Go far, but go together
What I like most about tech is the people: from the mailing lists that led to Bitcoin to a casual chat with a friend that may become my master's thesis.
Log
everything written, newest first
Most of a match rating can't be predicted
Position specialists didn't beat a single generic model, recent form was worse than no information at all, and nobody can predict a goalkeeper. The thesis answer was no.
When R² is too good
The first model explained 80% of the variance in forwards' ratings. That was the problem: it was rebuilding Sofascore's formula, not predicting football.
A season of Sofascore data, and where it lies
Scraping every player rating from the 2025 Brasileirão, and the three traps in the data: dates in the wrong timezone, zeros that never get sent, and rounds played out of order.