FootyChatter.com

The data genie is well and truly out of the bottle. The presence of data-driven consultancies, the rise of public data websites and the growth of its use in the media (cough, cough) highlight how integral statistics have become in the way we view and analyse the game. Knowledge sharing has been a crucial catalyst for the creation and development of many metrics and statistical models. However, the curtain every football fan wants to peek behind is the use of analytics within professional clubs. Understandably, these in-house data departments will maintain a high degree of confidentiality to maintain a competitive advantage over their rivals, but what does this landscape look like? Advertisement Having a ‘Moneyball’ approach remains the in-vogue term used to explain the data-led methods adopted by clubs such as Brentford, Brighton & Hove Albion and Liverpool. But any club that has had success with data knows it is not as simple as Oakland As baseball general manager Brad Pitt clicking his fingers and pointing at numbers guy Jonah Hill in the Moneyball movie. GO DEEPER Talking to Michael Lewis on the 20th anniversary of 'Moneyball' Analytics departments must focus their energy wisely. The simplicity of the message delivered by Dr Ian Graham, Liverpool’s director of research until 2023, was notable during Stats Bomb’s 2021 conference, declaring that “player recruitment and retention is the most important work — by a factor of 10”. Buy-in is also crucial. You might have the best statistical models and machine learning algorithms in the world, but aligning and integrating such work with key decision-makers is where the impact of analytics can be maximised at club level. Brighton’s owner/chairman Tony Bloom ensures club staff use data provided by his company Starlizard, which has helped turn lesser-known players into Premier League stars, including Kaoru Mitoma, Moises Caicedo, Alexis Mac Allister and Julio Enciso. Similarly, his Brentford counterpart, Matthew Benham, is the founder of statistical research company Smartodds — primarily designed for professional gamblers but crucial in helping Thomas Frank’s side find value in the player recruitment market. GO DEEPER Premier League: How to find the edge in data analytics - examining trends and what is to come If a club’s owners or sporting directors are less data-minded, a communication gap can often develop between analysts and the powers that be. Recently, companies such as Soccerment and Sentient Sports have used Generative AI to help bridge that by condensing complex statistical analysis into simple football language — think Chat GPT for player scouting — but challenges can still exist. Advertisement “Best-practice analytics is not creating the most ‘complex’ model or algorithm, it is analyses that are trusted and adopted by decision-makers that ultimately have an impact on their processes, ” says Dan Pelchen, founder of analytics company Traits Insights.  “Trust and understanding can empower more experts to use data daily, helping avoid biases and mitigating risk. ” There has been a lot of commentary on the growing world of football analytics in recent years, but — aside from a recent research paper published in April  — there has rarely been an objective, statistically-led depiction of the data ecosystems across the leagues. Traits Insights collected information on approximately 500 staff members from more than 90 clubs in the top four divisions of English football — categorised into data analysts (a catch-all statistically-based role), recruitment analysts, first-team analysts (for example, performance/technical/opposition analysis), and overarching heads of analysis — to better understand the challenges facing clubs to build “best-practice” analytics processes, which The Athletic can now exclusively share. Outlining best practices is one thing, but implementing them is another.   Setting up a coherent, self-sustaining analytics department requires a significant investment from board level, and making a business case for its long-term utility can be challenging. Traits Insights’ analysis showed the ‘traditional’ top six Premier League clubs (Manchester City, Arsenal, Liverpool, Manchester United, Tottenham Hotspur and Chelsea) have approximately 14 analysis-based staff members on average — which is double the average among clubs in the bottom half of the same division. Unsurprisingly, those numbers dwindle as you descend into the second-tier Championship, League One and League Two, the fourth level of the English game. For some, limited staff capacity can mean some analysts will often be asked to have a Jack-of-all-trades role — data engineer (collecting and managing large datasets), data analyst (interpreting the information and presenting to colleagues), and data scientist (building statistical models to provide insight) all rolled into one, for example. Advertisement “Data analysis is still a relatively new department within clubs, ” said one data scientist at a Premier League club, speaking anonymously to protect relationships.  “People from different backgrounds are often enthusiastic about introducing data into their workflows, but a club typically begins by dipping their toe with a small investment — for example, one junior data role and one data provider subscription. “Those who allocate this initial investment typically don’t come from a data background and understandably don’t know the different skill sets required between data analysts, scientists and engineers. When the first junior hire begins work, they can quickly become overrun with demands that cannot be met without the structure in place to produce quick and valuable insights. “This can quickly lead to frustration on both sides. It is no coincidence that the clubs with the most successful data departments have people at the very top of the club who have come from quantitive backgrounds. ”  This is a sentiment shared among other staff members throughout the English football pyramid. “A good data engineer is crucial for productivity and enabling other roles to succeed. It is a role that is often the hardest to fill and is frequently overlooked because it’s not flashy or particularly visible to day-to-day practitioners, ” said a data scientist at an EFL club, also speaking anonymously to protect relationships. “Data scientists are predominantly responsible for model generation and delivery of these insights, and data analysts are the most people-facing — responsible for the development of tools and delivering clear visualisations and presentations. “Each of these three roles have specific responsibilities and skills that are essential for fulfilling their tasks. Without one, the others would face increasing challenges. If a team member leaves, our skill sets are all deemed as the same when, in reality, they are very different disciplines. ” The desire to use analytics has grown exponentially in recent years but it is important to note that specialised expertise is required to manage and interpret data, build statistical models, and create interfaces (for example, dashboards and visualisations) that allow the analysis to be understood by others at the club. Advertisement This requires specific education and technical training to create such advanced models (for example, neural networks and machine learning algorithms) — stemming from backgrounds in hard sciences such as data science, economics, computer science, engineering and mathematics. Many staff members will have qualifications in sport-and-exercise science, performance analysis or similar — which requires a lot of technical training — but the statistical qualifications among staff are scarce within football. Traits Insights’ analysis found that 46 per cent of data analysts in the sample had a technical statistical education, with approximately five per cent of the remaining analysis staff having such a background. The limited number of support staff with expertise in data and statistical insight can put strain on specific individuals when internally building technical systems, with the core goal being that all team members can develop such systems and extract insight at all stages along the “production line” of a club’s workflow — from junior data analysts up to senior staff members, including sporting directors. Approximately 75 per cent of the 20 Premier League clubs have specialised data analysts, with 50 per cent having multiple. By contrast, only half of the Championship’s 24 clubs have a dedicated data analyst, which similarly dwindles when reaching the 24 sides in both League One (25 per cent) and League Two (less than 10 per cent). However cliched you might think it is, football is a results-based business. Sporting directors will often have a long-term view of the club, but that may not always be as stable when going down the three tiers of the Football League.   For support staff, being afforded the time to build statistical models and generate tangible insight can be easier said than done. At clubs with a higher turnover of coaching staff, these workflows and systems can naturally break down if a new manager or head coach has a different method of operating.  If this does occur, it can stifle the progression of an analysis department.   Similarly, analysts working at lower-division teams may also want to work at other clubs and climb up through the leagues, making staff turnover more likely further down the pyramid. This was reflected in Traits Insights’ analysis, which showed analysts at top-six Premier League sides had an average tenure of 4. 7 years, compared with 2. 5 years or less in League One and League Two. Broken down by role, a club’s head of analysis is most commonly in their role for the longest period. Notably, data analysts have been in their positions with the team concerned for 2. 5 years by comparison, which speaks to the infancy and potential transcience of the job compared with other support staff at a club. “These results are not surprising when you think how new data is within football, but also how valued these skill sets are within other industries, ” said the Premier League data scientist quoted earlier in this article.   “Within clubs, a data analyst role often begins as a junior role, but the skill set required at a club is more in line with a senior role within other industries. If you can meet all of the requirements to work at a club, you will be in huge demand outside of football, so it is understandable why people may move on quicker than other roles. ” Advertisement When building an analytics department, there is neither a single path to success nor an established method for clubs to develop their infrastructure. Best-practice is difficult to come by without stability, strong technical skills, and investment — and the complexity of such work means the Moneyball method is often idealised beyond reality. Naturally, clubs with bigger budgets can invest more in their analysis departments but work that influences player recruitment, player retention and talent development is where data analysis can find its best outcomes, and establishing clear lines of communication between departments is crucial. Whether outsourcing work to third-party consultancies or developing your own data team within the club, ample opportunities remain to gain a competitive advantage — at any level of the game. (Top photo: Nick Potts/PA Images via Getty Images) Get all-access to exclusive stories. Subscribe to The Athletic for in-depth coverage of your favorite players, teams, leagues and clubs. Try a week on us. Mark Carey is a Data Analyst for The Athletic. With his background in research and analytics, he will look to provide data-driven insight across the football world. Follow Mark on Twitter @Mark Carey93