Apex
Model performance

Analytics

Calibration, Brier loss, and training records across MLB, Tennis, and NFL models.

updated 11d ago
MLB Brier
stable
0.2469
0.12ExcellentMarketRandom0.27
21,458 train·val 20152023·season 2024
Hierarchical pyramid + MagNet GNN + BayesFlow
TEN Brier
strong
0.1751
0.12ExcellentMarketRandom0.27
15,801 train·val 20182023·season 2024
Tennis encoder + MagNet GNN + BayesFlow
NFL Brier
stable
0.2247
0.12ExcellentMarketRandom0.27
4,378 train·val 20152023·season 2024
NFL encoder + MagNet GNN + BayesFlow
Calibration reliability
predicted vs observed
No calibration data yet

The reliability diagram will populate once the backfill pipeline has labelled games to compare against.

Brier reference
excellent · market · random
MLB
0.2469
0.12ExcellentMarketRandom0.27
TEN
0.1751
0.12ExcellentMarketRandom0.27
NFL
0.2247
0.12ExcellentMarketRandom0.27
Live tracker

Betting performance

No tracker data yet

No settled predictions yet. Tracker populates after record_prediction_outcomes archives the first completed day.

Slate Ledger
12 rows
APR.17.26
tennis
Games
0
Avg
Max
Act
0
APR.17.26
nfl
Games
0
Avg
Max
Act
0
APR.17.26
mlb
Games
14
Avg
51%
Max
+57.2%
Act
14
APR.14.26
tennis
Games
0
Avg
Max
Act
0
APR.14.26
nfl
Games
0
Avg
Max
Act
0
APR.14.26
mlb
Games
15
Avg
52%
Max
Act
0
APR.13.26
tennis
Games
31
Avg
52%
Max
Act
0
APR.13.26
nfl
Games
0
Avg
Max
Act
0
APR.13.26
mlb
Games
10
Avg
51%
Max
Act
0
APR.12.26
tennis
Games
0
Avg
Max
Act
0
APR.12.26
nfl
Games
0
Avg
Max
Act
0
APR.12.26
mlb
Games
0
Avg
Max
Act
0