The world’s largest sports newspaper, MARCA, interviewed Comparisonator CEO Tarkan Batgün through its big data journalist Miguel Ángel García. Here are some key headlines from the interview:
K.Tarkan Batgün arrives for this interview on a special day: Spain and Turkey face off in a match he knows well from his dual perspective as a global analyst and Turkish expert in artificial intelligence applied to football . CEO of Comparisonator —a platform that contextualizes performance, compares players across leagues, and simulates how they would adapt to new competitive environments—he has worked for international clubs, agencies, and consultancies. He created the ‘Scouting Lab’ at Bursaspor , served on the board of Altınordu FK, advised companies like Wyscout and SoccerLab, and was responsible for NIKE Türkiye’s scouting program for six years.
From this multifaceted perspective, he argues that context is key to any data and that AI only makes sense if it helps us make better decisions . In this conversation, he explains how his technology translates football between leagues, detects hidden risks, and prevents multi-million dollar transfers that could go wrong.
Question: You’ve worked on four continents and always insist that data without context is useless. What’s been the biggest culture shock that forced you to completely reinterpret a piece of data or a player’s profile?
Answer: Working on four continents taught me one thing very early on: the same number can mean completely different things depending on where it comes from. And the biggest culture shock, the moment that really forced me to reinterpret the data, came when I went from the structured football of Australia to the emotional, chaotic, high-intensity environment of Turkey.
Let me give you a concrete example: In Australia, I analyzed a midfielder who had excellent passing accuracy: 92-93%. In that league, this usually indicates intelligence, patience, and well-trained positional play. But when I went back to Turkey and applied the same logic, I realized something shocking: a passer with 92% in the Turkish league is often not very creative. He might simply be avoiding risks, playing the ball back, or immediately releasing it due to pressure.
That was the moment I realized that context dictates the truth, and it pushed me to create Comparisonator as a contextual engine for sporting directors, coaches, and recruitment managers: to reinterpret figures through the prism of league pace, to adjust performance to tactical style, to understand how a player behaves outside their environment, to help clubs assess talent globally without falling into misleading statistical traps.
The same number can mean completely different things depending on the country it comes from.
Q: Your career combines club, agency, consulting, and teaching. What did you learn in each of those roles that you now directly apply to the design of Comparisonator’s artificial intelligence?
R. Each stage of my career gave me a different perspective on football, and today, all those perspectives are directly integrated into the AI of Comparisonator.
From the club environment, I learned that decision-makers don’t have time; they need clarity. They don’t want ‘big data’; they want to know if a player fits our style or not. That’s why our AI behaves more like a decision-support advisor than a statistics machine.
From the agency world, I learned that talent trajectories are as important as talent itself. From consulting, I learned that every club has a different reality. That’s why Comparisonator’s AI adapts to the user. It learns the club’s style, needs, and priorities, and tailors its recommendations accordingly.
From teaching and lecturing, I learned that understanding comes from explanation, not numbers. That’s why we created CompaGPT: an AI that explains football data to humans the way an experienced scout or coach would.
Q. At Bursaspor, you created the ‘Scouting Lab’. What part of that idea is still relevant today and what has become completely obsolete with current AI?
A: Thanks to my mentors Christophe Daum and his assistant Rudi Verkempinck, the Bursaspor Scouting Lab was my first attempt at creating a systematic, evidence-based way to evaluate players. Many parts of that idea remain relevant today, but others have been completely transformed by modern AI. Let’s say the ‘Scouting Lab’ was the seed. The methodology (structure, clarity, collaboration) remains relevant. But everything manual, repetitive, or subjective has been superseded by AI. Today, Comparisonator is the Scouting Lab transformed into a global, dynamic, and environment-aware intelligence engine.
Q. Many clubs believe they are using data, but in reality, they are only seeking to confirm their preconceived opinions. How much noise do these biases generate in the modern scouting process?
R. Bias is the biggest hidden cost of modern scouting, and it generates far more noise than clubs realize. Many clubs think they are using data, but in reality, they are using numbers to justify decisions they have already made emotionally. This creates three major problems: you stop discovering new players, because if data is only used to confirm an opinion, you never question your first impression and never uncover unexpected talent; you filter out the truth, as confirmation bias causes clubs to ignore red flags; and you lose your competitive edge, because if all clubs use data to support pre-existing beliefs, they all end up signing the same players.
The biggest hidden cost of modern scouting is bias: it distorts, limits, and leads to lost talent
Q. When a Comparisonator report contradicts the intuition of a coach or head scout, how is that conflict usually resolved? Who is wrong more often?
A: When data and intuition disagree, the first rule is simple: don’t choose either option, investigate further. A coach sees things that data can’t: body language, personality, behavior in training… Comparisonator sees things that a coach can’t: adaptation to the league, tactical stress, hidden risk factors…
In my experience, when conflicts arise, environmental projection (how the player will adapt to the league and the system) is usually where intuition underestimates the risk. That’s precisely where Comparisonator adds value: it doesn’t replace human judgment, but rather protects it from blind spots.
