More and more football clubs are relying on artificial intelligence when looking for new players. This can create scores for each individual player – sometimes with surprising results.
How good is the technology?
Every club wishes to finally sign a new star player who not only costs little but also is a direct reinforcement. The goal is to achieve the highest possible return with the lowest possible risk. Football is becoming more and more like mathematics. Scouting is crucial for low-risk when signing new players. The better a player is analyzed, the better he can be assessed.
This is where artificial intelligence comes into play. To put it simply, the AI evaluates data and then assesses players. That is precisely the job of scouts. A human scout sets specific criteria before observing a player. The AI was also taught particular rules at the beginning. For example, a player should win a duel rather than lose it.
Data about each player
Since professional players have been constantly observed and evaluated during training and games for years, the AI already has a lot of data. Based on this, the tool can then assess players. The software company SCOUTASTIC has been working with Bundesliga clubs for some time.
Christian Rümke from SCOUTASTIC explains that the artificial intelligence system uses pure player data and can also evaluate texts: “Many smart scouts drive around and write reports every weekend,” says Rümke. It is challenging to keep track of things and know what was written in the report three years later.
Better than human scouts?
Players also developed artificial intelligence to help football clubs find the right players. Jan Wendt, one of the founders of Praier, even claims that the AI is correct in its player assessments more often than humans: “If you choose ten players with us , then we will be correct 8.5 times, which is much better as if you were doing it with human means.” Wendt means that the AI would be right at least eight times if you chose ten players.
Many scouts would undoubtedly like such a quota. According to Wendt, it depends on the individual case; some clubs scout well, others less. He emphasizes that it’s not about categorizing scouting into right and wrong. The tool supports the scouts. “But if you consistently ensure that players have to go through our filter and only take players who we see suitable and who will make the club better, the scouting rate will improve,” says Wendt.
AI score for each player
Praier’s tool assigns each player a so-called AI score. This is calculated from the AI-collected data about the respective player. “A striker is evaluated using different criteria than a central defender,” Wendt explains. Overall, the player’s performance is measured with over 200 parameters, and the player’s influence on the team’s performance would also be included in the evaluation, explain the Player boss. The score can then be used as a comparison parameter for players worldwide.
In an exemplary scouting scenario, Wendt’s AI shows tha tStuttgart striker Deniz Undav is rated better than Erling Haaland. Wendt explains that, according to the AI, Undav is the slightly more complete striker, as he achieves better values, especially when working against the ball.
The limits of artificial intelligence
Even if the AI concludes that Undav is better than Haaland, this cannot be wholly proven. Both play in different clubs and leagues. For example, the coach or the player’s environment could influence performance.
This means that if Undav moves to another club and another coach, he could no longer perform as well. “The AI simply reaches its limits. And then, to say that the player fits exactly into this system, to this coach, and into this environment where he feels comfortable, there are so many factors the AI can certainly already support. “But at the moment, it doesn’t bring everything together,” explains Rümke from SCOUTASTIC.
Support for the human scout
Despite everything, artificial intelligence is changing scouting in football. Having a decision-making aid that is wholly data-based and does not involve personal preferences is undoubtedly an advantage. It also relieves clubs and scouts of a lot of work and can help scout players in more minor leagues abroad who might not otherwise be on the club’s list.
Both Player and SCOUTASTIC emphasise that AI does not replace any scout in the world. “We see ourselves more as support and help in making informed decisions. We have no claim to make player transfers,” said Rümke. In the end, a human still has to make the decision, and it’s not just about data and mathematical connections, but instead on a human feeling and a personal assessment of whether the player fits into the team.
Data science has made its way into the world of football. How are teams and businesses utilizing it?
Artificial intelligence is becoming prevalent across all sectors. As the World Cup approaches, one might ponder if AI also plays a role in soccer. Whenever there is data, machine learning models can be employed: football generates vast amounts of data, and there exist a century’s worth of statistics, audio, video, news, and social media posts. Over the past few years, companies specializing in AI for soccer have emerged, and football teams are hiring analysts and data scientists. Why are they doing this? What are the applications? This article delves into these topics.
The realm of predicting the future
As long as there have been sports, there has also been betting. In ancient times, the Romans would bet on chariot races and even turned to various methods to sway the outcome in their favor. Today, relying on magicians is a thing of the past , although in 2010, an octopus was used to predict the outcomes of the World Cup (one of my favorite moments is when the octopus predicted the final match). Nevertheless, sports betting alone generated a revenue of 4.33 billion in 2021 (and this figure For betting platforms, estimating the odds is crucial to avoid losses due to users’ winnings. Betting agencies make use of sophisticated algorithms to set these odds.
Predicting stunning not only intrigues bettors but also betting agencies. The challenge of foreseeing a team’s win or loss has fascinated mathematicians and statisticians. An article published in Plos One employed the double Poisson model to accurately predict six out of the eight teams in the quarterfinals, as well as Italy’s victory over England:
First developed in 1982, the double Poisson model, which assumes that the goals scored by each team are Poisson-distributed with a mean that depends on the offensive and defensive strengths, continues to be a popular choice for predicting football scores, despite the numerous newer methods that have been developed. […] These predictions won the Royal Statistical Society’s prediction competition, showing that even this simple model can yield high-quality results.
However, this was a retrospective study. The same authors are predicting Belgium’s victory in the 2022 World Cup. There are also other forecasts, each with differing predictions: Lloyd, based on the insurable value of each player (cumulative value), predicted that England would win the cup (a method that was successful in 2014 and 2018). Opta Analyst, using AI, predicted that Brazil would emerge victorious (with 16.3% odds, and 13% for Argentina). Electronic Arts also performed simulations with algorithms to predict the cup winner and placed its bet on Argentina.
The robotic scout that never misses talent
In 2003, the book Moneyball gained popularity by detailing how Billy Beane, the manager of the Oakland Athletics baseball team, utilized statistics to construct the team. Beane was able to demonstrate that skillful statistical analysis could help him identify players more effectively than traditional scouts.
