"Even if AI finds a good player, the final choice rests with humans": What K-League scouts never overlook... "Character, attitude, and adaptability cannot be measured by numbers" [22nd Anniversary Special Project ③]

*This content was translated by AI.

Jürgen Klopp, the head coach of the German national team, wanted Julian Brandt as a striker during his time leading Liverpool (England). However, Liverpool's data analysis team looked elsewhere.

It was immediately Mohamed Salah. It was a choice made by data analysis. The result was a great success.

It is not just Salah. Roberto Firmino, Andy Robertson, and Sadio Mane, who have long been key players for Liverpool, are also cited as representative success stories of data-based player evaluation scouting.

According to Sky Sports UK, former Liverpool data research head Ian Graham explained that the data analysis department played a significant role in the club's success. He developed a program called "Possession Value," which calculates how much each play increases a team's scoring probability every time a player touches the ball. Using this tool, Graham evaluated and recommended potential signings, which became a crucial stepping stone for Liverpool to embark on a new golden era.

Graham served as head of data research at Liverpool from 2012 to 2023. During this period, Liverpool lifted numerous trophies, including the UEFA Champions League (UCL) and the English Premier League (EPL).

A typical example is Mane and Salah. When manager Jurgen Klopp wanted Mario Goretzka, Liverpool's data analysis team recommended Mane. Additionally, when the club pushed to sign Brandt, the data analysis team gave Salah a higher evaluation. Sky Sports explained that "Liverpool convinced manager Klopp to agree to signing Salah."

As time passes, data analysis is advancing at an even faster pace by combining with artificial intelligence (AI) technology. While the focus was previously on collecting and organizing game records that had already occurred, it has now progressed to predicting situations that have not yet happened and assigning value to every single action of players.

In modern soccer, one of the fields where AI is most widely utilized is player evaluation. Go Sang-gi, an artificial intelligence researcher at the University of Seoul, recently told StarNews, "The most basic example is expected goals, or 'xG.' It uses machine learning models to calculate the probability that a shot in a specific situation will result in a goal." He added, "Recently, models have emerged that evaluate the value of all actions performed by players on the field, including passes, shots, and dribbles, as well as off-the-ball movements when they do not possess the ball."

The representative club that accepted such data analysis faster and more proactively than anyone else was Liverpool.

Professor Ko stated that overseas clubs are already utilizing data analysis and AI for player analysis and scouting, noting that when Liverpool signed players such as Mohamed Salah, Andy Robertson, and Roberto Firmino, it is known that the club's data analysis department employed AI-based match analysis models and player evaluation models to assess and scout them.

▶ AI to search for players more broadly

One of AI's greatest weapons is undoubtedly its 'scope' to identify more players. Professor Ko stated, "Human scouts can only gather information on a limited number of players and rely on watching games directly, which restricts the pool of players they can analyze," adding that "AI-based models can simultaneously analyze many players worldwide as long as data is provided, making such constraints far less significant."

He added, "It is understood that clubs currently utilizing AI employ it to generate a kind of candidate list," explaining that "when AI recommends an initial recruitment list, human scouts and analysis teams conduct more detailed investigations on the players within that list and review contract terms before proceeding with final acquisitions."

Liverpool had a background that allowed it to embrace data analysis more actively than other clubs earlier. The club's owner, Fenway Sports Group (FSG), had already experienced success by utilizing a data-driven decision-making system at Boston Red Sox of Major League Baseball (MLB) in the United States. In 2004, Boston reached the pinnacle of the World Series for the first time in 86 years.

ESPN in the United States reported that "at Liverpool, there was a clear directive from the club's management to reduce subjectivity in player scouting."

The greatest advantage of data analysis is also its "scale" and "consistency." Graham emphasized that "using data makes it much easier to scout a far wider range of players."

He added, "No human being can watch every soccer match held worldwide in person and simultaneously judge what value each player is creating." He further explained, "Moreover, using data makes it inevitable to systematize the scouting process, as the same filter can be applied to all players."

