Chances for a returning superbowl champion to repeat

[bmorepunk]

Have they stated the chances of a team winning the Super Bowl is even across all 32 teams?

 

[Sarlacc83]

Have they stated the chances of a team winning the Super Bowl is even across all 32 teams?

No. But what you said about not being able to reasonable predict the odds early played into my response. FOs model is reasonable, given that they gave Seattle a 20% or so chance to win the Super Bowl last year, which was quite high and as it turns out, pretty accurate.

Point being, there aren't an indefinite amount of variables. The key is finding the meaningful statistics in the noise.

 

[Cartire]

Pretty sure the only correct answer is 1/32. Just saying.

No.

Please elaborate.

The NFL season is not a statistically randomized process. If it were, the probability would be 1/32. But it's not, it is indeterminate.

Actually, at the start of the season. IT is 1/32. All teams are considered equal at this point, records are even across the board. No single schedule is technically harder yet. Analysis utilize past seasons for predicting strength of schedule, along with past personal stats and observations.

But currently, at the start of the 2014 season, all teams are even. Regardless if we assume the Raiders will suck again or what not.

As of right now, its 1/32.

 

[Ad Hawk]

So we're really discussing the difference between chance and odds?

If what you say, Cartire, is absolutely true, then believing we have a "better chance" than other teams is simply horse-manure, in spite of our team's previous success, players, coaches, homefield advantage, etc., right? I'm just clarifying.

 

[Cartire]

So we're really discussing the difference between chance and odds?

If what you say, Cartire, is absolutely true, then believing we have a "better chance" than other teams is simply horse-manure, in spite of our team's previous success, players, coaches, homefield advantage, etc., right? I'm just clarifying.

Correct. At this current junction. However, what you guys are trying to make a point of is basically our standard practice of odds. We use them for betting and making better informed decisions. Teams are given a better shot at winning due to popularity of opinion. We can take information from previous seasons, personnel, past results, and utilize that to make an informed popular opinion. This includes the use of saber metrics from previous seasons. But there is no definitive formula because its still a prediction of future results.

The only true odds right now are 1/32. If I have 32 cards with only 1 of them having an X on them. The chance that someone will pick that X is 1/32.

As records shift, odds will shift too.

 

[bmorepunk]

Have they stated the chances of a team winning the Super Bowl is even across all 32 teams?

No. But what you said about not being able to reasonable predict the odds early played into my response. FOs model is reasonable, given that they gave Seattle a 20% or so chance to win the Super Bowl last year, which was quite high and as it turns out, pretty accurate.

Point being, there aren't an indefinite amount of variables. The key is finding the meaningful statistics in the noise.

There is no way to sufficiently test the models that FO is using to predict a Super Bowl winner. The NFL season is very complex and it ends up being a one-shot process; in order to test their model we'd have to run the same season with the same initial conditions many times to get a distribution of winners. Unless they give a team a 0% probability and that team wins, we can blindly say any team they assigned a probability greater than zero was a reasonable probability. We can't tell the difference between this system and a system where someone eyeballs what look like the best teams and assign them higher probabilities.

There are not an indefinite amount of variables, but there are a very large number of variables. FO can choose the ones they want and design a model, but there is no determinate way to show the accuracy of that model without testing it repeatedly.

E.P. Box once said that "Essentially, all models are wrong, but some a useful." While FO's models are fun for fans of the sport they're not particularly useful.

 

[chris98251]

If you find a way to crunch numbers for the Super Bowl winner can you do it for the lottery next?

 

[bmorepunk]

So we're really discussing the difference between chance and odds?

If what you say, Cartire, is absolutely true, then believing we have a "better chance" than other teams is simply horse-manure, in spite of our team's previous success, players, coaches, homefield advantage, etc., right? I'm just clarifying.

Correct. At this current junction. However, what you guys are trying to make a point of is basically our standard practice of odds. We use them for betting and making better informed decisions. Teams are given a better shot at winning due to popularity of opinion. We can take information from previous seasons, personnel, past results, and utilize that to make an informed popular opinion. This includes the use of saber metrics from previous seasons. But there is no definitive formula because its still a prediction of future results.

The only true odds right now are 1/32. If I have 32 cards with only 1 of them having an X on them. The chance that someone will pick that X is 1/32.

