Predictions & Results
Every model pick, grouped by match day. Times shown in Indian/Mauritius.
Market Radar
Upcoming fixtures with model probabilities and bookmaker line movement, grouped by match day. Click a match for odds and details.
xG Standings
League tables next to expected goals (xG) and expected points (xPTS). Teams well below their xPTS have been unlucky; teams well above have been riding luck.
| # | Team | P | W-D-L | GF:GA | PTS | xG | xGA | xPTS | PTS − xPTS | Corners/G | Big ch. | Box touches | YC |
|---|
How to read this: xPTS is the points a team would expect from the quality of chances it created and conceded. PTS − xPTS of −3 or worse is marked Unlucky (results may improve); +3 or better is marked Overperforming (results may cool off). Click a team for its home/away split and next fixture.
Archive
Completed matches with full-time scores, post-match xG, corners and model pick results, grouped by match day.
How to read this: xG (H–A) is post-match expected goals from shot quality. When xGPANDA had a pick on a completed match, the WON or LOST badge links it to the Predictions & Results record. Click any row for full match stats.
Model Audit
How closely the model’s pre-match probabilities match real full-time outcomes across completed league fixtures.
Benchmark comparison (1X2 Brier score)
Shorter bar = more accurate probabilities. Lower Brier score is better.
Reliability by probability bucket
When the model predicts 60%, home teams should win roughly 60% of the time.
| Predicted bucket | Matches | Model expected | Actual win rate | Delta | Status |
|---|---|---|---|---|---|
| Loading calibration data… | |||||
A Brier score measures how close probability forecasts are to actual results on a scale where 0.00 is perfect and 0.2222 is blind guessing (33.3% home / 33.3% draw / 33.3% away). Beating a basic Poisson model (~0.212) means team xG and form adjustments add real signal; ~0.189 is the benchmark set by sharp closing bookmaker odds.
Historical fixture probabilities in the database were computed using season-to-date standings rather than strict pre-match snapshots, which flatters in-sample scores. Forward pre-match picks are logged separately in the prospective audit ledger before kick-off.
Reviews & Improvement Ideas
Share your honest experience with xGPANDA.AI or suggest leagues, features, and improvements we should build next.
Leave a Review or Suggest an Improvement
Did a pick land for you, or do you want a new league, filter, or UI improvement? Tell us below.