Identifying talent is no easy feat: in the summer of 2022 alone, 4.4 billion was spent in Europe on player transfers (this year, the most expensive transfer was Antony to Manchester United for 85 million, though it didn’t make the top 10 most expensive transfers ever). Moreover, there are numerous instances of players costing tens of millions but failing to live up to expectations.
While this approach is common in basketball today, it is not as straightforward in soccer. In baseball, statistics have been collected and utilized for many years, and there are fewer variables to analyze (for instance, only one team attempts to score points at a time). In soccer, several models have focused solely on the number of goals or goal-scoring actions, overlooking the contributions of players who may not have had possession of the ball at that specific moment.
Despite these challenges, many teams now rely not only on scouts but also on companies specializing in algorithms. Additionally, several teams have hired analysts and data scientists. One particularly fascinating example is Brentford, which has developed its own algorithm for identifying undervalued players with high potential (acquiring them at a low cost and selling them at a significant profit).
On the other hand, owner Bentham has made millions with his company Smartodds, where, with a team of statisticians, he calculated match outcomes more accurately than bookmakers.
Nevertheless, it’s not just about identifying the most underrated player; it’s also about identifying the best player for the team from among thousands of potential candidates. According to Brentford’s owner, the models must also account for player development.
Several companies have specialized in various aspects of this process. Some collect player data, others analyze this data and suggest potential acquisitions, and yet others recommend suitable salaries. For instance, SciSports employs its algorithm to track over half a million players for potential acquiring teams .
It’s all about strategy
Many teams have found that spending large amounts of money to acquire top players does not guarantee success. Soccer is a team sport that requires players to collaborate. Currently, various researchers and companies are concentrating on improving teams’ strategies and tactics.
The concept is not new. Back in 1950, Charles Reep examined games and concluded that most goals were scored from fewer than three passes, indicating the importance of passing the ball as far forward as possible. Over the years, more advanced approaches have been developed , such as the one created through the collaboration between the University of Lisboa and Barcelona. The authors used positional data from players to determine the hypothetical threat to the opposing defense.
During a game, there are numerous passes. For a team seeking to analyze strategy in preparation against another team, it would be necessary to study videos and calculate statistics. Currently, specialized companies analyze recorded footage using computer vision algorithms and then sell the results.
However, these images come with a high price tag. To address this, researchers have focused on predicting the movement of players when they are not in the frame. Recently, DeepMind and Liverpool FC collaborated on a similar approach, and a paper was recently published . The authors used a combination of statistical learning, video understanding, and game theory:
“We illustrate football, in particular, is a useful microcosm for studying AI research, offering benefits in the longer-term to decision-makers in sports in the form of an automated video-assistant coach (AVAC) system”
The researchers analyzed over 12,000 penalty kicks taken by players in Europe, categorizing them based on shooting technique and scoring success. The analysis revealed that midfielders employed a more balanced approach, being more inclined to shoot at the left corner and use their dominant side.
Moreover, stopping a penalty kick is a challenging task for a goalkeeper, who only has a split second to decide whether and where to dive. Therefore, goalkeepers now receive statistics on the typical penalty kick shooting patterns of players. There are also studies dedicated to free kicks, focusing on how to position the defensive wall to provide the best view for the goalkeeper.
Other studies are centered on analyzing the optimal timing for a player to shoot, pass, retain possession, make a run toward the goal, and so forth. Some of these studies leverage approaches derived from the same simulation algorithms used for autonomous machines. An example is StatsDNA, which was acquired by Arsenal and follows a similar approach, relying on telemetry and Markov chain-based algorithms.
It may appear that these studies have not yet had a significant impact and are still predominantly at the research stage. However, in recent years, the shooting distance for players has been reduced. Data analytics has conclusively calculated the probabilities, showing that the farther the shooting distance, the lower the likelihood of scoring. Supported by data and analytics, teams are encouraging players to take shots from closer range and avoid long crosses into the opponent’s area.
Additionally, determining when to substitute players during a game is no easy decision (consider the controversy surrounding Cristiano Ronaldo’s substitution). “There is no favoritism as AI removes the emotion from decision-making,” states Martin McCarthy, who collaborates with IBM Watson on pre- and post-match analysis, player substitution, and other strategies.
Only the ball remains the same
Indeed, artificial intelligence is anticipated to transform every aspect of soccer. Numerous startups are researching the optimal diet for players and training methods to prevent muscle injuries. When a player sustains an injury, there are studies on predicting recovery time and the best recovery strategies.
Other applications include utilizing algorithms to determine ticket prices based on factors such as the significance of the game, timing, and more. Moreover, during major events, the entry process into the stadium often results in queues and errors, prompting companies to explore the use of facial recognition for ticketing systems.
Furthermore, the Bundesliga has teamed up with AWS to enhance insights during broadcasts, produce highlights, and automatically tag players.
Tests have been conducted with robotic cameras that autonomously track ball movements (particularly during COVID-19). While this has not always been successful, in one instance, the algorithm mistook a linesman’s bald head for the ball, leading to complaints from fans who missed their team’s goal as a result.
A study conducted by the NBA revealed that referees make errors in 8.2% of instances, and 1.49% of calls made in the final minutes of the game are incorrect, potentially impacting the game’s outcome. The realm of soccer has seen its fair share of controversies , prompting the implementation of Video Assistant Referee (VAR) and Goal-line technology. Research is ongoing on AI referees to minimize contentious decisions, such as Diego Maradona’s infamous “Hand of God” goal in the 1986 World Cup.
Furthermore, there might be changes in sports journalism as advancements in language modeling enable coherent text generation. This could benefit lesser-covered minor leagues, as demonstrated by NDC, the Dutch local media, which utilized algorithms to produce match reports for 60,000 matches in a year.
Parting reflections
Football leagues generate vast amounts of data, encompassing videos, countless posts, newspaper articles, and extensive discussions. Many teams now incorporate sensors in training to gather additional data. Given the rise of artificial intelligence, it was inevitable that sports would be impacted.
