Data has played a role in football for decades, but in recent years its use has expanded rapidly. Football’s data analysts are working out how to encourage innovation within a world that remains built on tradition, ego and rivalry.
After watching his side lose 3-2 to Luton Town, Southampton manager Harry Redknapp turned to one of the club’s data analysts and said: "I’ll tell you what, next week, why don’t we get your computer to play against their computer and see who wins?"
The analyst had given a pre-match briefing to Redknapp and his players using statistics from player-tracking software Prozone, demonstrating how the team should approach the game.
Redknapp’s denouncement was symptomatic of an industry that has often struggled to balance technological advancement with ingrained traditions. This was, however, all the way back in 2005, and Redknapp has never been known as one of the more forward-thinking minds in world football.
The analysts took a back seat for the rest of the season and Redknapp went back to relying solely on experience and intuition. Southampton failed to gain promotion that year.
Fast-forward to 2021 and 19 of the 20 Premier League clubs are using Prozone. Data and analytics has fundamentally changed the way that football is played, managed and viewed.
This was underlined by Brentford’s recent promotion to the Premier League. The club – nicknamed the Bees - is owned by data-driven ex-banker Matthew Benham and operates on a data-led recruitment model.
Brentford’s approach has stolen the headlines, but clubs up and down the football pyramid are now utilising data either through investment in their in-house analytical capabilities or by collaborating with specialist providers.
The slow infiltration of data into the beautiful game can reveal a lot to data practitioners about the importance of patience, the benefits of building a data culture, and working out how to land analytical messages with an audience that, in many cases, might be directly oppositional to them.
A brief background to football analytics
Data is applied to football in three key areas: performance analysis, recruitment and strategy.
In performance analysis, data is used to support pre- and post-match analysis of key player traits, set-piece trends, chance creation, and team shape during various phases of play. Post-match this data can be used to evaluate how the team and individual players performed in relation to pre-determined KPIs as part of the game plan and football strategy.
In terms of recruitment, data is increasingly being used to benchmark and analyse players across different markets against current squad players and potential transfer targets. This is used in conjunction with traditional live scouting methods and video.
Recruitment is traditionally geared towards simply improving the squad, but consideration is increasingly being given to long-term financial gain. This approach has been key to Brentford’s success.
The Bees use mathematical and statistical modelling to identify undervalued players to buy, develop and sell on for large profits. The club profited around £67m from the sale of Neal Maupay, Said Benrahma and Ollie Watkins alone.
From a strategic perspective, data can be used to track and identify the characteristics of successful clubs. Data can also support pathway management and succession planning. Brentford’s data-driven approach to succession sees the club sell on its better players while remaining competitive. They won their first ever Premier League game 2-0. Sorry, Arsenal fans.
The widespread presence of data within football is a relatively new phenomenon that has accelerated thanks to technological developments. Embryonic forms of data analysis have taken place within football for decades. One of the pioneers of football analytics was accountant Charles Reep. Reep collected data in person at over 2,200 football matches throughout the 1950s and 1960s. Each game took 80 hours to analyse.
By the 1990s the use of video to analyse matches and training sessions was becoming increasingly common but poor technological capabilities held back the more analytically minded managers of the time.
At the 2002 World Cup, Argentina manager Marcelo Bielsa is said to have taken more than 2,000 DVDs of opponent matches out to the tournament. With no simple way of condensing the information into insights for coaches and players, Bielsa’s meticulous approach was stifled and the Argentines were sent packing in the group stage.
The acceleration came later in the 2000s when player-tracking technology was developed. The technology is now widely available in different software packages. It uses pitch side cameras and GPS-tracker vests to output thousands of data points for analysts to track and manipulate. This information is collected and used both by in-house analytics teams and third-party providers such as Wyscout and Opta.
Most decision-makers are from the football world, not the data world. The two remain largely distinct for now.
Over the past decade, performance analysts and recruitment departments have developed new metrics to look at this growing well of information. Many of these have since entered the mainstream. If you turn on your favoured sports news channel today you’re likely to hear reference to advanced metrics such as expected goals (xG) and expected threat (xT).
This may all paint a picture of an industry that has flung open its arms to the wonders of data, but to a large extent football remains a game of passion. The personnel in charge of strategy, be they a manager, chairman, or director of football are often big characters with big salaries and even bigger egos. Careers are short and insecure. Trust comes at a premium.
Most decision-makers are from the football world, not the data world. The two remain largely distinct for now. Data practitioners are working out how to capitalise on the advancement of analytics within the game to bridge that divide.
A team game
“Football is a dog-eat-dog world”, said Connor McGillick, lecturer in performance analysis at the University Campus of Football Business (UCFB). “Sometimes analysts are seen as more of a threat than a help.”
This dynamic will be familiar to data practitioners in any industry. Indeed, at a DataIQ roundtable held in June 2021 one attendee commented that: “Sometimes you can feel a bit like the Grim Reaper when you’re essentially having to demonstrate to users that a new product is superior to them and their knowledge.”
"Sometimes analysts are seen as more of a threat than a help." - Connor McGillick, lecturer in performance analysis at UCFB
This is a particular concern for football managers given their low-level of job security: the average Premier League manager lasts just 69.4 league games.
“The coaches have often been doing this for years, if not decades. When an analyst comes in its easy to perceive it as them telling the coaches what to do, which isn’t necessarily the case,” said McGillick.
In the corporate world this kind of oppositional dynamic can often be overcome by hitting KPIs, demonstrating value to a CFO or by reducing the amount of time spent on low-value tasks. Acceptance within the football world ultimately hinges on one factor: winning football matches.
