3 in 4 chance of undefeated regular season?

[sportsratings]

With seven games left, I've projected the Patriots' remaining schedule and came up with a remarkable 74.5% chance for a 16-0 finish. This is using a statistical system that compares all of one team's game performances against another team's. Normally odds of winning seven straight come out very low.

Odds of 19-0 are almost 50%

Funny thing is, I think most people would say those odds are the same--that is, it seems unlikely the Pats could go 16-0, then proceed to lose in the playoffs.

what's your take, if you had to put % odds on either question (16-0, 19-0)?

 

[drew4008]

What the hell? This isn't math class, it's football.

 

[Patriot_in_NY]

Whadda I think? I think we play Buffalo on the 18th.

 

[PATRIOT64]

This is stupid - How can you possibly have ANY odds of what 19-0 is when you don't have any idea of what the teams the Patriots will be facing in the playoffs??,Its impossible to calculate right now -
Hell we don't even know if NE will have the
#1 seed yet?

 

[alamo]

This is stupid - How can you possibly have ANY odds of what 19-0 is when you don't have any idea of what the teams the Patriots will be facing in the playoffs??,Its impossible to calculate right now -
Hell we don't even know if NE will have the
#1 seed yet?

Don't be ridiculous. Those should be directed at yourself, not at the original post.

Every single setting of odds requires some base assumptions and projections. Obviously a projection of results for the next 10 games is less certain than the projection of the next game, but that doesn't mean it's not valid as a projection. Who will have the #1 seed is part of the projection, your complaint about the #1 seed is like saying forecasting today's weather is impossible because we don't know for sure if it will rain or not.

This guy took his rating system and ran the rest of our schedule though it, including playoff matchups assuming the current standings hold. He was looking for a specific result, a calculation of how likely an undefeated season is. The results are notable because for almost any team in history the same projection technique would yield a much much lower probability of an undefeated season than it does here.

It's just another indication from an independent source that we are seeing something special here.

 

[Satchboogie3]

Stats are on our side. They simulated the rest of the season on accuscore, the Pats went 16-0 60% of the time... I like those odds. I really like those odds.

 

[Fencer]

Argggh. When I was a stock analyst in 1983, our investment strategists used the same fallacy -- they came up with some kind of percentage evaluation, and falsely referred to it as a "probability."

I hated it then and I hate it now.

The probability that the Pats beat the Ravens is NOT 100%. Ditto the Jets (although in that case it's pleasingly close).

So let's play around with some numbers.

Suppose the Pats have one game each remaining with .8, .85, and .9 probabilities of victory, and all the others are at .95. Then there's a 42.7% chance of 19-0.

If you suppose there are two each at .8 and .85, one at .9, and that the others are at .95, the probability of 19-0 is 32.2%.

But how realistic are those high figures, considering that:

• The Pats have to play Indy again.
• The Pats have to play three other playoff teams.
• The Pats have to play Pitt in the regular season.
• The Pats have to play the Giants Week 16.
• Brady -- durable as he is -- could get injured.
• Moss could get injured.

and so on?

 

[ctpatsfan77]

• The Pats have to play Indy again.

and so on?

How do you know that?

 

[PatsWickedPissah]

Argggh. When I was a stock analyst in 1983, our investment strategists used the same fallacy -- they came up with some kind of percentage evaluation, and falsely referred to it as a "probability."

I hated it then and I hate it now.

The probability that the Pats beat the Ravens is NOT 100%. Ditto the Jets (although in that case it's pleasingly close).

So let's play around with some numbers.

Suppose the Pats have one game each remaining with .8, .85, and .9 probabilities of victory, and all the others are at .95. Then there's a 42.7% chance of 19-0.

If you suppose there are two each at .8 and .85, one at .9, and that the others are at .95, the probability of 19-0 is 32.2%.

But how realistic are those high figures, considering that:

• The Pats have to play Indy again.
• The Pats have to play three other playoff teams.
• The Pats have to play Pitt in the regular season.
• The Pats have to play the Giants Week 16.
• Brady -- durable as he is -- could get injured.
• Moss could get injured.

and so on?

That is a much more rational assesment of individual probabilities, the point 8 and point 9 range. Note how when you multiply them, the end number starts to get small. But once the Pats WIN each next game, the probabilities will rise.

I know, I know, and the sun will set in the west today.

 

[mcbee]

He was talking regular season.

So let's do some simple math

Prob. of beating Buffalo = .95?
prob. of beating steelers = .7?
Raven? = .75
Miami =.95
Jets =.95
Philly =.8
Giants (hard one) = .8

Multiply those all together and it's 28%.

So I don't know where you get 75%.

In fact if you ASSUME that the pats are 100% to beat the bills, dolphins and jets and Eagles AND 90% to beat the steelers, ravens and giants, you get about 73%. ie .9 * .9 * .9 = .729.