However, sports often resist altering rules and adopting new technologies, particularly in official matches. The introduction of VAR and goal-line technology sparked substantial debate. Nevertheless, soccer is a multibillion-dollar industry, prompting teams to turn to data science for improved player signings to avoid costly mistakes.
The entire interconnected ecosystem of sports will also undergo changes, from tactics and coaching to injury prognosis and ticket sales, and even sports journalism.
Football is arguably one of the most challenging team sports to analyze due to its numerous players with diverse roles, infrequent key events, and minimal scoring, as highlighted in a DeepMind article.
On the other hand, soccer presents unique challenges compared to other sports, with additional external factors to consider. The anticipated revolution will take time. For instance, algorithms may suggest that players like Lionel Messi are overpaid relative to their value, yet their advertising returns are difficult to quantify. The controversies stemming from human errors garner significant attention, as they are integral to the sport’s appeal.
Analyzing game footage is a fundamental activity for football teams but is also labor-intensive and prone to human error. A groundbreaking solution developed by the computer science department at Brigham Young University has revolutionized the planning and execution of football game-tape analysis. This innovative The approach utilizes machine learning, neural networks, and computer vision to save significant time in tagging players, tracking their movements, and identifying formations accurately.
Football teams rely heavily on strategic planning, with the analysis of game footage forming the cornerstone of devising winning strategies. The NFL’s “Game Operations Manual” prohibits the use of video recording devices during games, highlighting the significance of the information-gathering process. As a result, scouts resort to observing coaches and their assistants from the stands in an attempt to gather insights into their strategies.
The strategic nature of football, characterized by its stop-and-start dynamics and intricate formations, lends itself well to analysis, distinguishing it from the more fluid sport of soccer.
Continuous preparation
In football, coaches and players have numerous occasions to execute diverse strategies and formulate specific tactics for each play, be it on offense or defense.
If you have thoroughly completed your homework and the footage deities have provided unique insights, you have an opportunity to use them to outsmart the other side.
Mark Lillibridge, an experienced football player and NFL scout, discusses how his team discovered a tell from a fearsome fullback on an opposing team by repeatedly reviewing tapes. The fullback had the habit of “ever so slightly cran(ing) his neck to get a view of the player he was about to block.”
Additionally, there is an AI chatbot able to summarize any PDF and address related questions.
Such revelations can make a significant impact, leading to game disappointing and enhancing pursuits. Lillibridge states, “There’s nothing better than being 90% sure what play was about to be run.”
This type of insight explains why players still begin their preparations for the next game by reviewing footage of the previous game. Teams often allow players to download footage onto their iPads from almost anywhere.
However, having footage alone does not guarantee success for a player. The actual challenging work occurs in the departments responsible for creating game tapes.
In these departments, team personnel must accurately identify players from opposing teams, their positions, movements, as well as offensive and defensive formations.
They must then make astute observations on everything from overall strategies being employed by the opposition coach to detailed player movements and tendencies, in order to devise countermeasures.
This level of analysis demands a substantial number of hours, considering that there are 55 players on each team’s roster and 32 teams in the league. Additionally, historical tape reviews require a significant amount of time.
Furthermore, getting the analysis right is a difficult task, particularly for humans. offline, it’s a straightforward task for machine learning.
When the engineering team at BYU began analyzing their college’s football tapes, they quickly realized a major issue regarding inconsistent camera angles.
At the college level, game camera placement tends to be inconsistent, and not all players are always visible from a single camera angle. Furthermore, the quarterback and defensive players closest to the line of scrimmage are often obstructed.
To address the issue, the BYU team decided to develop a proof of concept using the Madden 2020 NFL video game. This solution provided the control and consistency their algorithm needed.
The most useful camera view turned out to be an overhead, bird’s-eye-view, allowing almost all players to be seen. Coupled with end-zone views, every player could consistently be covered.
The solution worked, and the BYU team’s algorithm successfully identified and labeled 1,000 images and videos from the game.
The researchers reported greater than 90% accuracy on both player detection and labeling, and 84.8% accuracy on formation identification. Accuracy in identifying formations reached 99.5% when excluding the more complex I formation, which had several player views obstructed.
So, what does all this success mean for the immediate future of football analysis? According to Lee, “Well, you could get access to the broadcast video of NFL games, filter out commercials, graphs that they put on the screen, but it’s not as efficient. It’s a lot more work.”
“You don’t really need to have a bird’s eye view. You just need to be up high, so we can see the whole field. And if you cannot see from the overhead camera, you should be able to see from the end zone . Once you get that all synchronized, you’re in business,” Lee added.
The NFL has long made every NFL game in the season available in the All-22 format, which is a camera perched high up at the 50-yard line, providing a view of every player on the field.
Even enthusiastic fans can access this data for $75 a year.
NCAA college football conferences began doing the same thing last year, though the initiative is still in its early stages.
In essence, what BYU’s algorithm achieved with Madden 2020 can easily be applied to future developments in football analysis.
AI system will completely change your experience at sporting events
It’s football season officially, which means you might be heading to an NFL game soon. If you are, the lengthy, frustrating, and not always accurate metal detector process may soon become a thing of the past, thanks to Evolv body scanners.
Have you ever attended a sporting event and spent what felt like an eternity just trying to enter? Security technologies can slow down lines significantly, and they’re not always effective – your necklace, keys, and belt may trigger the metal detector, while weapons can slip through. At Cleveland’s FirstEnergy Stadium, it turns out a lot of football fans wear steel-toed boots.
“Everyone wearing these boots was setting off the metal detectors when they were coming in,” says Brandon Covert, the vice president of information technology for the Cleveland Browns.
The team has managed to resolve this issue with artificial intelligence, after implementing security screening technology from Evolv.
“I would say that through machine learning, at this point, I don’t believe that’s been a problem this season,” Covert states.
You may not be familiar with Evolv, but its technology is being used in stadiums across the nation. In fact, the company has screened over 350 million individuals since its launch in 2013, second only to the US Department of Homeland Security’s Transportation Security Administration. Evolv screens nearly 750,000 people daily and as many as 1.25 million on weekends.
