Ok, that's a ridiculous way to look at the projections. Off the top of my head, I picked San Francisco, Seattle, New England, Baltimore, Houston, and Atlanta at various times through the season, as well as Chicago and Tampa Bay to at least make it to the game. Here's a pretty graph of Baltimore's odds at the end of each week.
|Log Scale, because I can|
The lowest point for the Ravens came as the playoffs started, when I had them with a 1.692% chance of winning. At this point, there were only 12 teams, so the mean would have been 8.333%. Baltimore definitely beat the best to get to the Superbowl, as I had New England and Denver with a combined 83% chance of making it.
Finally, I picked San Francisco about 2 to 1 over Baltimore once the matchup was determined.
All in all, either Bill James' ideas or my implementation hasn't worked out the way we expected for the last 2 years. Of course, I can still maintain that for a large number of Superbowls in parallel universes - maybe some with power outages, some without, some with Beyonce wardrobe malfunctions - that San Francisco would likely win about 2/3 of them. But in ours, we got the less likely outcome. That's probably more fun, though I'd like for it to be more fun and the surprisingly successful team be one I like, such as the Rams, or even the Steelers.