The article takes an experimental approach to scouting, asking whether the output of a departing star player can be recreated through a combination of complementary data profiles.
We Can’t Replace Him
When your best player leaves, many clubs instinctively look for the next star or a like-for-like replacement. It is an understandable reaction, but it often becomes an expensive and uncertain pursuit, as almost 50 per cent of ‘big transfers’ don’t deliver the expected impact (‘How To Win The Premier League’, 2023). Clubs are not only betting on one player to adapt quickly, but they are also entering a market where every team is competing for the same “next big thing”, which inevitably drives prices up.
This raises a broader strategic question: is there a smarter way to replace departing stars and reduce the risk built into recruitment?
Inspired by Michael Lewis’s Moneyball (2003), later popularised by the 2011 film adaptation.
The image above recalls one of the most iconic scenes from the movie Moneyball, where Oakland A’s scouts debate how to replace Jason Giambi. Billy Beane cuts through by arguing that they cannot replace Giambi directly, but they might be able to re-create his value in the aggregate. Instead of chasing one expensive superstar, they focused on the key performance indicator (KPI) that contributed most to winning baseball games, On-Base Percentage (OBP). The idea was to distribute Giambi’s total OBP across several undervalued players, allowing the A’s to compete in an uneven market. No single signing was a perfect match, but collectively they managed to replace Giambi’s OBP value.
Translating this idea to football is less straightforward. Baseball is a highly structured sport with isolated events, while football is a dynamic invasion game with constant interactions. However, the underlying Moneyball principle remains the same. If we can identify a metric or a set of metrics that consistently correlates with increasing scoring probability and ultimately winning matches, then we can break down a player’s impact and examine whether that value can be reproduced across multiple roles rather than through a single high-risk signing.
Expected Threat as KPI
This is where expected threat becomes an interesting starting point. By analysing how a player advances the ball into dangerous areas, we can quantify the part of their game most closely tied to creating goal-scoring opportunities or increased danger. Since football is a low-scoring sport, the most valuable contribution to the target is a player’s role in generating chances. Once we understand the components of their xT profile, we can explore whether several targeted, cost-efficient signings can collectively reproduce that influence.
In order to not make the calculations overly complicated, we will focus on finding metrics that we believe represent progression and chance creation (essentially xT). This can also make it easier to scout and pick the metrics that certain positions should excel in inside our squad composition.
Re-Creating A ‘Star’ Playmaker
The scenario we are imagining is that our star creator leaves. This is a player who receives the ball both in the Create phase and the Finish phase. In practice, that means we’re trying to replace someone who helps progress play after the initial Build Up phase and someone who is central to our chance creation in the final third. Players who influence multiple phases like this are hard to replace. They don’t appear often, and when they do, they come with heavy price tags.
To approach this problem, we will break creative impact into clear components. The idea is to separate their progressive qualities through Forward Momentum actions, their ability to advance play through 8-metre Carries at Speed, and their Pass Attempt Share to Runs Ahead of the Ball. We then combine this with their Avg Expected Threat (xT) Per Pass Attempt into Off Ball Runs to assess which players that picks out the most dangerous passing options on average. By isolating these elements, we can understand which parts of the profile can be reproduced by different players. Later, we will also look at how we can replace their impact in the final third or the Finish phase.
Performance Index for Progression
- Count Forward Momentum Possessions In Create
- Count 8M Carries Above Running Speed (all phases)
- (Share, %) Pass Attempts to Runs Ahead of the Ball / Pass Opportunities to Runs Ahead of the Ball In Create
- Avg xThreat Per Pass Attempt to Off Ball Runs
Below are the results of our index ranking filtered for U23 Midfielders in the top five leagues from the 24/25 season.