Evolv was established in 2013 after both co-founders had personal connections to those who were put at risk due to inadequate security in large gatherings.
Co-founder Anil Chitkara had a close friend and college roommate who was on the 101st floor of the North Tower on 9/11. Then 12 years later, he was driving home from the Boston Marathon, where he had watched his wife cross the finish line with his kids, when he found out that an explosive had detonated. Co-founder Mike Ellenbogen also knew people directly affected by the Boston Marathon bombing.
The team developed the touchless screening system, Evolv Express, which has a similar build to a metal detector but can identify threats much more quickly. The scanners can screen up to 3,600 people per hour, 10 times faster than traditional metal detectors.
The body scanners utilize a combination of advanced technologies including sensors, machine learning, cloud analytics, and centralized data stores, which enable the scanner to detect potential threats such as knives, guns, and explosives. According to Evolv, there is usually around a 5 % alarm rate in a sports stadium.
“Instead of simply looking for a binary yes/no for metal, they’re searching for specific shapes, sizes, and density of things that could potentially be threatening and could potentially cause mass harm,” stated John Baier, the vice president of sports at Evolv Technology. This, he said, “allows patrons to keep their cell phones, earbuds, keys, etc on them and walk through with the normal pace of life.”
The technology also enables individuals to walk through without needing to remove their items in their pockets or bags without compromising accuracy.
Evolv reports that, since January 2022, its system has detected and prevented over 30,000 guns and 27,000 knives, not including law enforcement, from entering its customers’ venues.
By eliminating the need for time-consuming and burdensome bag checks, Evolv argues that people can enter the venue much faster, and staff can redirect their attention where it’s more needed, creating a more positive fan experience.
Another benefit is that the more precise readings can inform the staff exactly where the danger is, targeting a very specific area and avoiding uncomfortable, full-body pat-downs.
“Through a secondary screening, we also provide a targeted region where the person needs to search, so it’s not a ‘Please step aside so we can wand your entire body,’” Baier said. “It’s ‘Sir, Madam, what’s in your right pocket or your left ankle?’”
Also, what does cybersecurity mean? And why is it important?
The Cleveland Browns implemented the technology in August at the FirstEnergy Stadium. Since then, the scanners have been used for three events – a Machine Gun Kelly concert and two pre-season games. So far, the team is pleased with the implementation’s results.
“Our stadium operations groups have really fallen in love with these, both from a speed and service perspective. Getting fans in the stadium on time is a big challenge,” Covert says. “And when you’ve got 50,000 people that get up 15 minutes before kickoff, it creates a bottleneck. So they love them, because we can clear gates extremely quick, quicker than we’ve ever been able to before.”
The Browns have installed a total of 12 Evolv Express units at their two south gates, which made it possible to replace 100 metal detectors and reduce the number of staffers in half, from 150 to approximately 75.
The scanners utilize machine learning, allowing them to adjust to best suit the unique circumstances of each stadium and its surroundings. Evolv has screened 114,000 fans at the stadium so far, with only a 4% alarm rate.
The Browns have also made use of the analytics dashboard that accompanies the Evolv system. The dashboard provides insights into the performance of the security screening system, visitor flow, location-specific performance, and more.
“All of these machines are linked to a central dashboard. This allows us to monitor the entrance of people in real time, assess the popularity of the units, identify potential threats, and optimize operational efficiencies,” explained Covert.
In addition to the Cleveland Browns, other NFL teams, such as the Atlanta Falcons, LA Rams, New England Patriots, and the Pittsburgh Steelers, have also adopted the technology.
Evolv’s technology extends beyond football stadiums. It has been deployed in casinos, healthcare facilities, places of worship, and numerous educational institutions. Investing in these scanners is an investment in enhancing the overall experience for fans, which is particularly valuable to the sports industry.
“Safety is a crucial service that we offer our fans on gameday, so we’re always committed to enhancing safety measures,” Covert remarked. “The return on investment in this model isn’t just financial; it’s also about improving the fan experience .”
The realm of professional football is extremely competitive, with teams employing every possible means to secure an advantage. Recently, Artificial Intelligence (AI) has surfaced as a groundbreaking technology in the high-pressure area of player scouting and recruitment. AI systems have the ability to sift through extensive datasets and video clips to identify promising players much more effectively than human scouts. However, the application of AI in football scouting is still a topic of debate. Could it truly transform recruitment and offer clubs a significant competitive edge?
Don’t dismiss this as a passing trend; back in 2015, I attended a presentation at IBM regarding the capabilities of Watson, their leading AI. One developer I conversed with mentioned something that has stayed with me: rather than viewing AI as simply artificial, we should consider it as IA or “intelligence augmented”—a collection of tools and capabilities designed to enhance rather than replace our human abilities. Its impact on sports has yet to be clearly visible.
To grasp why AI scouting is an exciting development, it’s essential to consider the shortcomings of conventional scouting methods. Scouts generally depend on personal judgments and inconsistent information while evaluating potential signings. This method is labor-intensive, costly, and often subject to human mistakes and biases. Scouts frequently travel to observe players live but often find it difficult to make well-informed comparisons between prospects. The football transfer market has also become less effective recently, with exorbitant fees paid for untested talents. AI scouting offers a remedy for these issues.
AI scouting employs advanced algorithms to analyze complex metrics and video data on millions of players across the globe. These systems can assess footage to evaluate technical abilities, movement, positioning, and various other traits. By standardizing the evaluation process, AI scouting eliminates human prejudice and provides consistent, comprehensive insights. These models are also better at predicting player performance and growth statistically. AI analysis identifies promising talents much faster than traditional scouting networks. This enables clubs to spot undervalued players ahead of their competitors. Additionally, AI can assist in customizing training regimens and recommend positional roles that align with players’ strengths. The insights provided by this technology far surpass the limited observations made by human scouts.
