary of what happened during the session, and the subsequent one consisting of a more detailed breakdown of how each player is doing relative to their performance targets. This information is discussed in the afternoon and incorporated into the training plan so that, by the next morning, the coaching staff have a clear idea of how much work to load on to every member of the squad.

It’s a smooth, sophisticated process in keeping with Chelsea’s status as one of European football’s modern giants — and it’s just one example of how the club’s analytics operation has grown in influence and resources during the second decade of Roman Abramovich’s ownership.

Matt Hallam is not a name you’re likely to recognise, but he is a very significant figure in the story of Chelsea’s evolving relationship with data analysis.

Advertisement

A psychology graduate of Loughborough University, he arrived at Cobham on an internship to work under Tim Harkness, who was brought to the club with Carlo Ancelotti in 2009 and now holds the title of head of sports science and psychology.

So impressed was Harkness with Hallam that when Michael Emenalo was made technical director in July 2011 and signalled he was looking for someone to help him with the technological side of gathering and presenting information on players, Harkness told him he knew someone already at Cobham who would be suitable.

Emenalo explained to Hallam what he wanted: a bespoke database that would house all of Chelsea’s scouting reports and other information on players they were monitoring, and present it in a fashion that would be clear and accessible to any key staff who wished to view it, as well as owner Abramovich. Above all, Emenalo wanted a tool to help increase the level of internal transparency of player signing recommendations made to the club’s hierarchy.

Hallam asked for two weeks to come up with a solution, but returned to Emenalo after only one. The idea he presented was to use a data analysis tool called QlikView as the platform for a Chelsea-specific player database that would allow for the club’s scouts to upload their reports, video clips to be viewed and edited into highlights packages, and advanced metrics compiled.

The concept struck Emenalo as much more secure and discreet than the approach favoured by many other clubs of having their scouts upload reports to Scout7, a platform more vulnerable to unauthorised access (Liverpool paid Manchester City £1 million in a private settlement in 2013 after a complaint that three Liverpool employees had repeatedly accessed City’s Scout7 account). Hallam was given the green light, went away and created Chelsea’s player database.

Advertisement

Most top clubs now use sophisticated analytics programmes and databases to underpin their recruitment but, in 2011, Hallam’s advocation of QlikView and adaptation of the software to suit Chelsea’s specific needs put the club ahead of the curve. He has worked closely with head of international scouting Scott McLachlan in the years since to make their process of talent identification more data and analytics-driven, and is regarded internally as indispensable.

As the loan system Emenalo devised rapidly expanded in the following years, Hallam’s database was also used to monitor players Chelsea sent out to develop at other clubs.

Fabian Unwin, a former sergeant in the Norwegian air force, is the analyst specifically detailed to work on the club’s many loan players and his responsibilities include sifting through the stream of information that travels from loan clubs to Cobham, as well as compiling tailored video packages from their matches to feed back to them.

For the players involved, the process is efficient and individually-focused.

Richard Nartey is a Chelsea academy graduate who spent this past season on loan at Burton Albion in League One, before being released by his parent club. “When I was at Burton I could go on this app that they’d set up and I could view all my clips that the loan analyst person puts together,” he tells The Athletic.

“The loan analyst sends it to Chelsea, and Chelsea send it to you. I have had all my clips from this year and I have put them in a shorter highlights reel from my time at Burton lasting eight, nine minutes. If people want to see more, I can get it. Thanks to Chelsea I have them. Clubs that are looking for players in a certain position can see what they are looking for.”

Nartey was already accustomed to this level of detailed feedback from his journey through Chelsea’s renowned academy, where every age group from under-14s onwards have a dedicated performance analyst. “If you go and speak to them about it, they will give you any stat from any game,” he adds. “One time, they gave everyone their pass completion rate for short distance, medium distance and long distance.

Advertisement

“After every game, you get the amount of distance you covered, you can look back at distances you covered on every game you played, the fastest speed you’ve hit in any match, the deceleration time. They have everything for you. You kind of learn as they tell you. I didn’t know they could track some of the things they give you.

“One of the main things I learned was… it’s called a VC load, it’s how much force you put through your body. The higher your VC load, the more force is going through your body and that means potential injuries can be worse, if you don’t learn how to deal and cope with different things. They can manage your training, depending how your VC load is. They also showed my peak speed at the start of the year and halfway through the year, to show whether it was working or not.

