I'm confused by the champ weekend percentages. For example, Thunderbirds and Union still both control their own destiny for a playoff spot whereas the Radicals are (nearly) eliminated from contention already at 1-7. How is Madison ahead of both in %?
Great question David. More details on how those percentages are calculated can be found in my first post "Chase for Championship Weekend", but in short, the model used to calculate the probability of making Championship Weekend doesn't at all know the records of the teams. Instead, the model only knows how teams have performed in games offensively and defensively on average. I believe the Radical's % is above the Thunderbirds' and Union's because of how "well" Madison played in the games they lost relative to the way the Thunderbirds and Union have played this season.
But, you point out a major flaw in the model -- it cannot eliminate teams from Championship Weekend based on what is actually possible record-wise. However, I think that flaw is also a unique strength to the model in that it's focused solely on in-game performance regardless of the outcome of the game.
Gotcha, ya I actually saw that post after making my comment, makes sense. But really, since the Radicals are essentially eliminated from contention, I find it less useful since it shows them as third highest chance of making Champ weekend, even though that clearly doesn't pass the smell test.
Super interesting content so far, thanks for producing this!
I'm confused by the champ weekend percentages. For example, Thunderbirds and Union still both control their own destiny for a playoff spot whereas the Radicals are (nearly) eliminated from contention already at 1-7. How is Madison ahead of both in %?
Great question David. More details on how those percentages are calculated can be found in my first post "Chase for Championship Weekend", but in short, the model used to calculate the probability of making Championship Weekend doesn't at all know the records of the teams. Instead, the model only knows how teams have performed in games offensively and defensively on average. I believe the Radical's % is above the Thunderbirds' and Union's because of how "well" Madison played in the games they lost relative to the way the Thunderbirds and Union have played this season.
But, you point out a major flaw in the model -- it cannot eliminate teams from Championship Weekend based on what is actually possible record-wise. However, I think that flaw is also a unique strength to the model in that it's focused solely on in-game performance regardless of the outcome of the game.
Gotcha, ya I actually saw that post after making my comment, makes sense. But really, since the Radicals are essentially eliminated from contention, I find it less useful since it shows them as third highest chance of making Champ weekend, even though that clearly doesn't pass the smell test.
Super interesting content so far, thanks for producing this!