Trailblazing clubs have already showcased the potential of AI scouting. In 2020, Inter Milan acquired defender Pitaluga from Fluminense after evaluating his attributes through AI analysis. Midtjylland in Denmark has gone even further, attributing their remarkable league title victory to their AI scouting system.
The integration of AI in football is likely to remain confidential, but it is undoubtedly part of the success narratives of Tony Bloom and Matthew Benham, the innovative owners of Brighton and Brentford, respectively.
Rumors indicate that they employ teams of “quants” focused on identifying undervalued players in global markets, akin to the Moneyball strategy. Machine learning (ML) is probably already a facet of their business endeavors and will rely on players maintaining their trajectories over the next decade while establishing themselves in the European football landscape.
These instances highlight how smaller clubs can compete effectively against elite teams by adopting AI scouting. This technology provides an affordable pathway to high-level insights once exclusive to financially dominant clubs like Real Madrid.
Gamechanger?
AI technology offers professional football teams an exceptional chance to revolutionize their scouting practices and secure a significant advantage over competitors. By leveraging AI’s analytical capabilities and vast database, clubs will be able to make more informed signings, discover hidden talents, and maintain a competitive lead. While this might seem improbable, an article in Wired magazine revealed that Liverpool recently partnered with DeepMind to merge computer vision, statistical learning, and game theory to help teams uncover patterns in the data they gather.
Though traditionalists might resist, innovation is crucial for success in the intensely competitive football market. Clubs that do not adapt risk falling behind. Although the use of AI may raise concerns about reducing players to mere statistics, if implemented ethically, it can benefit clubs, players, and fans by allowing talent to flourish. The moment has arrived for clubs to adopt this transformative technology that has the potential to change the landscape of player recruitment.
When Yaya Touré relocated to Europe in 2001, it was made possible by the personal link between his youth team ASEC Mimosas and the Belgian club Beveren.
He was one of several players who transitioned from the Côte d’Ivoire club to Beveren. The expenses tied to properly scouting youth players meant that unless they were signed by top clubs like Arsenal, which directly acquired Yaya’s brother Kolo Touré from Mimosas, there were limited pathways for elite young Ivorian athletes to reach Europe.
Fast-forward 22 years, and any club in Europe can now carry out thorough research on any player in the ASEC Mimosas youth academy for less than the price of a round-trip flight to Abidjan.
Instead of traveling multiple times to observe various youth matches among the best Côte d’Ivoire teams, scouts can examine every player in detail on their laptops.
The system facilitating this process is known as Eyeball. It has been utilized by clubs such as AC Milan, Lille, and Benfica to recruit over 150 youth talents. David Hicks, the Director of Eyeball, mentions that ASEC Mimosas previously received one visit per month, but thanks to this system, they now receive 30 to 40 inquiries monthly about players. Instead of traveling, people are now reaching out and saying, “we have been monitoring this player for several months,” “we are impressed with him,” or “can you provide more details,” prior to deciding whether to visit Mimosas in person or invite the player for a trial in Europe.
Eyeball operates by using a high-resolution camera positioned high above the field to capture 180-degree views and create angles for artificial intelligence software to analyze. This software monitors each player and generates individual clips of their actions along with statistics comparable to those from OPTA.
Scouts can then utilize the system to search for specific attributes such as age, height, or speed, and view recent matches of that player. They can also identify the individuals responsible for the player, ensuring they know whom to contact regarding them. Twenty-five leading academies in West Africa are part of the system, allowing scouts from clubs like Liverpool or Manchester to watch matches complete with detailed data by Tuesday.
This enables scouts to review all these games before making a decision about a player, meaning if a player isn’t a fit, they haven’t wasted numerous trips trying to determine that.
Consequently, acquiring players from these clubs no longer necessitates a personal connection, as was the case with Beveren, or a significant scouting budget like that of a top Premier League team. Hicks describes this as “revolutionary.”
The Eyeball system is also implemented in various other countries, including France, where it captures all youth clubs within the top regional and national leagues, allowing teams to seek out the best young talents who might have been overlooked by the academy system. Since it targets professional clubs, Eyeball is focused on the top youth leagues in the countries where it has expanded its reach.
One of these nations is Iceland, where a Champions League club in mainland Europe used Eyeball to scout a top youth talent, extending beyond their usual scouting regions.
In the UK, Brexit has complicated the ability of clubs to easily recruit youth players from many of the aforementioned countries.
Hicks notes that within England, professional clubs tend to be quite secretive about their youth players and are reluctant to adopt the system, which he believes could assist youth players who have been released in finding new clubs. Currently, following the disappointment of being let go, players often have only a brief opportunity at trial matches to demonstrate their skills, but Hicks argues that having an easily searchable database of all youth matches for those players could aid clubs in deciding whether to sign players released by their competitors.
However, the Eyeball system is operational in Northern Ireland and is set to go live in Scotland soon, two regions where English clubs are showing more interest in scouting post-Brexit.
In addition to enhancing scouting, this technology is also aiding youth clubs in raising their standards. For instance, in Côte d’Ivoire, it can be utilized to enhance training and coaching sessions and help players become accustomed to the data analysis of their performance that is standard at top-tier clubs in Europe.
Looking ahead, Hicks envisions that comparing players across leagues will become even simpler, enabling clubs in one country to understand the specific areas they need to improve to compete with youth players on the opposite side of the globe.
Brighton & Hove Albion are trailblazers in integrating AI into football. They are revolutionizing the conventional methods of evaluating prospective new players.
What distinguishes Brighton & Hove Albion from Chelsea over the past year? One club prioritizes financial power in making transfer decisions, while the other heavily utilizes artificial intelligence to identify new talent.
Chelsea is known for its excessive spending. Since Todd Boehly acquired the majority ownership from Roman Abramovich in June 2022, the club has invested over 1 billion pounds, approximately IDR 19.2 trillion, to sign 31 players for “The Blues.”
In contrast, Brighton has spent a total of 497.06 million euros (around Rp 8.15 trillion) across seven seasons in the Premier League, England’s top tier. Meanwhile, “The Seagulls,” as Brighton is nicknamed, have consistently increased their revenue from player transfers each season.
