• Have something to say? Register Now! and be posting in minutes!

Harvard Sports Analysis Collective: Dolphins & Seahawks NFL's Top Teams

HammerDown

Well-Known Member
Supporting Member Level 3
68,257
5,320
533
Joined
Dec 21, 2012
Hoopla Cash
$ 198.00
Fav. Team #1
Fav. Team #2
Fav. Team #3
As the NFL's 32 clubs prepare to hit the dog days of training camp, a Harvard undergraduate has entered the prognostication game with a new mathematical model that predicts some surprises in next season's race for the playoffs.

If you ask Kurt Bullard of the Harvard Sports Analysis Collective, the Patriots' reign in the AFC East may be coming to an end. The Dolphins, who added Ndamukong Suh this offseason, rank third in Bullard's final table --two spots ahead of the Patriots -- with a 77 percent chance to make the playoffs, to New England's 62 percent odds. (He stays out of predicting the fate of Brady, who is currently suspended for the team's first four games.)

Bullard also has an interesting prediction for Western New York. After an offseason of record season ticket sales, the Bills come in with just a 40 percent chance to make the playoffs in the Bullard model, and he writes that "the much touted Bills revamping...may not have the desired effect." Buffalo has added LeSean McCoy, as well as the bombastic, high-flying Rex Ryan.

In explaining his approach (in much more detail here), Bullard writes that he's looking for "a more quantitative method [to predicting the NFL season] than the likes of Trent Dilfer." Shade aside, though, many of Bullard's other predictions mirror the expert zeitgeist. Seattle and Green Bay top the rankings, and New England and Denver are comfortably in the top ten. The Raiders, predictably, are last (with a one percent shot at the playoffs), joined near the bottom by Tampa Bay, Jacksonville, and Tennessee.

Turning his gaze to California, Bullard gives the 49ers just a 10 percent shot at the playoffs and highlights an exodus of talent from Santa Clara.

The skinny on Bullard's model: he worked with Pro Football Reference's Approximate Value statistic as well as depth chart information to try and quantify team strength and depth across position groups. He then ran schedule simulations and calculated playoff odds.


Harvard student's model picks Dolphins, Seahawks - NFL.com
 

HammerDown

Well-Known Member
Supporting Member Level 3
68,257
5,320
533
Joined
Dec 21, 2012
Hoopla Cash
$ 198.00
Fav. Team #1
Fav. Team #2
Fav. Team #3
Interesting but I'd still like to know more about how he's doing this. Does the algorithm take into account such things as odds for injury on specific positions or decline in performance based on the age of player?
 

jarntt

Well-Known Member
34,298
12,672
1,033
Joined
Aug 19, 2013
Hoopla Cash
$ 1,000.00
Fav. Team #1
Fav. Team #2
Fav. Team #3
Any model that gives the Dolphins a 77% chance to make the playoffs is broken beyond repair...
 

TDs3nOut

Well-Known Member
13,504
2,382
293
Joined
Jul 12, 2013
Hoopla Cash
$ 100.00
Fav. Team #1
Fav. Team #2
Fav. Team #3
For a number of years I have thought it was interesting to try using statistical models to predict the outcomes of sporting events. I'm sure that the kid from Harvard is smarter than I am, but I'm still skeptical about how reliably a statistical model might predict sports outcomes. Lots of noise in the data.
 

TDs3nOut

Well-Known Member
13,504
2,382
293
Joined
Jul 12, 2013
Hoopla Cash
$ 100.00
Fav. Team #1
Fav. Team #2
Fav. Team #3
The Seattle and Miami predictions don't strike me as totally "off the wall", though Seattle certainly seems more plausible than Miami. Lots of potential bumps in the road, of course, lie ahead for both those two and the other 30 teams in the league.

If Seattle and Miami do end up as the last two standing, I wonder if the Harvard undergrad gets some nice job offers.
 

flyerhawk

Well-Known Member
96,749
33,373
1,033
Joined
Aug 18, 2014
Location
Hoboken
Hoopla Cash
$ 500.00
Fav. Team #1
Fav. Team #2
Fav. Team #3
The Seattle and Miami predictions don't strike me as totally "off the wall", though Seattle certainly seems more plausible than Miami. Lots of potential bumps in the road, of course, lie ahead for both those two and the other 30 teams in the league.

If Seattle and Miami do end up as the last two standing, I wonder if the Harvard undergrad gets some nice job offers.

I think he is only predicting the odds of making the playoffs. Statistically that is the only thing you really can do.
 

