Ice_Ice_Brady

The dictionary defines "favorite" as "a competitor judged most likely to win." That said, most analysts consider the Patriots to be favorites to win the Super Bowl since their odds are higher than any other team.

Most simulations have the Patriots winning the Super Bowl in 20-25% of simulations. When forecasting the season using really smart computers, the Patriots ultimately fail to win it all in 75-80% of simulations, the lowest percentage in the NFL. If the Patriots were to have a 20-25% chance of winning a given football game, they would likely be be a 10+ point underdog. In other words, the Patriots chances of winning the Super Bowl is roughly equivalent to the Bills beating the Patriots tonight.

In a simulation of "The Patriots vs. The Field" (where the field is the other 31 teams), the Patriots are not likely to win. Does anyone really think they should be called "favorites" to win the Super Bowl?

Personally I consider a team to be "favored" if their odds are better than 50%. I'm tired of hearing about how we are the "favorites." Bottom line is, every team is the underdog going into the season.

Just a thought about expectations....

Another interesting note.

When Tom Brady went down after game 1 last season, the Patriots went from winning the Super Bowl in 30% of simulations to winning it 5-6% of simulations. When Richard Seymour was traded last week, the simulations were barely changed. So much for the idea that Seymour is the "Brady of the defense." The two players that are far and away the most valuable are Brady and Moss. Welker would be third, and then it's a gigantic dropoff.

RobAllan

1. Where are all these numbers you're using coming from?
2. Who is the "favorite" based on the simulations you're referencing?
3. What criteria was used for the players in the sims / how were teams weighted? How did strength or weakness of schedule figure into the percentage odds? (I personally think strength of schedule is a useless stat until about mid season because teams are NEVER the same as we think they are when the season starts)

Just some thoughts/questions .

-Rob-

Ice_Ice_Brady

Just basic observations with Accuscore and a lot of Vegas books. Nothing scientific about the data, just about the way the data is interpreted by the media and fans.

Spiral

The way the term is applied to contests with multiple contestants, it's relative to each of the other teams, not all of the other teams. By your definition, unless the "favorite" has less than even money odds, that team is not a "favorite."

PatsFanSince74

that's a big question, but first, thanks for the information and as a bit of a stats admirer, i'd appreciate a link to those simulations.

i'll be honest and say that I am very skeptical of the value of trying to simulate an entire NFL season, unless the model is extremely complex to the point where one would need a supercomputer to run it a meaningful number of times. To me, the Vegas line is a much more reliable indicator of who is going to win the SB, since it incorporates literally thousands of years of experience of observers who are putting their money at risk.

At the end of the day, I doubt very much that even so complex a model would explain any more than the Vegas line.

As to the difference between predictions of individual games and the SB, think about it like this. using your 10-point underdog analogy, saying that there is a 20% chance of the Pats winning the SB is kind of like saying that, on September 1st, they are the equivalent of a ten point single game underdog to win the SB when the comparison for the Steelers might be being a 10.5 point single game underdog, for the Colts a 13 point underdog all the way to the Lions as a 50 point underdog.

another way to look at it is to take a common theme on this board, to the effect that the Patriots will be "favored" to win every game they play this season. if the patriots were a 3--1 favorite to win every game then they would win 3/4 of their games over an indefinite period of time; so, if you ran a million simulations using the 3/1 odds, then the median would be a 12 win season; but, some of those simulations would yield both 0 and 16 win seasons. being favored to win every game we play says very little about how many games we actually will win.

As to the impact of losing Richard on the models, it all depends on how the data are put in and how they are weighted. i could build a model that factored in the identity of the Gatorade guy on the sideline and, properly weighted, his identity could determine the winner. More realistically, if you erected a model that overweighted the D, i'm sure the impact would be greater.

__________________
It is what it is. It wasn't what it wasn't.

MoLewisrocks

It means nothing.

PatsFanSince74

yeah, that too.

__________________
It is what it is. It wasn't what it wasn't.

PatsFaninAZ

I think in gambling terms, the word "favorite" largely as evolved from horseracing, where the "favorite" is the horse most likely to win, even if he or she is less than 50/50 to win versus the field.

So I'd say that in common parlance and given its horseracing roots, the patriots are the favorites. I also think it meets the dictionary definition -- the patriots are the competitor most likely to win.

If you ask what "America's favorite color is," I would think that "blue" would be a correct answer even if the fact is that 49 percent say they like it best, as opposed to 51 percent.