To date, they have earned 447.92 million euros (about IDR 7.34 trillion) from selling players to other clubs. Their highest transfer income was achieved in the summer of 2023, reaching 190.2 million euros (around IDR 3.12 trillion).
Despite their significantly different financial positions, Brighton has outperformed Chelsea. This was particularly evident during the 2022-2023 season, when Brighton qualified for the Europa League for the first time, finishing in sixth place, while Chelsea ended up in twelfth.
At the beginning of the season, Brighton was among six teams that achieved three wins in the first four matches. On the other hand, “The Blues” found themselves in twelfth position at the international break, having garnered only four points from one victory and one draw.
Brighton’s success in transfers can be attributed to their adoption of cutting-edge technology. Unlike many teams that still rely on traditional scouting methods, the Seagulls’ management utilizes an artificial intelligence-based application to analyze thousands of player data.
This application, named Starlizard, was created and developed by Brighton’s owner, Tony Bloom, since 2006. Over its 17 years of existence, Starlizard has focused on offering data analysis to assist individuals in making informed choices when gambling at casinos, whether for sports or poker.
Bloom, who earned a Bachelor’s degree in Mathematics from the University of Manchester, applies his knowledge of calculations and formulas to enhance the application that aids his activities as a professional poker player and sports bettor. He established Starlizard as a pioneer in AI for sports.
According to The Sun, Bloom has employed advanced statistical evaluations through Starlizard, including expected goals (xG), which have surged in popularity over the last three years, originating in the early 2010s. He leveraged this data to elevate Brighton from a League One club, the third tier in England, to a competitive mid-table team in the European Premier League.
Through Starlizard, Brighton gathers crucial player metrics globally that align with their playing philosophy, such as passing skills, chance effectiveness, and potential injury risks. This methodology enables Brighton to sign talented players that larger clubs often overlook, including Alexis Mac Allister, Leandro Trossard, Moises Caicedo, Kaoru Mitoma, and Evan Ferguson.
Brighton feels confident replacing key players from last season—like Mac Allister, Caicedo, and Robert Sanchez—who left this summer. They have successfully filled those gaps with suitable alternatives at significantly lower costs, such as Mahmoud Dahoud, Carlos Baleba, and Bart Verbruggen.
“We have a method to analyze data and use it to inform our decisions,” Brighton’s CEO Paul Barber stated in an article by The Telegraph in January 2023.
In terms of player recruitment, Starlizard categorizes the collected data into three types: players acquired for immediate impact, players beneficial for both the present and future, and those signed for future prospects.
Mac Allister and Mitoma belonged to the third category when Burung Camar acquired them. Upon arriving from Argentinos Juniors, Mac Allister was temporarily loaned to Boca Juniors for a season, while Mitoma, who came from Kawasaki Frontale, was initially assigned to another Bloom club, Royale Union Saint Gilloise, in Belgium.
Facundo Buonanotte and Julio Enciso are two South American players that fall into the second player category. Players older than 25, like Dahoud and James Milner, are placed in the first player category.
Brighton also utilizes indicators in its player database that resemble traffic lights. A green light signifies a perfect match with the club’s playing style, yellow indicates players nearing the criteria, and red is for those who require closer monitoring.
Even though Brighton primarily relies on data for player evaluations, they still employ professional talent scouts. However, they do not send scouts worldwide to gather information and keep direct tabs on players.
Instead, Brighton has innovatively organized talent scouts to focus on specific positions. Thus, instead of having scouts for regions like Europe or Asia, Brighton assigns them to specializations such as goalkeeper, central defender, wingback, midfielder, winger, and striker.
For example, John Doolan was hired as the talent scout manager for midfield strikers. He previously held the position of head talent scout for Everton in the UK for a decade.
Brighton manager Roberto De Zerbi acknowledges that he has gained new insights while spending six months in Brighton. Although he is recognized for his sharp acumen in identifying young talent during his time with Sassuolo and Shakhtar Donetsk, De Zerbi finds Brighton’s use of AI to be very beneficial for assessing potential new players.
“At my former club, my scouting team would provide me with player names, and I would evaluate them solely through video footage, without using data. Now, I have begun to adapt to utilizing algorithms to discover new players in the transfer market,” De Zerbi shared with The Athletic.
Through Starlizard, which employs around 160 staff members, Brighton has already stepped into the future of football. If they continue to perform well in Europe, it could greatly enhance the Seagulls’ financial standing. With a mix of AI and increased funding, Brighton holds the potential to emerge as a new powerhouse in England.
We often discuss the concept of football intelligence. In the current data-driven era, this is increasingly being rivaled or complemented by artificial intelligence. Numerous teams are implementing AI technologies to evaluate their players’ performance, decide on tactical approaches, and predict the movements of their rivals. This development carries both sports-related and financial implications.
For over thirty years, football enthusiasts have engaged with Football Manager (and its predecessors), a video game where players oversee their professional football team largely based on statistics, even though actual football encompasses non-statistical elements, and excelling in Football Manager may not translate to real-life coaching, where factors like charisma and interpersonal skills are crucial. However, today, AI is bridging the gap between authentic football and the management style of Football Manager.
The revolution in performance assessment
Artificial intelligence (AI) has fundamentally transformed how performance and market worth are measured in sports analytics, particularly within football. As the French Olympic Games draw near, this evolution is becoming increasingly advanced. AI is revolutionizing performance evaluation by examining and interpreting vast amounts of data to provide new insights that can enhance player development, team strategies, and overall game outcomes.
Traditionally, the valuation of football clubs depended on financial indicators like earnings from ticket sales, merchandise, and broadcasting rights, while player performance was assessed using basic metrics, including goals, assists, and defensive actions. Although these statistics hold significance, they fail to capture the complete picture of a player’s or team’s impact on the game, leading to misrepresentations in valuations and missed investment prospects.
