← Portfolio
XGBoost · Flask API coming soon

FIFA World Cup 2026

Score predictions for every match across USA, Mexico & Canada — powered by gradient-boosted models trained on international form, xG, and historical fixtures.

48 Teams
12 Groups
8 Preview fixtures
Model pipeline: Feature engineering → XGBoost regressor/classifier → Flask REST API → this dashboard. Currently showing mock predictions for UI development.

Tournament groups

Upcoming predictions

Knockout stage

Tournament bracket from the Round of 16 to the Final — winners feed into the next round on the right. Tap any fixture for predictions and model rationale.

How predictions are made

Each match prediction is produced by an XGBoost model trained on international match results from 2010 onward. Input features include recent form, expected goals (xG), Elo ratings, head-to-head records, home/host advantage, and squad strength indices.

The Flask backend (in development) will serve live inference endpoints. This frontend consumes /api/dashboard and /api/matches/:id — swap mock data by setting WC_API_BASE in js/data.js.

Match detail pages show the primary scoreline, alternative outcomes with probabilities, and SHAP-style feature breakdowns explaining the model's reasoning.