02-09-2022, 04:01 PM
The mistake so many people make when analyzing Super Bowl matchups is relying on season-long body of work data and modeling. I understand the mindset -- it's a larger sample size and therefore it should be more reliable.
The problem is: models of NFL success are almost universally terrible at the level of game-to-game prediction. They're just bad. Moreover, the Super Bowl isn't won by the team that performed at the highest level throughout the season. It's won by the team that performed at the highest level in the game. Perhaps postseason stats alone provide a better gauge just for recency's sake, but even that is dubious.
Both of these teams have had to overcome very difficult opponents to get to this point. The Bengals especially have had to get through at least two that were better on paper by every meaningful metric in the book. No model on Earth said the Bengals were favored in Kansas City. We can either say "they defied probability and got lucky" or we can say "the Bengals were a better football team than the Chiefs on 30 January, 2022". To scoff at the latter is nonsense in my view.
The Super Bowl will be no different. All of the prognosticating and especially the smack talk -- it's irrelevant noise. There is no model good enough to answer the question "who will win?" with any degree of reliability, and therefore we cannot really determine what to expect until we're watching the game live.
The problem is: models of NFL success are almost universally terrible at the level of game-to-game prediction. They're just bad. Moreover, the Super Bowl isn't won by the team that performed at the highest level throughout the season. It's won by the team that performed at the highest level in the game. Perhaps postseason stats alone provide a better gauge just for recency's sake, but even that is dubious.
Both of these teams have had to overcome very difficult opponents to get to this point. The Bengals especially have had to get through at least two that were better on paper by every meaningful metric in the book. No model on Earth said the Bengals were favored in Kansas City. We can either say "they defied probability and got lucky" or we can say "the Bengals were a better football team than the Chiefs on 30 January, 2022". To scoff at the latter is nonsense in my view.
The Super Bowl will be no different. All of the prognosticating and especially the smack talk -- it's irrelevant noise. There is no model good enough to answer the question "who will win?" with any degree of reliability, and therefore we cannot really determine what to expect until we're watching the game live.
