Section One – Pre-match Preparation

Pre-match analysis was first introduced to football by Charles Reep, a military accountant who would observe top-level football in his spare time. He would analyse the patterns of play in the matches by creating notes using codes and symbols, and later became known as the first ever analyst in professional football because of this (Pollard, 2002). In the 1950s, Brentford Football Club adopted Reep’s tactical analysis as they were in danger of relegation from the second division. They used his analysis on recorded passes, shots and players’ movements to improve their attacking sequences and movement on the pitch (Haywood, 2022). His technique, which emphasised regaining possession closer to their opponents’ goals and reducing the number of passes between regaining and shooting, not only helped Brentford but was later adopted by more clubs who integrated it into their pre-match training and preparation (Haywood, 2022)

As the game has continued to evolve, clubs have learned from the basics set out by Charles Reep and started to adopt new technologies such as Artificial Intelligence into their pre-match tactical plans. Top football sides are already using Artificial Intelligence to process their data and develop more effective pre-match tactics for formations, passing patterns, movement and analysing their opponents (Garratt-Stanley, 2025). By introducing artificial intelligence tools such as statistical learning, game theory and computer vision into their pre-match preparation, it allows coaches to combine their knowledge and experience with computer-based models, which help enhance how their team trains and perform on and off the pitch (Tuyls et al., 2021)

Liverpool Football Club became one of the first clubs in the world to adopt an AI system (Newsround and BBC, 2024) when they began using TacticAI, a technology developed by Google DeepMind to create pre-match predictions and simulations. TacticAI is an AI system developed to analyse set-piece routines, specifically corner kicks and provides recommendations on what type of corner routine to use and how to position players on the pitch to produce an effective corner (Garratt-Stanley, 2025). It uses statistical learning to process previous matches and help evaluate which tactics were successful in the past and dictate the probabilities of further success by continuing to use those tactics for corners, allowing coaches to analyse and plan their counter-tactics as well as train and inform players pre-match (Wang et al., 2024). Liverpool have seen significant improvements since introducing TacticAI into their pre-match analysis. Last season they averaged “seven goals from 230 corners in all competitions”, a statistic which means Liverpool scored from a corner set up once every 5 games, and furthermore their analysts and coaches’ favour “up to 90% of corner kick scenarios” provided by TacticAI (Garratt-Stanley, 2025)

Due to their success using TacticAI, Liverpool Football Club are continuing to explore how other AI systems and data models can offer a competitive edge for predicting

outcomes and patterns for other set pieces such as throw-ins or free kicks (Lichtenthaler, 2022). The club has already began experimenting with video tracking data models and are assessing the potential impact on their pre-match preparation and team performance if they were to implement it in the future (Lichtenthaler, 2022).

Another form of artificial intelligence utilised by coaches to process previous matchday footage and player tracking data for pre-match analysis is Computer Vision. It can provide insights for coaches on how their players perform during different game phases such as pressing, build up and transitions and help prepare the team for upcoming matches. (BBC Bitesize, 2025)

Computer vision analyses the video feed of individual players during certain events in previous matches, focusing closely on their body orientation and movement. This creates a 3-dimensional skeleton which can be analysed to show a player’s position on the pitch in relation to their teammates, the ball and the goal (Tuyls et al., 2021). Clubs can use statistical learning alongside computer vision to help build predictive models which use and learn from the video feed on individual players’ behaviours and decisions while on the pitch, such as their passes and tackles, especially when leading up to a goal. It can evaluate what tactics would enhance that player’s style of play and the impact that could have on the pitch, something which humans cannot do alone. (Tuyls et al., 2021)

In a recent interview with Ukrainian and Glasgow City footballer Nicole Kozlova, she explained how Artificial Intelligence is currently helping to improve her individual game and track her progress throughout the season (BBC Bitesize, 2025). Nicole Kozlova uses Earpiece, an AI technology developed by Swedish company Twelve football that processes match data and studies player performance patterns, producing reports and rankings of each player’s overall contribution to the teams’ match success. She discusses how it has helped turn her complex performance data into insights, which she has been using to understand her strengths as well as address areas where she could improve. Alongside her coaches, Nicole has been able to use detailed metrics on her positioning, speed and tactical shape to help focus her pre-match training, make more informed decisions going into matches and feel more prepared for the games ahead. (BBC Bitesize, 2025)

As Artificial Intelligence continues to evolve and become more accessible to players at all levels, the data collected will become richer, resulting in more comprehensive analysis and advanced models (BBC Bitesize, 2025). Due to the limitations of current performance models, Artificial intelligence can only currently analyse players’ actual performance against their expected performance, highlighting any strengths or weaknesses but currently overlooks physiological and psychological attributes (Liu and Liu, 2025).