▶ Teams that use data become stronger

Recently, not only Liverpool but also Brighton & Hove Albion and Brentford, among other teams in England that prioritize data and AI technology, continue to increase.

Notably, Brighton and Brentford are representative cases that have established themselves in the English Premier League (EPL) by prioritizing efficient player acquisitions and club operations despite relatively limited capital. Brighton was promoted from League One (third division) to the Championship (second division) in 2011 before rising to the EPL in 2017. Brentford also secured promotion from League One to the Championship in 2014, finally achieving EPL promotion in 2021.

Graham cited Liverpool, Brighton, and Brentford as clubs that effectively utilize data analysis, revealing behind-the-scenes stories: "When Liverpool discovered a young player from the lower leagues, rumors would arise that 'Brentford might take him.'" This underscores the superior player scouting capabilities of clubs that leverage data.

Professor Ko also predicted that how well a team leverages AI could impact its future competitiveness. He noted, "The difference between teams using AI and those not using it may not be immediately apparent," but added, "If better decisions are repeatedly accumulated, the gap will gradually widen."

▶ The Hiring Market for Soccer Clubs Transformed by AI

The global football community is already rapidly transforming alongside the adoption of artificial intelligence. The change does not stop at player acquisition methods alone; it is reshaping club organizational structures and even the recruitment market.

The UK Financial Times reported on the 29th of last month that "clubs are intensifying their competition to secure data," noting that "the share of professional football clubs in the overall UK recruitment market has tripled over the past five years, with a particularly sharp increase in demand for data experts."

According to reports, more than seven out of every 10,000 job postings coming from the United Kingdom are from professional football clubs. This represents a significant increase compared to 2.4 per 10,000 in 2020.

In particular, data and analysis-related positions accounted for about 5 percent of job openings posted by EPL clubs this year. The media outlet explained that this is double the level seen two years ago.

Indy's Senior Economist Jack Kennedy said, "Data and analytics are areas where teams are focusing more to gain an edge over rival clubs in areas such as player acquisitions and opponent analysis."

In fact, Tottenham Hotspur recently recruited analysts for scouting insights, while Brentford also posted job openings related to its insights and strategy departments. Over the past five years, Brighton, Newcastle United, and Manchester United have been cited as clubs that most significantly increased their hiring of personnel in data and analysis roles.

Kennedy recently described Brentford and Brighton, both of which have been promoted to the Premier League in the last decade, as "teams that achieve results exceeding their size by leveraging data utilization capabilities."

He explained, "Since the owners of both clubs are professional gamblers who made their money through betting, they have always placed great emphasis on data and applied that approach to running football clubs. Traditional large clubs can almost solve problems with money. Even if one signing fails, they can spend another £100 million on a different player. Small clubs cannot do the same." In this way, the advancement of AI is creating new roles and organizational structures within football clubs.

For those preparing to enter the soccer industry, AI utilization capabilities are becoming a new competitive edge. Professor Ko stated, "It is important to quickly learn what tasks AI can perform in the soccer field," and added, "One must understand which parts of the work they wish to do can be assisted by AI. This is not a matter specific only to soccer but a challenge that will apply to nearly all professions moving forward."

He also emphasized that "in the future, regardless of one's profession, an era will come where people work alongside AI," and added that "whether one wishes to be scouted or become a leader, it is essential to clearly understand how AI can assist in their respective roles."

▶ From Seville and Chelsea to MLS... AI scouting spreading globally

Change is not limited to England. Even in Spanish football, traditionally regarded as somewhat distant from data analysis, a new trend is emerging. At the center of this shift stands Sevilla.

Sevilla has been utilizing the AI scouting program "Scout Advisor," developed in collaboration with global technology firm IBM, since 2024. The initiative aims to enable clubs to search more quickly and easily for players who meet specific criteria within a vast database of player data.

Thanks to this, scouts can significantly reduce the time they previously spent sifting through a vast pool of players to identify candidates. Instead, they can dedicate more time to building direct relationships with potential signing targets to understand their personalities and attitudes, observing games firsthand, and making final decisions based on data.