Wolverine830872

2018 DCFFL Champion!
52,441
17,236
1,033
Joined
Jul 2, 2013
Location
Your head
Hoopla Cash
$ 500.87
Fav. Team #1
Fav. Team #2
Fav. Team #3
It looks like these projections are only for odds at making the playoffs. and it also sounds as though he used Brady's suspension in his calculations from my interpretation of the OP.

With all that being said, how is Indy not #1 on that list? Good team in a weak division should be very high odds.
 

Manster7588

I Support Law Enforcement.
46,055
13,480
1,033
Joined
Jul 17, 2013
Location
Las Vegas, NV 89129
Hoopla Cash
$ 920.85
Fav. Team #1
Fav. Team #2
Fav. Team #3
No problem with Seattle being on top but I don't see Miami there.
 

TDs3nOut

Well-Known Member
13,504
2,382
293
Joined
Jul 12, 2013
Hoopla Cash
$ 100.00
Fav. Team #1
Fav. Team #2
Fav. Team #3
I think he is only predicting the odds of making the playoffs. Statistically that is the only thing you really can do.

Absolutely false. One can easily use historical data on any subject to predict future outcome in most any context. Coming up with a reliable model, particularly when human behavior is involved, of course, is pretty tricky.
 

flyerhawk

Well-Known Member
96,749
33,373
1,033
Joined
Aug 18, 2014
Location
Hoboken
Hoopla Cash
$ 500.00
Fav. Team #1
Fav. Team #2
Fav. Team #3
Absolutely false. One can easily use historical data on any subject to predict future outcome in most any context. Coming up with a reliable model, particularly when human behavior is involved, of course, is pretty tricky.

We've had this discussion before. And you are still incorrect.

The sample set is simply too small. A 16 game season gives you a pretty good data set to come up with some sort of predictive model.

A single elimination tournament involves far too much chance. Even a team that is a 70% chance to win every game is still only has less than a 50% chance to win 3 straight games.

Which is No. 1 seeds fare far worse in results than they should.
 

The Derski

No Fat Chicks
38,839
6,124
533
Joined
Jul 11, 2013
Location
Tucson, AZ
Hoopla Cash
$ 418.10
Fav. Team #1
Fav. Team #2
Fav. Team #3
A lot of the fancy analytical equations and numbers crap that these Ivy League schools come out with is just a load of BS.
 

pennstatenut

Working with my wood
2,854
292
83
Joined
Jul 19, 2013
Hoopla Cash
$ 6,000.00
Fav. Team #1
Fav. Team #2
Fav. Team #3
He's probably taking into account that Brady has to play with regulation balls now.
 

Tgann69

tOffficial Avi Welcher
5,168
285
83
Joined
Aug 31, 2011
Hoopla Cash
$ 100.00
Fav. Team #1
Fav. Team #2
Fav. Team #3
It looks like these projections are only for odds at making the playoffs. and it also sounds as though he used Brady's suspension in his calculations from my interpretation of the OP.

With all that being said, how is Indy not #1 on that list? Good team in a weak division should be very high odds.
Maybe he thinks Miami has a big season NE drops off and the AFCE is weak? Plausible.
 

Tgann69

tOffficial Avi Welcher
5,168
285
83
Joined
Aug 31, 2011
Hoopla Cash
$ 100.00
Fav. Team #1
Fav. Team #2
Fav. Team #3
Absolutely false. One can easily use historical data on any subject to predict future outcome in most any context. Coming up with a reliable model, particularly when human behavior is involved, of course, is pretty tricky.
Yup... Do it everyday... Very doable
 

Wolverine830872

2018 DCFFL Champion!
52,441
17,236
1,033
Joined
Jul 2, 2013
Location
Your head
Hoopla Cash
$ 500.87
Fav. Team #1
Fav. Team #2
Fav. Team #3
Maybe he thinks Miami has a big season NE drops off and the AFCE is weak? Plausible.
True, but does he think that the NFCW and NFCN are going to be easier divisions for the Seahawks and Packers than the AFCS will be for the Colts?
 

Tgann69

tOffficial Avi Welcher
5,168
285
83
Joined
Aug 31, 2011
Hoopla Cash
$ 100.00
Fav. Team #1
Fav. Team #2
Fav. Team #3
True, but does he think that the NFCW and NFCN are going to be easier divisions for the Seahawks and Packers than the AFCS will be for the Colts?
Dunno but the AFCE has been weak other then NE. If NE had a fall off then possibly Miami is sitting alone atop with a decent schedule to play. Except when they lose to the Ravens of course. ;)
 
Top