A special announcement will replace today’s recap and the usual Friday preview:
the SFM website has undergone a significant revamp
We have updated both the website and the model itself.
Spoiler Alert 🐐
Based on more than 300,000 player-level observations, the SFM has updated its GOAT ranking!
👉 All discussions settled now!?
SFM Website Update
The website now appears in a completely new design, has undergone significant performance updates and comes with many additional features:
🥅 Goal Scoring Probabilities
Real-time and historical probabilities of scoring a certain number of goals in a given game for more than 1,000 players.
sfmG
A new metric, quantifying a player’s expected goals per game as predicted by the SFM.
Leaderboard
A leaderboard, ranking players based on their observed goals per game and showing the corresponding sfmG.
👉 SAR & PAR
Plots and rankings for Skill-Above Replacement (SAR) and Performance-Above-Replacement (PAR) as estimated by the SFM.
SFM Model Update
Larger Universe of Players
We have significantly increased the universe of players in our estimated sample, and re-estimated the model accordingly.
To motivate a finance analogy: the previous version was estimated on the DOW 30 – and even some vintage version of it. The new model has now been estimated on an extended universe comprising the S&P 500, NASDAQ, and even Russell 2000.
Who is the GOAT?
With the re-estimation on a larger universe of players, the SFM has updated its ranking of the GOAT!
And here, the beauty of the Bayesian framework speaks for itself: Erling Haaland and Kylian Mbappe entering the Top 5...but how confident can we be in that estimate?
Well, see and judge for yourself...
Want to learn more?
Reach out to us directly, or contact us via the website.
And for the geeks: read the Paper: https://arxiv.org/abs/2412.05911
#socceranalytics #pymc #sportsanalytics #bayesian #statistics #soccer #football #prediction #forecasting #sfm #factormodels #alpha