Consequently, the intricate nature of football, characterized by its fluid and dynamic qualities, requires a more detailed approach to performance evaluation. The emergence of cutting-edge technologies, particularly AI and machine learning, has revolutionized this area. AI can analyze data from diverse sources, such as player tracking systems, video analysis, and physiological metrics, delivering a more thorough and objective performance assessment.
A fresh perspective on football
AI technologies allow for the analysis of extensive data with unparalleled accuracy and speed. This capability facilitates a more intricate understanding of both individual players and team performances, accounting for variables that conventional methods might neglect. For instance, advanced AI models like the Complex Multiplex Passing Network (CMPN) categorize different types of passes and interactions in a match.
These models reveal insights that traditional statistics may overlook, such as the tactical significance of particular passes or the adaptability of players. This detailed analysis contributes to more precise evaluations of clubs. Additionally, machine learning models, including multiple linear regression and random forest models, have been created to forecast player salaries based on performance metrics and attributes. These models consider non-linear connections between variables, offering more accurate predictions compared to traditional techniques.
A significant area where AI has progressed is in the analysis of player movement and positioning. AI algorithms can monitor players’ movements during matches, evaluating their positioning, speed, and decision-making processes. This information aids in understanding how players contribute to both the offensive and defensive phases of the game, surpassing traditional statistical measures.
For players and supporters
AI also plays a vital role in managing and preventing injuries. By monitoring players’ physical condition and workload, AI can identify early signs of fatigue or potential injuries. This proactive approach allows for timely interventions, reducing the risk of injuries and ensuring that players are fit for crucial games. For example, AI systems in top teams have decreased injury rates by around 30% by providing real-time data on players’ physical strain and recommending rest periods or personalized training plans.
Moreover, AI enhances fan engagement by delivering deeper game insights. Augmented reality (AR) and virtual reality (VR) applications powered by AI offer fans immersive experiences, such as interactive match analyses and virtual stadium tours. This enriches the viewing experience and strengthens the connection to the sport. AR and VR technologies present fans with innovative ways to engage with the game, resulting in a more immersive experience.
To improve decision-making
During the Olympics or UEFA Euro 2024, teams will utilize AI to refine performance in real-time. By examining live match data, AI can provide actionable recommendations, such as proposing tactical changes or pinpointing players who might require rest or substitution. Indeed, in a pivotal match, AI can notice a decline in a key striker’s sprint speed, alerting the coach to substitute him before fatigue results in a possible injury, utilizing solutions like those offered by Statsports, which is employed by multiple Premier League teams such as Arsenal and Liverpool.
Consequently, private equity firms investing in football clubs greatly benefit from the capabilities of AI. AI can evaluate historical data, forecast future performance, and uncover potential investment opportunities. This leads to more knowledgeable decision-making, enhancing returns while reducing risks. There is a clear connection between the performance of football clubs and their stock prices.
Performance of football clubs and stock prices
For example, victories can significantly elevate stock prices, demonstrating the direct influence of match results on financial valuations. Borussia Dortmund FC, for instance, experienced a rise in its stock price from 2.80 euros to 4.50 euros per share following their successful semi-final in the 2012-2013 UEFA Champions League.
AI models can explore such relationships, yielding insights that assist investors in making more informed choices. AI also facilitates the examination of social media sentiment, which can forecast stock price fluctuations. For instance, positive sentiment on platforms like X or Instagram can prompt stock price hikes, whereas negative sentiment can lead to declines, just as Manchester United’s stock price fell in 2021 after the announcement of forming a European Super League. Hashtags such as #NoToEuropeanSuperLeague and #GlazersOut trended globally.
As the adverse sentiment intensified, especially with the threat of fan boycotts and potential loss of sponsorships, and due to the abrupt drop in stock price, the club began to withdraw from the ESL in response to the backlash. Merging social media analytics with conventional data sources offers a more thorough understanding of stock price movements.
Overall, the integration of AI in football valuation provides a competitive advantage. AI not only improves the precision of valuations but also delivers deeper insights into performance metrics and market dynamics. By adopting AI-driven analysis, football continues to progress, guaranteeing that every element of the game is carefully assessed and enhanced. The future of football performance valuation is here, characterized by intelligence, data-driven approaches, and remarkable accuracy.
In this article, we will explore how artificial intelligence is utilized in the realm of sports.
1. Customized training and diet programs
A team’s achievement relies significantly on the nutrition that empowers them on the field. The LaLiga Santander football club Granada CF recognizes the importance of nutrition for their performance and has partnered with the University of Granada and Tabalu to implement artificial intelligence in automating nutritional planning for athletes, tailoring players’ diets.
Over recent years, the club has integrated biomechanical assessments and body composition evaluations that aid in creating a comprehensive database of information about each athlete. By tracking metrics such as weight, body fat percentage, skin and muscle mass, and intracellular and extracellular fluid, Granada CF can determine what and when their players should eat. Together with the University of Granada, they have created an innovative app called Readiness Soccer, which assists the club in monitoring and tracking their athletes.
Food technology firms are also collaborating with football clubs to prepare all meals, including those eaten at the club facilities and those taken home. By working alongside team doctors and leveraging available technologies, the club formulates tailored plans for each player and employs artificial intelligence to streamline processes, identify patterns, and efficiently strategize diets.
Athletes across various sports are increasingly turning to apps like FoodVisor, which employ object recognition technology to identify over 1200 types of food, estimate portions, and generate a quick nutritional breakdown tailored to a player’s requirements. These AI-enhanced fitness applications also use computer vision techniques for real-time human pose estimation, offering players guidance on exercising correctly.
The capacity of AI to evaluate which diets result in the least risk and maximum efficiency paves the way for highly personalized nutrition for athletes across different sports. Furthermore, the growing availability of data will allow researchers to develop improved models that enhance diet plans while optimizing the fitness and performance of athletes.
a. Evaluation of player performance
Performance analysis is a field that supplies coaches with objective insights, helping them comprehend player performance. This analysis is essential for recognizing players’ strengths and weaknesses and identifying areas needing improvement.
