Academy Selection Simulator
Every young player a club's scouts have rated across their region over several seasons: 5,000 of them, each scored on technique and on athleticism, with no link between the two. Over those seasons the academy takes the best. Among the players it picks, the more technical tend to be the less athletic. Change how the academy picks and see where that link comes from.
- Picked
- Not picked
- Trend among the picked
- Trend across all 5,000
- Pearson's r, all 5,000
- Pearson's r, the picked
Try an example
How to read it
- Each dot is one player. Further right, the better his or her technique; higher up, the more athletic. The centre lines are the average player.
- The bright dots are the players the academy picked. The faded ones missed out.
- The dashed line is the trend across everyone. Flat means no link: knowing a player's technique tells you nothing about his or her athleticism.
- The gold line is the trend among the picked. Sloping down means the more technical the player, the less athletic, but only inside the academy.
- Nothing about the players changes when you move the controls. Only who gets picked does.
How it works
Each player gets two scores drawn at random, technique and athleticism, so across all 5,000 there is no link between them unless you set one. The academy then picks:
- Combined score: the top players on a mix of the two, weighted as you choose. A player can make up for weak technique with real pace and strength, or the other way round.
- Technique only: the top players on technique, athleticism ignored.
- A minimum in both: everyone above the same bar on each, set so about as many players pass as there are places.
The two numbers are Pearson's correlation coefficient, \(r\): one across all 5,000 players, one across only those picked. 0 is no link, −1 a perfect "more of one, less of the other" and +1 a perfect "more of both" (glossary). The lines are the least-squares trend through each group. With 200 players, an \(r\) within about ±0.14 of zero is what luck alone gives (two standard errors, \(2/\sqrt{n}\)), so the line under the numbers calls that no clear link.
What it shows
Only the combined score creates the link. When one strength can cover for the other, every picked player with modest technique has to be very athletic, and every picked player with modest athleticism has to be very technical. That fills the top-left and bottom-right of the picked group and leaves its bottom-left empty, and the trend slopes down. Picking on technique alone, or demanding a minimum in both, lets nobody trade one for the other, and no link appears. The pickier the academy, the stronger the false link: about −0.47 for the top 2,500, −0.78 for the top 200 and −0.82 for the top 50, on average over many intakes. Set a real link of +0.3 and the top 200 still come out at about −0.62: the selection can turn a link round, not just invent one.
Where it goes wrong
- The players are made up. Real academies weigh more than two things, and real technique and athleticism may well be linked; the point is what selection does to whatever link there is.
- Each intake is random, so the numbers move a little with "New intake". The figures above are averages over 60 intakes.
- It's the share picked that matters, not the totals: 12 picked from 300 rated gives the same link on average, it just wobbles more from one intake to the next.
The article behind it: Are gifted players less athletic? Collider bias and the academy
These are simulated players for analysis and education.