Playoff scenarios

Team ratings and quarterbacks

Raise or lower a team’s rating (points better or worse than an average team on a neutral field) if you think the market has it wrong, or take its starting quarterback out for the rest of the season. The quarterback figure is the model’s estimate of the drop to his backup (or to a replacement-level quarterback when he has started every game).

Ratings and quarterbacks don’t count while every game is a coin flip.

How the tool works

• The same model as the published odds: team ratings from the betting market with their uncertainty, the posted lines for this week and next, realistic scores and ties, and every NFL tiebreaker. Your browser plays 20,000 seasons each time (the published odds use 100,000); “More seasons” plays 100,000.
• A picked game is played with the result you chose (its score taken from real games with a similar line and that result), and the ratings react to it as the market reacts to any result. A pick changes nothing else: a team you pick to win in week 17 doesn’t become a better team in its other games.
• Changes are measured, not guessed: each simulated season is also played without your changes, with the same random draws, and the difference is added to the published odds. So a change shows its own effect rather than simulation noise, and with no changes you see the published odds.
• A chance you set replaces the model’s chance for that game (ties stay possible, about one game in 300). A rating change moves the team’s rating today; like every rating in the model, it drifts slightly back toward average over the weeks. A quarterback out counts in every remaining game, the playoffs too.
• Every game a coin flip shows what the standings and the schedule alone imply, as FanGraphs and the New York Times have done.
• The pages’ “if … wins” numbers are computed exactly like a pick here (the game played with each result, the same 20,000 seasons), so picking one game reproduces them. Games that matter in your scenario covers this week’s games: it compares the simulated seasons in which each result happened, which for this week’s games matches a pick (for games further away it would also reflect how strong the teams turned out to be).
• Final standings: with every game picked, the standings come straight from our standings engine, which reproduces every official seed and division rank since 2002. Ties that last past strength of schedule depend on points, so the simulated seasons decide them with realistic scores.
• Nothing is sent anywhere: the simulation runs in your browser. Your changes are kept in the page’s address, so you can bookmark or share them.