## Live Super Bowl Probabilities

They're back. Details here.

LIVE SUPER BOWL PROBABILITIES

### 11 Responses to “Live Super Bowl Probabilities”

1. Anonymous says:

New England with a 9 percent chance to win the super bowl? I'll take those odds.

2. Anonymous says:

This site has been pretty hard on the Patriots. Much harder than footballoutsiders, which thinks they were a top 5 team. Not sure why.

3. Anonymous says:

They are using a weighted efficiency which arbitrarily drops older games I believe. For example they had the Pats getting much better on a week that they didn't even play.

4. Unknown says:

It's not arbitrary. Recent performances have more predictive power.

5. Anonymous says:

Maybe depending on a number of other things like injuries. And its arbitrary by choosing the say the last 4 games instead of the last 5, this probably results in over fitting. A team that doesn't play shouldn't have any change in its efficiency as there is no new information.

6. Anonymous says:

Just because a team isn't playing doesn't mean that nothing new is happening. Suppose for example that Peyton Manning broke his leg in a car crash during the Wildcard round and was IRed for the rest of the playoffs. Even though Denver and New England had byes that week, I'd say this would significantly change both team's super-bowl odds.

7. Anonymous says:

Uhm, you don't really understand FO stats, huh? Since the stats are adjusted for opponent, a bye week can very well give us significant information about the strength of a team, because we get new information on the teams they already played.

8. Unknown says:

I don't know what FO specifically does, but what Wayne Winston describes in "Mathletics" is giving the most recent week weight 1, the week before that weight .95, the week before that weight .95^2, etc.

This removes the arbitrariness of choosing exactly 4 games.

9. Anonymous says:

16 games in a season is already a ridiculously small sample size. Filtering for more recent games seems even more so.

10. Anonymous says:

The probability is actually always almost always predetermined, but too many unknown, ultra sensitive, correlated variables for prediction to ever be near 100% accurate.

In other words... It is like the weather, chaotic, but predetermined

11. Mike says:

The notion that football is mostly luck is an absolute joke. Instead you have the results of very small differences in variables playing a huge role because the skill is so evenly matched that something as simple as concentration level of a WR on a single play can be enough to make or break the game.

So instead you have the illusion of luck because there are so many variables that we have not learned how to predict and because so much uncertainty exists as observers. Assigning probabilities is only attempt to handicap the data based upon what we do no and discounting it against that which we do not.

Putting it any other way is a misunderstanding that statistics is not the study of "chance" but instead the study of past data in attempts to assess the future. It is a linear assessment of non-linear systems.