In the past, coaches relied on their acquired knowledge through extensive sports experience to make decisions. Initially, this involved handwritten notes, which have now developed into sophisticated computerized systems and technologies that gather extensive performance-related data, including qualitative data, player acceleration, speed, and video sequences for in-depth performance assessment.
Recently, the Cornell Laboratory for Intelligent Systems and Controls has created an algorithm capable of predicting 80% of volleyball players’ in-game actions accurately. This algorithm merges visual data—such as an athlete’s position on the court—with information regarding an athlete’s specific role on the team. Coaches can use this data to enhance their competition preparation by training players with existing game footage from competitors, thereby gaining an edge. Basketball teams have also harnessed the advantages of AI, utilizing a computer vision-based application called HomeCourt. This app aids players in assessing their basketball skills by monitoring their performance metrics, including shooting statistics.
Technology is providing coaches with access to data through AI-powered platforms, offering them quick insights into areas requiring improvement and helping them recognize and anticipate the strengths and weaknesses of their opponents. This capability influences coaches’ decision-making regarding tactical choices and team selection, allowing athletes and teams to better exploit their rivals’ weaknesses while addressing their strengths.
2. Talent scouting and recruitment
Annually, recruitment agencies seek out promising young players who might become the next stars. Unfortunately, many of these potential superstars end up as squad members or, worse, become irrelevant. Nevertheless, the integration of technology into sports is opening up new pathways for discovering talent.
Currently, AI technologies such as computer vision assist recruiting agencies in analyzing player movements on the field, while machine learning algorithms predict skill levels, attacking and defending qualities, and overall player performance.
Machine learning algorithms also help scouts gain a clearer understanding of each player’s strengths and weaknesses, highlighting areas that need focus. AutoStats, a computer vision technology powered by AI, utilizes body recognition technology to produce detailed basketball tracking data that showcases the attributes and playing styles of prospective NBA talents, creating a comprehensive overview of each player’s potential.
Looking ahead, AI is set to enhance the recruitment process, making it faster and more efficient while minimizing biases, such as dismissals based on a player’s ethnic or social backgrounds, through objective assessments and evaluations of players. By utilizing AI to analyze player characteristics, scouts and coaches will gain better insights into which players align with their selection criteria, determine a player’s optimal position on the field, and enable them to design training programs that help players adapt and improve.
Conclusion
It is difficult to question the impact of AI, considering the significant contributions it has made to the largest industries globally. The sports sector is no different. Artificial Intelligence is gradually establishing its presence in the sports industry, significantly enhancing the competitiveness of teams and athletes. With such a wide array of applications, the entire sports field will inevitably seek to implement AI, leading to further innovations and improved results.
In the realm of soccer (known as football in most countries), where each pass, move, and goal is significant, this sport is undergoing a groundbreaking transformation. The driving force? Artificial intelligence (AI). From Major League Soccer (MLS) in the United States to the elite European Premier Leagues, AI is redefining the beautiful game in ways that were once only envisioned.
Exploring AI’s Contribution to Soccer’s Advancement
The introduction of AI into soccer is an exhilarating experience, merging tradition with state-of-the-art technology. This combination is fostering smarter strategies, improving player performance, and connecting fans more closely to the action. Leading European teams such as Liverpool and Barcelona are at the forefront of integrating AI into their tactical approaches. By analyzing extensive data, AI aids in deciphering the play styles, strengths, weaknesses, and strategic possibilities of opponents. This data-driven methodology allows coaches to create more effective strategies and secure a competitive advantage.
Furthermore, AI’s influence on player performance and health is significant. Teams in the MLS, including LA Galaxy and Atlanta United, employ AI for thorough performance evaluations. Sophisticated algorithms track player movements, monitor fitness, and evaluate injury risk factors. This not only enhances performance but also considerably decreases the chances of injuries, ensuring players maintain peak condition.
In terms of recruitment, AI is transformative. Teams like Manchester United and Bayern Munich utilize AI algorithms to scout talent worldwide. By examining data from various match levels, AI assists in identifying promising players, evaluating not just their current abilities but also their growth potential.
Off the field, AI is revolutionizing the fan experience by making it more interactive and tailored. Clubs such as Real Madrid and Manchester City utilize AI to deliver personalized content, match analyses, and predictive insights. This strengthens the bond between the club and its international fanbase.
Operational Efficiency Beyond the Field
Beyond the game itself, AI simplifies operations within clubs. From dynamic ticket pricing to merchandise sales, AI empowers clubs to make informed decisions based on data, maximizing revenue streams and improving the overall business model. The practical application of AI in soccer is as varied as it is inspiring. A notable example is FC Barcelona’s Barça Innovation Hub, demonstrating the club’s dedication to using technology, including AI, to stay ahead.
However, the real charm of AI in soccer lies in its partnership with human intelligence. Coaches, players, and staff are not being replaced; instead, they are being equipped with insights that were once unattainable.
The Future Field: AI in Soccer’s Horizon
As we gaze toward the future, AI holds promise for even more thrilling advancements in soccer. Envision AI-driven virtual reality experiences that bring fans from their homes into the stadium’s heart, or enhanced player analytics that transform training and gameplay strategies. For aspiring players, coaches, and enthusiasts, AI has unlocked a myriad of opportunities and welcomed them to partake in a revolution, inspiring innovation and early adoption in an area full of potential.
Nevertheless, as AI becomes increasingly embedded in soccer, ethical issues, especially concerning data privacy and fairness, are crucial. It is essential to balance technological progress with ethical responsibility for the sustainable integration of AI into the sport.
As we look forward, AI in soccer represents not only a technological progression but a new chapter in the sport’s illustrious history. It narrates the story of how data and algorithms can amplify human talent and passion. As teams worldwide embrace AI, they are crafting a new strategy for success, one that melds the best of human skill and machine intelligence. In this new chapter of soccer, let’s participate as players, fans, and innovators, bound together by our love for the game and our anticipation for its future. The playing field is advancing, and with AI, we are all part of this beautiful, unfolding narrative.
Leave a Reply