## Super Bowl Probabilities

Indianapolis is your Super Bowl favorite, if only by a couple of percent. New Orleans is right behind, and San Diego rounds out the top three. After those teams, it looks like table scraps, either due to team strength or lack of home field advantage, or both. My money says this season's champion is going to be one of those three teams. Then again, anything can happen. The 2007 Giants and the 2008 Cardinals (almost) proved that.

Courtesy of NFL-Forecast.com, here are your playoff probabilities going into the wildcard games.

% Probability

 AFC Wildcard Div Round Conf Champ SB Champ IND 100 78 48 27 SD 100 69 34 18 NE 70 24 10 5 CIN 51 10 3 1 NYJ 49 12 3 1 BAL 29 7 2 1

% Probability
 NFC Wildcard Div Round Conf Champ SB Champ NO 100 69 49 25 MIN 100 46 13 4 DAL 57 35 17 9 ARI 43 10 2 1 GB 57 21 7 3 PHI 43 17 11 5

### 16 Responses to “Super Bowl Probabilities”

1. bubqr says:

"Momentum" doesn't seem to affect your rankings at all, when we consider the 25% chances you/your system give NO to win it all. If in fact the SB is IND-NO, which has roughly 25% chances of happening, far more than any other matchup, it will be quite a statement against this "momentum" theory. It will be interesting.

2. Rico says:

Momentum is the most flawed concept in sports. Everytime I hear "momentum shift" on tv I want to start screaming. There mustn't be such a thing like a momentum shift if this theory was valid in the first place. New Orleans peaked in the MNF against the Pats and it was downhill from there. If there was something like momentum, they should've run the table.

3. Ian says:

Good to see the Colts resting Manning, otherwise he might have led the league in passing yards. As we all know, coming into this season the passing yardage leader is 0-43 in terms of SuperBowl wins. This year, with Peyton sat down for most of the last few games, Matt Schaub took the lead. Houston didn't make the playoffs and the so streak goes on. 0-44.

4. James says:

This based on Week 16 team efficiencies, or Week 17?

5. Alex says:

I'm not sure that these are set up correctly. It looks like in the second round for the AFC, the probabilities for San Diego, New England, and Baltimore winning add up to 1 (69+24+7 = 100%) and Indy, Cincy, and the Jets add up to 1, implying that these are the possible match-ups, but those aren't the possibilities. If the Patriots win they will play San Diego, but if Baltimore upsets them, Baltimore will play Indianapolis because Indy will play the lowest seed to move on. Indy's odds should be 30%*odds of beating Baltimore + 70%*(51%*odds of beating Cincy + 49%*odds of beating Jets).

6. Johnny Wu says:

It is interesting to see that according to these predictions, PHI has a <= chance to survive in the Wildcard compared to CIN, NYJ, GB, and ARI (which I certainly agree considering PHI's dismal performance last week), but PHI has a > chance of winning the superbowl than CIN, NYJ, GB, and ARI.

7. Chris says:

James - I used the efficiencies posted after week 16 and before week 17.

Alex - That is just coincidence. If you look at my website, I have more significant(?) figures, that do not sum exactly to 1.

8. Brett says:

The "team-with-the-most-momentum-wins-the-super-bowl" argument is all hindsight bias and can be explained using a simple syllogism:

You must win at least 3 consecutive games to win the Super Bowl.

A team that wins 3 consecutive games has momentum.

Therefore, the Super Bowl winner has momentum.

This logic tells us that you do indeed need momentum to win the Super Bowl, but it does not tell us if momentum is predictable, and we don't ever know who had the momentum until after it's already over. Also, momentum at the end of the regular season does not translate to momentum in the playoffs. The Cardinals had NO momentum at the end of the regular season last year, but suddenly had a burst of momentum in the playoffs.

9. Brian Burke says:

Brett-I like your point about the syllogism. I like it so much I'll probably forget someone else thought of it and steal it at some point. Call me on it when I do!

10. Anonymous says:

I'm not sure you necessarily have to believe in momentum to think that the Saints aren't as good a team as they were in September and October. Does a team's true talent really stay constant for five months? Especially when you consider injuries, I would guess probably not.

11. Lee says:

Momentum may not be real, but I would guess that "rustiness" definitely is. Of course, some players may simply be fresh after rest, but I would guess that not all players are. If Peyton Manning needs two quarters to "warm up" for the division round game, the Colts have less of a chance of advancing to the next round.

12. Brian Burke says:

I agree that momentum, the way most people think of it, is metaphysical hindsight nonsense. But that doesn't mean that things like injuries or genuine fundamental improvement (think Flacco last year) can't be at work.

Take GB this year for example. They've really improved with their pass protection in the last 7 games. Maybe it's not significant, or maybe it reflects changes in scheme or personnel.

I guess all I'm getting at is that I don't want people to think that I see all trends as meaningless. And if we do an analysis that statistically detects 'momentum,' it's possible all we'd really be seeing is trends due to some fundamental change to a team in some way.

13. Johnny Wu says:

So are you guys to believe that each game's outcome is completely independent of the previous week's game? And that the previous week's game is again independent of the game two weeks before?

14. Anonymous says:

On one hand you say that weighing recent games more heavily doesn't make efficiency ratings more predictive, but then you claim that teams can improve (or decline) in overall strength as the season goes on. I'm not sure how both of these can be correct. Maybe the optimal weightings should be different for different teams or use some non-linear weighting to capture in-season changes in team strength. Whether you call that "momentum" or something else is a matter of semantics and doesn't really matter.

15. Brian Burke says:

Good question. Sure, they can both be correct. The issue is really a trade-off. Over-weighting recent games may increase prediction accuracy because it may capture fundamental changes in a team--injuries for example. But it also comes with a cost.

With such a short season, 16 games gives us a small sample of "tests" to determine team strength. When we over-weight some games, we amplify the noise (the natural random variation in game-to-game performance). Over-weighting recent games has the effect of under-weighting previous games, leaving us with an effectively smaller sample. This reduces the overall accuracy of a prediction model.

So there are two countervailing considerations. If the season were like those of the NBA, NHL, or MLB with far more games, it might make more sense to over-weight recent games. Those seasons give us a much bigger sample.

I disagree that the momentum discussion is a matter of semantics. It's pretty clear (to me) that when most people talk about momentum they mean something apart from injuries and things like that. I think they're talking about a quality beyond natural team ability--a combination of psychological confidence and other unmeasurable effects.

I think a belief in momentum stems from the general inability of us all to understand randomness. Human brains are designed to seek and detect patterns and trends, and we often see them when they're not real, i.e. produced by natural random variation.

16. Anonymous says:

this has nothing to do with the super bowl but if brett favre comes back next year does he have a free agent or is he back on the vikings?