So you're making some bold predictions.

If we assume the AFC east teams are 100% prob. to win and the others 90%, then it's .9^4 or 65%.

So what probabilities are you using to beat the Steelers, Ravens, etc? You must be assuming NE has AT LEAST 75% chance of beating the Steelers.

 

[patchick]

Multiply those all together and it's 28%.

So I don't know where you get 75%.

He got 75% by starting with some bizarre assumptions. He has 4 of the remaining 7 games coming it as literally guaranteed (100%) wins!

"Sportsratings," the fact that your model yields ANY 100% win probabilities should instantly tell you that your model isn't working.

 

[speed]

There are 3 kinds of people in this world...

Those who can count and those who can't.

 

[FirstAndGoal]

There's a 75% chance that this guy doesn't know what "stats" is short for never mind the fact that he doesn't know how to use them.

 

[FrontSeven]

You have a sore leg.

Vegas set the line at 1 to 3. Even that is pretty spectacular. But odds are what? Odds ..... that's about it. We knew that. We knew there were odds. Odds are beaten every day.

 

[TruthSeeker]

Here are the projected odds in the article:

Team odds NE win
at Buffalo 100.00%
Philadelphia 90.27%
at Baltimore 100.00%
Pittsburgh 86.11%
New York Jets 100.00%
Miami 100.00%
at New York Giants 95.83%


All I can say is - ridiculous.

No NFL team hs a 100% chance to win a game. I saw an NFL 10-0 team lose to an 0-10 team years ago. It happens.

And then to say that the best chance a team has to beat the Patriots is about 14% is also ridiculous.

Here are my odds on each game:

Team odds NE win
at Buffalo 85%
Philadelphia 80%
at Baltimore 75%
Pittsburgh 70%
New York Jets 85%
Miami 90%
at New York Giants 70%

This corresponds to a 19.11% chance of going undefeated. (The math is simple; just multiply them all together i.e. .85 * .8 * .75 * .7 * .85 * .9 * .7.)

 

[PatsFaninAZ]

Here are my odds on each game:

Team odds NE win
at Buffalo 85%
Philadelphia 80%
at Baltimore 75%
Pittsburgh 70%
New York Jets 85%
Miami 90%
at New York Giants 70%

This corresponds to a 19.11% chance of going undefeated. (The math is simple; just multiply them all together i.e. .85 * .8 * .75 * .7 * .85 * .9 * .7.)

Those look pretty reasonable. I actually would lower our odds just a bit on the road games. Buffalo is going to be a tough game -- prime time, rowdy fans, surging team. Even before Buffalo got "good," they gave Dallas a pretty rough time at home in a prime time game. Not saying it's 50/50 by any means, but it's a tough one against a resurgent team that played us tough for a half at home. I think Baltimore is also tough to figure -- depends on which Baltimore shows up. They are traditionally tough at home too.

The Giants game is very tough to figure right now. It's possible that one or both teams will have little to play for. Or that one will have everything to play for and one won't.

 

[Pass_The_Rum]

This sort of reasoning is totally fallacious. Let me demonstrate.

Assume a dominant team has an average .9 probability of winning any given game over the season (averaging cupcakes and powerhouses both). This corresponds to an expected 14-2 or 15-1 record at the end of the season. In the past ten years there have been nine such teams. In theory, the possibility of a dominant team having a perfect season under these circumstances is .185, a bit less than one in five. It's been 25 years since the Dolphins went undefeated. At that rate of dominant team occurrence, there should have been 22-23 dominant teams. (14-2 or 15-1 teams)

By those odds, the probability of NO dominant teams going undefeated is the odds of such a team losing at least once (.815) raised to the power of 22. This is 1%, more or less. If the average really good team will defeat an average team 9 times out of 10, we should have seen a couple 16-0 seasons.

So the average percentage must be lower than .9, right? No, because then the number of 15-1 and 14-2 teams becomes a statistical abberation of its own. The probability for a team to go undefeated is too complicated to figure by such simple approximations.

 

[patchick]

Pass The Rum, thank you for bringing some serious food for thought to an otherwise pointless thread. If I interpret that correctly, a possible take-home lesson is that the 16 outcomes are not independent of one another. The mere fact of being undefeated entering a game changes the odds of winning that game.

 

[Fencer]

How do you know that?

I was oversimplifying.

It would be more precise to say that to go 19-0 the Pats have to beat the NFC champion, the eventual AFC runner-up, and so on.

 

[Fencer]

Pass The Rum, thank you for bringing some serious food for thought to an otherwise pointless thread. If I interpret that correctly, a possible take-home lesson is that the 16 outcomes are not independent of one another. The mere fact of being undefeated entering a game changes the odds of winning that game.

Example that further screws the analysis up: If the Pats are 14-1 going into the Giants game with HFA sewed up, they're apt to rest a lot more starters than if they're 15-0.