Elias Zamorano, the chief data officer of Sevilla, praised it as "the most revolutionary technology I have seen in football."

Another EPL club, Chelsea, is utilizing artificial intelligence for youth scouting. Through its "Future Blues" program and AI scouting tools, Chelsea evaluates young players not only in the United Kingdom but also around the world. The system works by having athletes download an AI scouting application onto their smartphones, perform designated exercises and technical drills specified by Chelsea, and then upload videos of their performance for AI analysis.

AI tracks not only players' physical movements but also the movement of the ball. It analyzes how player and ball movements are connected during passing or shooting processes and evaluates them using various metrics.

Subsequently, Chelsea's scouts and coaches reassess the player's potential based on data provided by AI. The AI first filters through a vast number of players, while final evaluations are left to human experts.

Major League Soccer (MLS) in the United States is also moving in the same direction. MLS has partnered with 'ai.io' to introduce a system that allows players worldwide to receive initial scouting evaluations through AI without having to travel to the United States or Canada at great expense.

It is similar to Chelsea's approach. When players record designated training sessions on their smartphones and upload the videos, AI analyzes their abilities and assigns scores. Players who receive high ratings are given the opportunity to participate in events held in the United States and Canada, where they train in front of actual MLS officials.

AI is breaking down the biggest barriers in traditional scouting: cost, distance, and time constraints.

▶ AI scouting has also permeated the K League

Signs of change are gradually emerging in the K League as well. Some clubs have begun utilizing AI programs for scouting and player analysis. While this is not yet at the stage of full-scale formal adoption, it is closer to a phase of accumulating data and testing possibilities for how such tools can be applied in actual field operations. Nevertheless, the global trends in football are slowly beginning to permeate the K League as well.

A club's scouting official told StarNews in an interview that "in the player acquisition process, we utilize both general statistical data and qualitative assessments from the technical team." The official added, "When initially reviewing candidate players, scouts evaluate them through live games and video footage, while the operations team compiles meaningful data such as appearance records and various performance metrics to reflect in the player acquisition report."

He added, "Even at the final stage of recruitment decisions, judgments are made based on player recruitment reports that integrate these evaluations and data," explaining that "the approach involves not simply selecting players with good statistics, but also considering how well they fit the current squad composition and the coach's football direction."

In the case of AI data, the framework for its utilization is still being concretized. A relevant official stated, "We are in the process of finalizing necessary indicators to determine how to quantify and evaluate a player's detailed movements and tactical execution capabilities that align with a coach's game model, going beyond existing statistics such as simple goals, assists, and pass success rates."

However, it is expected that AI will be utilized in even more areas going forward. A relevant official stated, "Once such standards are established in the future, we plan to utilize AI to conduct more detailed analysis of aspects that are difficult to verify with existing statistics alone, such as player movement when there is no ball, actions under pressure and during transition situations, and positioning selection, for use in player scouting and evaluation."

He further emphasized that "even after recruitment, we continue to accumulate performance records and key performance data for player evaluation, using them as reference materials for future contract renewals, salary negotiations, and determining the direction of team operations." He added, "Ultimately, while we are actively utilizing existing data now, the stage is being set to establish standards for the practical application of tactical and behavioral data analysis through AI."

Like overseas clubs, the advantage of AI was undoubtedly in the scope and speed of scouting operations. While Club A previously relied heavily on recommendations from scouts' personal knowledge of specific leagues or players and their human networks, it now uses data and AI to first filter candidates across a much broader range who meet the club's desired criteria.

A club scout official noted, "It has become possible to go beyond simply looking at records such as goals, assists, and pass success rates; instead, the coach's game model first defines what movements a player in a specific position should make, and then identifies players who closely meet those conditions."

He added, "In the past, we first looked at which player was a good player, but now I believe the process of defining what kind of player our team needs and then finding someone close to that profile has become much more sophisticated."

Changes have also occurred in administrative areas. In the case of Team A's operations team, while their primary role previously involved contracts, administration, and basic data organization, they now additionally analyze and compare player data evaluated by the technology team to support decision-making.

