Coverage Explorer
Which game each US market gets in every Sunday afternoon window: what aired, the networks' own maps once they are set, and our prediction for the rest.
Loading the maps…
Which game each US market gets in every Sunday afternoon window: what aired, the networks' own maps once they are set, and our prediction for the rest.
Loading the maps…
Which game each US market gets in every Sunday afternoon window: what aired, the networks' own maps once they are set, and our prediction for the rest.
On the 2025 games our model never saw, the game we expected aired about 70% of the time.
On the 2025 games our model never saw, the game we expected aired about 70% of the time.
We use regularized logistic regression, not an LLM, to learn from past broadcast choices. Inputs combine geography, home-market ties, broadcast windows, historical team draw and preseason strength. The highest-scoring competing game becomes the pick. Scores are adjusted against past results, then normalized within that window to estimate confidence. While games can still move, percentages allow for one moving to primetime and the game it displaces coming back, not for early-late kickoff swaps.
Map shading strengthens at 45%, 60%, 80%; these are display bands, not broadcast guarantees. The accuracy note above checks an unseen season. Network-set maps remain distinct from predictions. What-if controls change inputs, not the trained model.
Sportsbook win totals stand in for how good the teams look until FTN's DVOA ratings arrive.
Predictions weigh each market's distance to the teams, home-market rules, the broadcast window, season projections, and how big a draw each team is, over 2025 · 2024 · 2023 · 2022, with recent seasons counting for more. What aired, and what the networks have set, is read from 506sports' weekly coverage maps. To replay a finished season week by week and see what the advice would have been worth, run it back through the Season Simulator.
Showing all 3 windows this week.