As records shift, odds will shift too.

Sports gambling uses statistical models to determine how to best leverage the guesses of a reasonably large population to make money. Betting odds and probabilities of systems are not the same things. These two things related to NFL games and NFL seasons are related, but the idea that betting odds accurately represent the probability of events is absurd and unscientific.

Your poor understanding of probability is demonstrated by your treatment of the NFL season's winner as a stochastic process; it is not. We do have more information as the season goes along to make more informed guesses as to who will win the Super Bowl; the actual odds of any team winning the Super Bowl is 1/32. It is not a random process, and we don't understand the system nor know how to measure proper initial conditions. Not knowing these does not make the probabilities even.

 

[bmorepunk]

If you find a way to crunch numbers for the Super Bowl winner can you do it for the lottery next?

The lottery is a essentially a random process. All you can do with that optimally is to buy one ticket; your chances of winning go from zero to something really small (but the increase in probability is obviously huge between the two).

Maybe someone can come up with a good model for equities so I can predict the performance of stocks and make a bunch of easy money, since apparently complex systems are made simple.

 

[Cartire]

So we're really discussing the difference between chance and odds?

If what you say, Cartire, is absolutely true, then believing we have a "better chance" than other teams is simply horse-manure, in spite of our team's previous success, players, coaches, homefield advantage, etc., right? I'm just clarifying.

Correct. At this current junction. However, what you guys are trying to make a point of is basically our standard practice of odds. We use them for betting and making better informed decisions. Teams are given a better shot at winning due to popularity of opinion. We can take information from previous seasons, personnel, past results, and utilize that to make an informed popular opinion. This includes the use of saber metrics from previous seasons. But there is no definitive formula because its still a prediction of future results.

The only true odds right now are 1/32. If I have 32 cards with only 1 of them having an X on them. The chance that someone will pick that X is 1/32.

As records shift, odds will shift too.

Sports gambling uses statistical models to determine how to best leverage the guesses of a reasonably large population to make money. Betting odds and probabilities of systems are not the same things. These two things related to NFL games and NFL seasons are related, but the idea that betting odds accurately represent the probability of events is absurd and unscientific.

Your poor understanding of probability is demonstrated by your treatment of the NFL season's winner as a stochastic process; it is not. We do have more information as the season goes along to make more informed guesses as to who will win the Super Bowl; the actual odds of any team winning the Super Bowl is 1/32. It is not a random process, and we don't understand the system nor know how to measure proper initial conditions. Not knowing these does not make the probabilities even.

Hmm, I think you just want to argue. So im done with you. I said nothing about probability. I said that betting odds are made using popular opinions based on past results. Thats exactly what it is. They create odds that will best reflect the majority of bets evenly distributed. If the popular opinion is that Seattle will repeat, their odds will be less.

But like I said, since you jsut want to argue and tell me what I dont know, even though i never even mentioned the word probability, then I say good day to you sir.

 

[bmorepunk]

You replied directly to a statement about probability in which you state "it" is 1/32. Neither the odds that are made for gambling nor the determinate probability of any given team winning the Super Bowl is 1/32. What is "it"?

 

[Seahawk Sailor]

Actually, at the start of the season. IT is 1/32. All teams are considered equal at this point, records are even across the board. No single schedule is technically harder yet. Analysis utilize past seasons for predicting strength of schedule, along with past personal stats and observations.

But currently, at the start of the 2014 season, all teams are even. Regardless if we assume the Raiders will suck again or what not.

As of right now, its 1/32.

You sound more like a Browns fan at the beginning of a season than anything else I've heard in a long time. Currently, Johnny Manziel and Russell Wilson have the same quarterback rating too. That must mean they both have equal chances at that MVP award. Now where did I put that damned 32-sided dice?

[youtube]yCFB2akLh4s[/youtube]

 

[bmorepunk]

If "it" is 1/32 I can easily exploit the odds in Vegas to win lots of money.

Now I just have to figure out what "it" is. Is it the Matrix?

 

[pocketprotector]

In the last 10 years, the returning champion has a 10% chance of winning the superbowl.

In the last 10 years, how many have actually done it?

New England 2004-05.

One in the last ten years. Ten percent chance. I'm not very good with math. Can someone tell me if this sounds right?