A source said, "As the use of AI becomes more widespread in the future, we expect to see increased interaction not only verifying players scouted by scouts but also proposing players discovered through data to the technology team."

Youth talent scouting is also a field with significant potential for AI application. A source positively assessed that, in the long run, the value of AI in this area could grow even larger.

In the case of lower leagues or youth teams, sufficient data is often not accumulated, making it difficult to compare them as precisely as first-division players. However, if game footage is sufficiently secured, AI can play a role in narrowing down candidates initially in areas where scouts previously had to personally review every single match.

It is expected that cases where players discovered and recommended by AI make their debut in the actual professional league will emerge in the future. A related official explained, "In traditional scouting, there are realistic limits to the number of games a person can watch, so more attention inevitably goes to players from famous leagues or top teams. However, AI and data allow for comparisons based on what players actually do during games, rather than relying on the fame of the league."

He further emphasized, "Therefore, even if a player would not have attracted attention in the past due to league level or team affiliation, they are now more likely to be considered as a candidate if they possess exceptional skills." He added, "From the players' perspective, how they play will become just as important as where they play in terms of evaluation."

▶ Still, the final choice is up to human eyes

At first glance, it may seem that the role of a human scout has diminished significantly. However, when examining the player recruitment processes of the world's top clubs, the story is quite different.

Even as AI and data continue to advance, the human eye ultimately reappears at the final moment.

Liverpool also did not fully recognize Roberto Firmino's value from the very beginning based on data alone. While data analysis played a significant role in the acquisition process, it is possible that Firmino would never have worn Liverpool's jersey without human judgment.

Graham introduced the fact that Michael Edwards, who was Liverpool's technical director at the time, strongly advocated for signing Firmino. Firmino could play in three positions: striker, attacking midfielder, and winger, and the club's data analysis team struggled to decide which position should be prioritized when evaluating him.

Initially, Firmino was classified as a forward. However, his evaluation as a forward based on data was not high. Edwards then argued that Firmino should be viewed in the 'No. 10' role. The data analysis team re-analyzed Firmino, and the results were surprising. When classified as a 'No. 10', Firmino was rated as one of Europe's top young players.

The British Guardian introduced this anecdote, explaining that "at the time, young players who received higher evaluations than Firmino included Alexis Sanchez, James Rodriguez, Isco, Oscar, Paul Pogba, and Aaron Ramsey."

Graham reflected, "If Edwards hadn't insisted that we continue to analyze Firmino in depth, we likely wouldn't have spent so much time analyzing him."

Ultimately, while the data strongly supported the acquisition of Firmino, it was human judgment that prompted a re-examination of that data and ensured its proper utilization.

A K League A club also shared a similar view. The A club has not yet connected an AI-discovered player to actual signing discussions. However, a representative stated, "We believe such cases may emerge in the future," adding, "For instance, even if a player does not stand out based on general statistics like goals, assists, or pass success rates, they could receive high evaluations from AI analysis when examining their movement without the ball, pressing timing, penetration into space, defensive transitions, and positioning required by the coach's game model."

He also predicted, "If a player like that is selected for the candidate pool, people will likely start with the question, 'Why was this player highly rated?' and re-examine the footage. Through that process, they may discover the value of players who were previously overlooked under the old system."

A source emphasized that what is ultimately expected is not for AI to pick players on behalf of scouts, but rather to enable a second look at players who are easily overlooked by traditional scouting methods.

▶ Things AI Still Can't Read

There are areas that cannot be seen through AI or data alone. A source said, "Aspects such as how a player communicates with teammates during the game, whether they carefully observe their surroundings before the ball arrives, what kind of reaction they show after making a mistake, whether they actively request the ball when the team is in trouble or try to hide, and how quickly they adjust their actual play after receiving instructions from the coach are still difficult to explain perfectly with numbers."

Ultimately, this part must be verified through human eyes and judgment. A relevant official explained, "We can see more during training," adding that "training intensity, competitiveness, attitude toward teammates, and even behavior while on the bench are all crucial factors in determining what impact a player will have when joining the team."