“You can call them any time. Last season, when I was trying to go on loan I spoke to Michael Emmerson (under-23s’ lead performance analyst). I told him that I was trying to go on loan and wanted a highlight reel of my under-23s clips. He sat down with me and we made a six, seven-minute video of different parts of my game: heading, long passing, short passing, tackling. He does that for anyone who asks to go on loan.”

Some of the more senior figures in Chelsea’s analytics operation have been at the club as long or longer than Hallam and Harkness.

James Melbourne, head of first-team analysis, joined from Prozone in 2005. First-team match analyst Christy Fenwick was recruited from Wolverhampton Wanderers two years later and is credited as being the first person at the club to collect and analyse performance data on individual players and the team as a whole over differing periods of time.

Chelsea are also open to hiring analysts who have worked in other sports.

One of the more recent additions is lead recruitment and data analyst Dan Pelchen, who arrived at Cobham in the summer of 2018 after spending six years with Melbourne-based Aussie rules football club Collingwood. Pelchen initially came to England to learn more about how analytics work in football, but is now considered key to the unit.

Advertisement

First-team performance analyst Paul Quilter is one of several members of staff who, like Hallam, were brought in on internships and impressed enough to be offered permanent roles and contracts.

Giving opportunities to smart young people has always been prominent in Chelsea’s thinking when building out their analytics operation, and brings the added benefit of engendering a deeper sense of loyalty among those hired. The vast majority have academic backgrounds in sports science, and all pride themselves on keeping a low profile.

“As a club, we’re not data-led,” Ben Smith, Chelsea’s head of research and development, said in a presentation at the Sports Analytics Innovation Summit in 2013. “We’ve got a huge range of experts and we try to use our data as a support tool. Very rarely are we going to present data and make a decision based solely off that — it’s just an extra source of information that our coaches will take on board; and if it’s good data, hopefully it’s going to have an impact and be seen in our results.”

It’s fair to say, however, that advanced data analysis informs virtually every decision made on the football side of the club.

Many of the most significant recent advances have come in the sports science department, reformed during Antonio Conte’s two years as manager and overseen by Harkness.

QlikView is still used at Chelsea, but a technology partnership with Microsoft has enabled the club to transition to using a more powerful data visualisation tool, Power BI, for much of their data analysis. It is this software which allows the club’s analysts to monitor live data during training sessions and to quickly create the reports which are sent to Lampard’s fitness coaches each afternoon and inform the next day’s session.

Training workload is now highly individualised, subject to adjustment based on proximity to matches and a number of other variables. Lampard has benefited this season from the strides data analysis helped Chelsea make in maintaining the fitness and sharpness of substitutes and squad players during Conte’s title-winning 2016-17 campaign, when Cesc Fabregas was one of the club’s best performers despite only starting 13 league matches.

Advertisement

Data gathered on academy prospects who train with the first team is particularly useful when assessing whether or not they are ready to play senior minutes, and is regarded within the club as more indicative than anything that can be gathered from youth football. Sources have told The Athletic that Billy Gilmour’s impressive physical metrics in training and games backed up Lampard’s belief that the teenager was capable of making an immediate impact in senior football, despite his relatively slight frame.

Some first-team players are more interested in their performance data than others. It should come as a surprise to no-one that Cesar Azpilicueta is always keen to learn everything he can about his physical output on the pitch, from the total distance he covers to the number of high-intensity sprints he completes.

The sophistication of Chelsea’s analytics operation has other uses. Devising the individual rehabilitation regimes to help Callum Hudson-Odoi and Ruben Loftus-Cheek come back from their respective achilles ruptures was a joint effort between the club’s medical and sports science departments, with data analysis fundamental to the decisions of when both players could be given the green light to step up their recoveries in matches for the under-18s and development squad.

Whether it is scouting, managing loan players, assessing first-team performance or injury prevention and recovery, the professionalism and progressive mindset underpinning Chelsea’s analytics operation stands up to comparison with any of Europe’s better run elite clubs.

The work they do off the pitch will continue to be vitally important to any success achieved on it.

(Photo: Darren Walsh/Chelsea FC via Getty Images)

ncG1vNJzZmismJqutbTLnquim16YvK57kHJsamthaHxzfJFpZmlwX2WBcK%2FHnqOsnZFirq%2Bty7KropujYrGiwMBmnqmrXw%3D%3D