As AI models continues to develop and use bigger, more representative datasets, researchers believe they will become smarter and produce more personalised tactics for each player on a team. This will amplify how analytical insights are used to create individual training plans, tactical drills and more defined match plans. (Liu and Liu, 2025).

Besides producing data insights on their own team, analysts can utilise statistical learning AI tools during a pre-match period to analyse their opponents and produce actionable insights. Just like when they produce statistics on their own players, they can also make data-driven conclusions based on the opposition’s historical game characteristics and unique behaviours, for example, how they attack or defend the ball (Tuyls et al., 2021). Formerly, analysts could only take into consideration the on-ball actions of their opponents, such as shots and passes; however, the introduction of artificial intelligence has introduced reinforcement and deep learning models, which can examine off-ball movements and situational decisions even when it doesn’t relate directly to a goal (Tuyls et al., 2021). Using these models to help analyse their opponents enables coaches to make more informed and strategic decisions when selecting their own team and starting eleven formations (Nouraie et al., 2023). Deep neural network is an AI approach currently used by analysts to rate each players suitability in different positions. The model assigns each player a suitability score based on performance data such as a player’s speed, attacking ability and defending strength. From these data-driven insights, the system can generate multiple strong lineups with different player variations and formations allowing the coaches to use their professional judgement to make the final decision (Nouraie et al., 2023)

As artificial intelligence continues to develop more advanced models, football teams are expected to rely less on analysis of past events and use predictive modelling more in their pre-match preparation (BBC Bitesize, 2025) as well as adaptive models which analyse unobserved on-pitch factors such as unpredictable play, player chemistry and match context. (Garratt-Stanley, 2025). In the future analysts hope that predictive modelling could simulate more counterfactual scenarios, allowing them to model how specific opposing teams and individuals react to different tactical approaches. This will allow teams to refine their own techniques and adjust their pre-game tactics based on their opposition. (Tuyls et al., 2021)

Section Two – Mid-match Analysis

Prior to the introduction of Artificial Intelligence and digital platforms for performance analysis, football clubs manually analysed match play using pen, paper and a code sheet. They would focus on the players involved, the type of actions, the outcomes of their actions and the time it occurred during the match. Some analysts even used pitch diagrams allowing them to visualise which third each event occurred (Carling and Court, 2013). This allowed coaches and analysts to highlight any areas of the pitch being exploited by their opponents or areas where possession was repeatedly lost. It provided coaches with real time insights and allowed them to make tactical changes during the match and strengthen a team’s style of play.

However, as this process used human observation and manual notetaking alone, analysts couldn’t keep up with the speed of play and therefore couldn’t maintain a high level of accuracy when collecting the data (Carling and Court, 2013). When artificial intelligence was introduced to analytic processes in football clubs, it completely changed how teams gather and analyse their tracking, event and biometric data during matches. It not only quickened the process of analysing and identifying data patterns but was able to provide more in-depth insights which were missed by the human eye. (S. Dedgaonkar et al., 2025).

Although real-time analysis is still developing, companies such as Catapult have developed AI technologies solutions and wearable devices to help clubs at all levels across the world further analyse and manage their players and overall teams’ data, using it to create meaningful outcomes on and off the pitch (CATAPULT, 2025). Using Catapults AI-powered tools such as MatchTracker or Focus, teams like FC Köln, Frankfurt and Luton Town have been able to use real time analysis to provide coaches and analysts with immediate insights on tactical issues such as defensive gaps starting to emerge due to the opponent’s movement or strategies. This enables analysts to make immediate tactical adjustments by altering formation or briefing their players with new instructions (CATAPULT, 2025).

Beyond the match tactics, clubs can combine the GPS data gathered from player with the multi-angle live video analysis to collect data and build a wider picture on a player’s performance as well as highlight any issues such as fatigue, players struggling with the game intensity, or players not performing at their usual standard (CATAPULT, 2024). This prepares coaches to make smarter substitute decisions as they can not only identify which player needs to be subbed off but also identify which substitute is ideal to come on based on how their style of play will exploit the opposition, how they fit into the team’s current tactical approach and how well they work alongside other players on the pitch (CATAPULT, 2025)

Catapults real-time analysis technologies have provided both coaches, analysts and players with accurate live match data which can strengthen their decision-making and tactical understanding, allowing them to make more data-driven real-time decisions for tactical adjustments and individual player insights (CATAPULT, 2025) As AI analysis continues to evolve, it is anticipated that it will allow systems to create more sophisticated strategic in-game recommendations for substitutes, formations and tactics as well as incorporate predictive modelling to help forecast match scenarios for specific opponents based on historical patterns and trends (CATAPULT, 2025)