The first two matchdays are in the books and the last round of the group stage is already underway. Time for an honest assessment on how our SFMMO model has fared over the first 48 games of the tournament!
SFMMO Validation
That match outcome prediction is a tough field and that bookmakers pose a very competitive benchmark is no secret. Still, we’re here for the fun and the excitement of the challenge!
The stats at a glance for calling correctly W/D/L:
- 📣 SFMMO – Accuracy
- 52.1%
- 📚 Bookmakers – Accuracy
- 64.6%
- 📣 SFMMO – Ranked Prob Score (RPS)*
- 0.190
- 📚 Bookmakers – Ranked Prob Score (RPS)*
- 0.152
- 🎲 Coin-flip – Ranked Prob Score (RPS)*
- 0.229
- *RPS
- lower is better.
The verdict. No sugarcoating: the performance so far has been quite disappointing. The SFMMO has called only 25 out of 48 results correctly, trailing the bookmakers by 6 matches.
Furthermore, our SFMMO put on average too much weight on the wrong outcomes, as documented by the RPS – the MSE for probability forecasts.
🤓 To find out more, check out our validation page, where we show results for each of the 48 matches: https://www.soccerfactormodel.com/world-cup-2026/validation
Despite a performance of the SFMMO that is so far rather akin to Belgium than to Messi, still:
Follow the SFMMO
🌍 On our website: https://www.soccerfactormodel.com
📱 On our app (only Apple iPhone): https://apps.apple.com/app/id6758757755
X On twitter @SFM_Predicts
#socceranalytics #football #prediction #forecasting #sfm #factormodels #alpha #assetpricing