10% have done it, but that doesn't equal chance, I don't think.

Yes, I agree that the OP has no idea what the difference between probability and simple division is.

My language was sloppy. But my point was that the media paints it as predictive, but that narrative doesn't add up.

And your guess about my knowledge of statistics is 100% wrong.

 

[pocketprotector]

Also for the troll who would rather nitpick language than discuss the point of the post, the word chance is perfectly acceptable from the context of 'pick any returning champion from the last 10 years at random and there is a 10% chance they will be repeat winners.' This is a common usage.

The point, is that over the last 10 years returning champions have won at a rate 3+ times higher than the rest of teams. A small sample size, a meaningless number, but the sort of stupid narrative peddled by the likes of Peter King doesn't add up.

 

[bmorepunk]

I agree that the idea that gets thrown around of the predictability of the repeat is silly.

"Troll" is inappropriate. I'm not trying to get people riled up. I do think the language is very important when talking about mathematics and statistics. While a lot of these terms are loosely used in common writing we should work hard to make sure what we're saying is reasonably accurate.

I don't think Peter King has any idea what numbers are.

 

[BlueTalon]

Logic doesn't work when you start with a false premise.

So if it weighs the same as a duck, it's made of wood?

 

[Ad Hawk]

I agree that the idea that gets thrown around of the predictability of the repeat is silly.

"Troll" is inappropriate. I'm not trying to get people riled up. I do think the language is very important when talking about mathematics and statistics. While a lot of these terms are loosely used in common writing we should work hard to make sure what we're saying is reasonably accurate.

I don't think Peter King has any idea what numbers are.

Agreed.

I, too, strive for precision in language (and spelling/grammar; tips hat to Roland) as well. Without it, misunderstandings lead to needless disagreements. Nobody here is trolling intentionally. I for one just want to rightly comprehend the OP's message, and accurately hear those following it.

If, knowing what you know about all the teams in the league at this moment, you were told that the Bills and Rams would eventually face off in the Super Bowl this season, would you be surprised? If so, then you believe they don't have the same "chance" to win as other teams, whatever the reasons. Do we mean potential instead of chance? I'm still trying to wrap mind around this because the way I look at it, I can't see the Hawks not having a winning season based on a lot of factors. I don't believe they have the same "chance" as every other team. I just need help with my terminology. In fact, I believe they have more potential to repeat than any of the last several teams that have won.

 

[bmorepunk]

I think a good way to put it is that every team has the same opportunity at the beginning of the season to win the Super Bowl. They all start with the same record and they have play under the same basic circumstances (number of games, rules structure, etc.) to get there.

Where this breaks is that not all teams experience the situation where their personnel, schedules, conditions (officiating and other things they can't control), execution, and injuries are the same. We can certainly say that teams who have better personnel (coaches and players), particularly those that have proven themselves in the past, should have a better chance of winning the Super Bowl than teams that have it much worse. We can try to use schedules as a measure at the beginning of a season, but every year the perception of a schedule's difficult at the beginning of the year can be quite different by the end of the year. Injuries can drastically change a team's performance and things that are out of the teams' control can decide games.

An analogy to this might be that we have two students at a high school that both want to go to a prestigious university. One is from a stable upper middle class family and the other is from a poor family that has a lot of issues at home. While both these kids have the same academic opportunities, one of the kids has more advantages starting out in the same system. It's possible that the poor kid can gain some advantages through working harder and/or having a knack for being a better student, but if we were to try to guess which kid had the best chance of getting there it's pretty clear what people would choose.

We know just due to uncertainty of this complex system that we can't accurately assign probabilities to who will win the Super Bowl. We can look at it from the perspective of Bayesian probability. This allows us to make a hypothesis based on the knowledge on hand and adjust it as more information comes available. At the beginning of the season, we can guess what we might think the chances of any given team winning the Super Bowl are, but as the season wears we adjust those guesses. This is essentially what sports writers (unscientifically) do in terms of making picks at the beginning of the season then changing them as it progresses.

Nate Silver released a book a couple years ago called The Signal and the Noise. It's a great read for just about anyone and he talks a lot about a Bayesian approach to trying to predict sports performance.

 

[ZagHawk]

As long as we have a better chance than the Niners (which I think we do). Works for me.