Tiago Rodrigues

It was during a university activity that I first encountered ScoutDecision. That day, Miguel showed me that even someone with a background seemingly distant from football (yes, we are engineers) could have a significant impact on this beautiful game. From that moment on, I was like a kid with a dream—one fueled by an obsession with data. (By the way, Moneyball is a masterpiece!) My goal for my master's thesis was clear: to identify technically gifted players in lower leagues who could perform at the same level as those in top-tier competitions, allowing the data to express itself and define an efficient frontier. Scouting is an incredibly complex process. First, because there is no single “correct” way to evaluate players—each scout has their own methodology and beliefs, meaning the same player can be evaluated differently by different scouts. Second, because there is no golden formula that defines talent. Talent is an abstract and multifaceted concept, making each player unique and difficult to assess with absolute certainty. Lastly, scouting is highly subjective; sometimes, it’s hard to put into words what we see (that unexplainable “coach’s eye” feeling). My mission was to address some of these challenges by structuring the scouting process in a clear and objective way—helping clubs and stakeholders rationalize their convictions and support decision-making. To achieve this, I developed a Multi-Criteria Player Performance Model—a starting point for real-world observation tasks: DataScouting. But I quickly realized that scouting isn’t one-size-fits-all. The needs and preferences of each club vary based on their vision, playing style, and desired player characteristics. That’s why incorporating the preferences of decision-makers (Chief Scout, Sporting Director, or Head Coach) into the model became essential. This approach allowed me to: 1. Identify the group of criteria most valued for each position and/or function based on the team’s context. 2. Move away from the idea that all criteria carry equal weight—some attributes matter more for certain positions and roles. 3. Acknowledge that criteria aren’t isolated—some attributes create synergy, enhancing overall performance. 4. Make the process transparent for all stakeholders, establishing a clear and intuitive player profile for each position. Nothing would have made me prouder than implementing this project at my club — the one that shaped me — Sporting Clube da Covilhã. The first step? Adapting the model to the club’s context: What is the club’s vision? Where do the players typically come from? The next step was data collection. The club had never had a formal scouting department. Given the widespread success of the Football Manager database in global recruitment, I used its data as a foundation for my process. The team’s priority was reinforcing the forward position, leading to the definition of two distinct profiles: a fixed forward and a mobile forward. Considering the club’s financial constraints, the target players came from the fourth division (Campeonato de Portugal). However, to establish a benchmark for quality, I also gathered data on forwards from the second and third divisions (Liga 2 and Liga 3). The decision-maker selected the group of criteria for each scenario and ranked the criteria they valued most for each forward profile. Then, they identified synergy effects between attributes and adjusted the scales accordingly. The results? • Fixed Forward: Out of 229 evaluated forwards, 38 were on the efficient frontier, but only 7 played in the Campeonato de Portugal. • Mobile Forward: 57 forwards were classified as efficient, with 18 coming from Campeonato de Portugal. • Only one striker was common to both profiles. This group of identified efficient players was then referenced for real-world scouting task. The forwards who were not on the efficient frontier were also evaluated, allowing us to identify the criteria in which they still had gaps compared to those on the efficient frontier. Additionally, ranking the inefficient forwards helped assess how close they were to reaching the efficient frontier, providing insights into their potential development. At the end of the project, the model also served as a validation tool for the scouting process. Players identified by the decision-maker were tested against the model: two out of three were classified as efficient, while the third ranked seventh, very close to the efficiency threshold. This provided data-driven confirmation of the decision-maker’s assessment. Ultimately, a scout is more than just a talent spotter—a scout is a decision facilitator. After all, if an engineer could lead Portugal’s national team, maybe there is room for engineers in football.