Nate Silver: Pats 2003-2014 Best 12 Yr Stretch in NFL History

[Soul_Survivor88]

@FourierSeries what does Cortana say about the Pats getting to and possibly winning the Super Bowl?

 

[brownj12]

and soon enough, 2001-2015 will be the best 15 year stretch in history.

If the patriots manage to keep their elo at the Average for Brady's career, they will take the 13,14,15, and 20 records after this season, 16,19,and 21-24 would fall after the 2016 season.

 

[venecol]

Absolutely. The sample size is so ridiculously small that forecasting is a crap shoot.

My guess is that Nate's model shows a distribution of possible W/L outcomes for the Pats season that ranges from 7--9 or 8--8 to 15--1 or 16--0 out to the fourth or fifth Standard Deviation. The 11.6 wins is the mean of the distribution with a standard deviation of some sort, probably 1.5--2.0 games.

So, in the second week, if the Standard Deviation is 1.5 games, Nate's model would be saying that there is ~67% confidence that the pats will win between 10.1 (10) and 13.1 (13) games and ~99% confidence that the pats will win between 8.6 (9) and 14.6 (15) games, etc. In other words, "Interesting" but not "Helpful."

Now I remember why I fell asleep in that statistics class. Lol

 

[Urgent]

538 shows NE on their week to week with a high probability of defeating everyone they face except Denver which they have at 50%/50%. But their simulations show an average 12-4 season. I should have stayed awake in that statistics class.

Imagine that there is a series of transactions with a 60% probability of outcome A and a 40% probability of outcome B over 16 repetitions. At the end of 16 repetitions, statistically you would not predict outcome A 100% of the time; you would predict it at 60% of the time.
So, unless you are favored at 100% to 0%, just because you are favored in each match-up it does not mean you are projected to win all match-ups.

"60% of the time it works every time."

 

[venecol]

Imagine that there is a series of transactions with a 60% probability of outcome A and a 40% probability of outcome B over 16 repetitions. At the end of 16 repetitions, statistically you would not predict outcome A 100% of the time; you would predict it at 60% of the time.
So, unless you are favored at 100% to 0%, just because you are favored in each match-up it does not mean you are projected to win all match-ups.

"60% of the time it works every time."

Thanks to all of you brainiacs for clarifying my stupid post. I get it. I hate statistics, always have always will.
I'm just a dumb engineer. Lol

No seriously thanks guys. But I really do hate statistics.

 

[RecoveringCowboy]

The Patriots* obviously got to 1849 intentionally just to mock the 49ers. Wouldn't expect anything less from those classless cheaters

It was a message from God

 

[Fencer]

But, if I had to venture a guess in this case, it would be that the Standard Deviations around the mean outcomes of games more than two or three weeks out in the future are pretty large. The result would be that a few hundred thousand simulations of potential season W/L records for, for example, the Patriots most likely yield a wide range of possible records, the mean of which has 11.6 wins after week 1. The result will change week by week.

1. So you're saying it's probably Monte Carlo vs. closed form (or more than "probably")?

2. Do you think he's taking into account possible injuries and so on? What I mean by that, is he taking into account the chance that deviations from current predictions might be highly correlated over time (because of some common causes, most likely injury)?

 

[PatsFanSince74]

1. So you're saying it's probably Monte Carlo vs. closed form (or more than "probably")?

2. Do you think he's taking into account possible injuries and so on? What I mean by that, is he taking into account the chance that deviations from current predictions might be highly correlated over time (because of some common causes, most likely injury)?

1) I'm guessing he's doing simulations , but I really don't know.
2) I don't know how sophisticated the model is. Knowing Nate's other work, I'm sure he's described it somewhere; I just haven't looked for it. His political forecasting was based on adjusting polls for a whole range of factors, so it's possible he's developed some algorithm that accounts for injuries...or he could just be messing around to see what works.

 

[FourierSeries]

@FourierSeries what does Cortana say about the Pats getting to and possibly winning the Super Bowl?

Nothing yet. Last year she predicted NE would win the SB back in December. I am guessing that's when her SB prediction will be made this season as well.

 

[FourierSeries]

Absolutely. The sample size is so ridiculously small that forecasting is a crap shoot.

My guess is that Nate's model shows a distribution of possible W/L outcomes for the Pats season that ranges from 7--9 or 8--8 to 15--1 or 16--0 out to the fourth or fifth Standard Deviation. The 11.6 wins is the mean of the distribution with a standard deviation of some sort, probably 1.5--2.0 games.

So, in the second week, if the Standard Deviation is 1.5 games, Nate's model would be saying that there is ~67% confidence that the pats will win between 10.1 (10) and 13.1 (13) games and ~99% confidence that the pats will win between 8.6 (9) and 14.6 (15) games, etc. In other words, "Interesting" but not "Helpful."

That's not necessarily true.

The standard error is . So it would decrease with the number of n, which is the number of simulations, not the number of games. If you run enough simulations, you can generate a fairly useful result.

It would be like using a Monte Carlo simulation to price an Asian Option with a short maturity.

The real problem is that these methods themselves are most useful in the hard sciences. Outside of that, you can't rely on them too much.

 

[FourierSeries]

So both Nate Silver and Cortana had very poor records picking games this week after a successful week 1. Through 2 weeks, they both have an accuracy of 59.38%.

Nate Silver's model has updated predictions for the number of wins and for the SB. NE is now projected to win 12 games, and they have a 19% probability of winning the SB. I believe this is slightly higher than last week. GB moved into 2nd place with a 12% probability of winning the SB. Denver and Dallas are tied for 3rd with 10%.

Both Nate Silver and Cortana give the Pats over 80% odds of winning this week. Not really surprising.