Game Outcome Win Probability Forecasting on NFL regular-season play-by-play 2015-2024 (2024 test)
-0.0079Delta (Brier)Direct − Masked-CoT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Direct − Masked-CoTn=5185, resamples=10^42026.06 | -0.0079 | -0.0101 | |
| Direct − DeepSeek-V4n=5185, resamples=10^42026.06 | 0.0005 | -0.0019 | |
| DeepSeek-V4 − Marketn=5185, resamples=10^42026.06 | 0.0083 | 0.0064 | |
| Direct − Marketn=5185, resamples=10^42026.06 | 0.0088 | 0.0068 | |
| Masked-CoT − Marketn=5185, resamples=10^42026.06 | 0.0167 | 0.014 |