This point was illustrated by Stats Perform, a sports data and analytics company that operates a football analytics course at Birkbeck University. When asked how data teams can encourage buy-in within the football world, team performance specialist at Stats Perform Andy Cooper said: “Demonstrating, using terminology that is understandable to key stakeholders, how the applied use of data and analytics can help you win matches…it is as simple as that.”
Football being football, passion and intuition will always play a role. McGillick mentioned that it can, for example, often be difficult to persuade a coach to reduce a player’s training load on fitness grounds if that player hadn’t performed well in a recent loss.
This is where data foundations, if established correctly, can come into action. “Once you’ve got the message across that you’re working with the coaches, and not against them, they’ll start to have trust in you,” said McGillick.
Trust in these foundations can help to ensure that blame for a loss doesn’t lie squarely on the analysts and that long term faith in the analytical process isn’t damaged off the back of a bad run of form. After all, it took Brentford nine years to get promoted to the Premier League on its data-driven model. “Brentford’s success didn’t happen overnight,” said Cooper. “Their promotion was the culmination of a long-term strategy.”
Patience is key, and in the high-pressure world of football patience can only be afforded when robust data foundations have been established.
Landing the message
Given the importance of productive engagement between analysts and the manager, players and coaches, a premium is put on the ability to effectively land messages and explain analysis and insight.
The football analytics course at Birkbeck includes a lecture dedicated to “managing conversations and negotiating with decision makers”. It outlines the importance of taking the time to understand the culture of a football club and the people that work there.
Most of the personnel that UCFB graduates will go on to support will not have a background in advanced analytics. Identifying relevant communication techniques is key.
McGillick said: “Are you the world’s best or worst analyst if you can’t get your message across? You could write your own code and algorithm to churn out a heat map within 10 seconds of the game, but that is useless if your audience can’t understand it.”
A heat map, by the way, is a data visualisation technique used to show where players spent the majority of their time on the pitch.
For football analysts, strategy dictates the statistics that are relevant.
Data visualisation is a vital tool both for analysts relaying insights to the club and for punditry teams relaying insights to fans watching at home.
“In most cases, people aren’t going to have the time or inclination to read a 2,000-word report, they need the information in a format where they can grasp the main takeaways without the wider methodology,” said Cooper.
“If someone is more receptive to taking in performance findings which are plotted on a pitch map, or on a chart, this should be a key factor in establishing how visual templates can be developed and coded, so they can be utilised regularly during the season.”
Managerial-playing style also plays a key role in the insights that should be delivered. The reality for analysts is that most managers are only interested in insights that support their particular brand of football.
The direct playing style of a manager like Sam Allardyce might lend itself to a focus on defensive statistics, crossing accuracy and heading success rates. This information would be largely irrelevant for Pep Guardiola, who would be more predisposed to statistics relating to line breaks and possession in the final third.
For football analysts, strategy dictates the statistics that are relevant.
At the University Campus of Football Business, McGillick prepares his students for this by conducting a role-playing exercise.
In the exercise Connor adopts the role of manager. He adopts a different footballing style and disposition to data for each session. The task for students is simple: land the message.
McGillick said: “In the early lessons, students will often just come to me and say ‘we’d sign this player because his crossing accuracy is 65%’.”
Without further context around how that information is relevant given the manager’s preferred playing style, or how it fits to a broader strategy outlined by a director of football, the response is likely to be “so what?”
Further context in footballing terms would mean outlining how, statistically, a player is a more suitable fit given the coach’s style of play when compared to a similar player in the squad or transfer market.
Landing the message is all about making that message relevant to the context of the team. The same is true for practitioners working within the context of the broader business culture.
In an ideal world, data would already be embedded within business strategy, but most organisations are a way off from reaching that point. DataIQ research shows that just 22.5% of organisations are currently at an Advanced level with their adoption of data and analytics.
The reality for most data practitioners is that they, like the UCFB students, need to work out how to “land the message” when relaying information to boards lacking in analytical expertise.
Football analysts have identified data visualisation and matching insights to strategy as springboards for the acceptance of data.
Like in the footballing world, the correct approach for practitioners will change depending on the culture of the business. Analysts need to have the skill to identify this culture and tailor their analysis to it.
Football is a simple game, made simpler with data
Harry Redknapp missed the point when he landed the blame for his side’s loss on his analysts. Analysts can’t ensure that a team wins, but they can ensure it has the best chance of success.
Analysts have been able to show that football, despite its reliance on passion and tendency for randomness, has constant and predictable patterns that can be mapped and exploited.
Gary Lineker once said: “Football is a simple game: 22 men chase a ball for 90 minutes and, in the end, the Germans always win.”
Analysts aren’t going to be able to ensure that England wins its next penalty shootout against die Mannschaft. They can however provide useful insights into what direction the German stopper is likely to dive.
In the recruitment department, analysts have been able to show that data can combine with traditional scouting methods to uncover bargains.
In general, analysts have been able to show that football, despite its reliance on passion and tendency for randomness, has constant and predictable patterns that can be mapped and exploited.
This gradual acceptance of data into football has relied on cultural factors as much as it has technological advancement.
Overcoming data-scepticism and poor data literacy. Establishing a data culture. Identifying relevant insights. Issues that analysts in the football world face, and continue to face, are familiar to analysts everywhere else.
The message from football analysts is to have patience. For every boardroom equivalent of a data-driven Pep Guardiola you’re going to have a traditionalist like Sam Allardyce. Identify the right approach for your environment. Demonstrate that you’re part of the team, not part of the opposition.