So, who makes the mistake more often? Usually, the side that ignores the context. And in modern football, context is non-negotiable.
Q. Standardization across leagues is one of the biggest challenges in the industry. Which competition is showing the most resistance to the algorithm, and why?
R. The league that creates the most resistance to any algorithm is the one in which football is least standardized, where the pace, structure, and tactical discipline vary enormously within the same match.
For us, these are usually leagues with significant differences in pitch quality, unpredictable pace of play, inconsistent defensive organization, and extreme emotional intensity. Standardization is the problem; contextual intelligence is the solution.
Q. You talk a lot about AI points, trends, consistency, and functional role. Of all these indicators, which one best predicts a player’s future progression?
A. The most reliable indicator of a player’s future progression is the consistency of their performance across different environments. AI points, trends, and role metrics are important, but the real indicator is this: Does the player continue to perform when the context changes? Different pace, different pressure, different tactical demands, different quality of opponent.
Virtual Transfer has already prevented signings that would have cost clubs millions.
Players who maintain their performance across multiple environments almost always progress. Players who crumble outside their comfort zone almost never do. That’s why Comparisonator focuses so heavily on performance stability, league translation, adaptability indicators, and role behavior under pressure.
Q. Virtual Transfer allows you to simulate a player’s performance in another league. Do you have any documented cases where the model prevented a club from making a bad signing?
R. Yes, several, but I can’t reveal the names of the clubs or the players. What I can say is this: Virtual Transfer has already saved clubs millions. A recent case was that of a highly sought-after striker from a fast-paced, open league. His raw stats were spectacular: dribbling, progressive runs, expected goals… Everything suggested he was an essential signing.
But when we ran him through Virtual Transfer and simulated his performance in one of Europe’s top five leagues, two red flags immediately appeared: his efficiency dropped by almost 50% under increased defensive pressure, and his decision-making slowed significantly in structured tactical environments. The club blocked the transfer. Two months later, he signed for another European team and struggled in precisely the areas our model had predicted.
This is Virtual Transfer’s main objective: not to say “no,” but to reveal the truth about how a player behaves outside their comfort zone. In modern recruitment, that clarity can mean the difference between a successful signing and a very costly mistake.
Q. AI platforms and models promise to eliminate biases, but they can also generate them. What has been the biggest ‘false positive’ or system failure that has forced you to revise the model?
R. The biggest false positive we’ve had came from a player who seemed exceptional because the environment of his league artificially inflated his strengths. He played in a competition with very low defensive pressure, very open spaces, chaos in transitions, and extremely high ball recovery zones.
On paper, his metrics were elite. Our initial model placed him very high in his position. But when he moved to a more structured league, everything fell apart. Not because he lacked talent, but because his environment had created a statistical illusion.
AI doesn’t become dangerous by making mistakes, but by not understanding the context.
That was a turning point for us. We realized the model needed deeper weighting. We rebuilt the engine so that environmental distortion is now one of the first things the system checks.
The lesson was simple: AI doesn’t become dangerous when it makes mistakes; AI becomes dangerous when it doesn’t understand context. That failure made Comparisonator stronger, more cautious, and far more adaptable.
Q. You’ve worked in undervalued markets and in top leagues. What common pattern do you find in players who adapt best when they make a sudden competitive leap?
A. Across all continents, the players who adapt best after a major competitive shift share the same pattern: they learn quickly, not just because they’re fast players. The players who succeed are those who can recalibrate their habits almost immediately when the environment changes.
Q. More and more clubs are looking for the next Haaland before he bursts onto the scene. Is it realistic to think that AI can anticipate generational talent, or are we still chasing unicorns?
A: AI can identify extraordinary patterns early on, but it can’t create a Haaland. Generational talent isn’t predicted; it’s confirmed over time. What AI can do is recognize the signs that typically appear before a major leap. A unicorn becomes a unicorn because of its environment, training, personality, and mindset, not just metrics. AI finds the potential. Human scouting finds the destiny.
Q. After 20 years in football and technology, what uncomfortable truth do you think the scouting industry needs to hear if it wants to take the next step?
R. That most clubs don’t have a scouting problem, but a decision-making problem. Clubs gather tons of reports, videos, statistics, and opinions… but when push comes to shove, many still make decisions based on emotions, politics, hierarchy, or panic.
The next step isn’t more data. It’s more discipline in how decisions are made. And that’s precisely why we created Comparisonator: not to replace scouts, but to force decisions to be clearer, fairer, and harder to manipulate.
Click here to read the original interview: https://www.marca.com/futbol/2025/11/18/tarkan-batguen-datos-e-intuicion-chocan-hay-elegir-hay-investigar.html