The same applies to the recruitment of foreign players. A representative stated, "For a foreign player, it is important not only to have soccer skills but also to be able to adapt to a new country and culture, food, family living environment, and language. Even if a player is excellent, they will find it difficult to fully demonstrate their abilities if they cannot adapt. In this regard, the experience of scouts and on-site experts certainly plays a role."

He added, "I do not want to describe this as merely a scout's simple 'gut feeling.' I believe it is a form of pattern recognition and accumulated experience gained by watching many players and numerous games over a long period, repeatedly experiencing both successes and failures."

Ultimately, human judgment is indispensable. A relevant official stated, "While AI can present strong opinions, the most important criterion is that humans must ultimately bear responsibility for those decisions."

He further emphasized, "I do not think I will discard the AI's evaluation simply because it differs from my own judgment. Instead, the process of re-examining why such different results were obtained is important."

He also explained, "There may be parts we missed, or conversely, the data used by AI might not have fully reflected the player's actual role or the characteristics of the league. Therefore, we re-examine the data and video, and share opinions within our technology team."

Caution is also needed in scouting young players. A related official stated regarding the use of AI scouting, "Young players require particular care because their physical growth stages differ and current performance does not necessarily predict future performance." The official added, "In youth scouting, I believe it is far more important to assess how to judge growth potential and the direction of change than to rely solely on current data figures."

Liverpool also abandoned plans to sign a new right-back after former manager Jurgen Klopp observed Trent Alexander-Arnold, who was then merely a young player, during training. The club kept a spot for him in the squad to allow Alexander-Arnold to grow. He has since developed into a world-class player. This case highlights the crucial role Klopp’s eye for talent played in his development.

Graham also said, "It was really important that the scouts understood that the evaluation we created is not perfect."

Victor Orta, sporting director of Sevilla FC, also emphasized at a conference held during the 2023 World Football Summit that "we will never sign players based solely on data. However, we will never sign players without utilizing data either."

Continuing, Director Orta explained, "Good players always have good data. However, human eyes are always present after that. Ultimately, it is the human eye that evaluates and decides everything."

Ultimately, what matters is how humans interpret the data discovered and generated by AI. The Irish Times emphasized a structure in which AI does not make all decisions independently, but rather humans review AI's proposals to make final selections.

The media outlet explained that "for an AI tool to perform its function properly, humans must first provide industry-scale relevant data and pose appropriate questions," adding that "if humans accept the suggestions offered by AI, those ideas must be implementable in actual stadiums."

▶ AI as the 'Advisor' to Human Scouts

A club official also explained that "scouts judge aspects of a player's character, attitude, growth potential, and tactical understanding that are difficult to fully explain with numbers through games and training," while noting that "data and AI excel at identifying patterns humans might miss, comparing a wider range of players, and objectively verifying those judgments."

He added, "In the future, the role of scouts as assistants will become extremely important in supporting a scout's judgment with data to the same extent as their individual capabilities, establishing criteria that align with the manager's game model, and systematizing player evaluations."

He also emphasized that "ultimately, acquiring good players does not come from either scouts or AI working well alone, but rather from a 'team' that organically connects the qualitative judgment of the field with quantitative analysis based on data."

A veteran scout from Club B also stated, "While AI can provide significant assistance in terms of data analysis, football involves other evaluation dimensions," adding, "For instance, a player's playing style or personality are key factors. It is difficult to perfectly quantify such aspects through data."

He also stated, "Foreign players face similar challenges. In other countries, lifestyles and cultures, including dietary habits, are clearly different. Simply put, it ultimately comes down to the issue of adaptation." He added, "If a player we sign cannot adapt to K League culture, I want to question whether their data is absolutely important."

He further emphasized, "For youth players, the changes are even greater. There may be physical changes, but there are also psychological changes. Even if AI predicts based on data, it cannot be considered 100% accurate."

He added, "While AI will provide assistance, soccer is not something that can be determined solely by statistics," and noted, "Therefore, it seems we cannot rely entirely on AI."

<© STARNEWS. All rights reserved. No reproduction or redistribution allowed.>

*This content was translated by AI.