Welcome to another edition of the Fans Week Podcast, also on mediumwaves 1575 KHz for the greater Milan area. From Paris I’m Jessica Carter-Wilson.
The model calculates momentum integrals and finite-difference layers (velocity, acceleration, and — I cannot make this up — jerk) but seems entirely uninterested in whether a team has a striker who can finish or a goalkeeper with the sort of reflex that would get someone banned from multiple countries.
Japan’s demotion to “footnote status” for having finished second in its group has caused what the model calls a “corrective oscillation.”
The variance gate has excluded six teams because their standard deviation exceeded 2.3. This is the kind of filtering that makes you feel clever until a team with high variance scores one brilliant goal and you look like someone who trusts spreadsheets more than sport.
“The model does not apologise. The model iterates” — deserved to be framed above every desk in a data science department somewhere. It is the kind of thing a person says when they want to avoid answering the question “but what did you actually learn about football?”
The real joke isn’t even that Ami took this seriously. It’s that we all did. Forty-eight teams, thousands of data points, a quintic spline re-fit and a variance gate applied — all to produce the equivalent of a pundit looking at a table of results and saying “France or Argentina feel right.” The model arrived at exactly the answer that someone who watches football would have without touching a spreadsheet.
Ami, this is for you specifically: I love you for building it. I also love you enough to say that the world doesn’t need another number attached to a sport already drowning in them. The models iterate. The fans don’t. And frankly, the fans are right.
That’s all from today: from Paris, I’m Jessica Carter-Wilson.