Win probabilities for week 4 NFL games are listed below. The probabilities are based on an efficiency win model explained here and here with some modifications. The model considers offensive and defensive efficiency stats including running, passing, sacks, turnover rates, and penalty rates. Team stats are adjusted for previous opponent strength.Pwin Visitor Home Pwin 0.47 
SF
NO0.53 0.35 
HOU
JAX0.65 0.21 
CLE
CIN0.79 0.38 
ATL
CAR0.62 0.61 
ARI
NYJ0.39 0.44 
MIN
TEN0.56 0.72 
DEN
KC0.28 0.36 
GB
TB0.64 0.64 
SD
OAK0.36 0.72 
BUF
STL0.28 0.56 
WAS
DAL0.44 0.46 
PHI
CHI0.54 0.36 
BAL
PIT0.64
Game Probabilties Week 4
published on 9/25/2008
Subscribe to:
Post Comments (Atom)
I assume the redskins are favored over the cowboys in this model due to interceptions?
I assume the Redskins are favored over the Cowboys in this model because Tony Romo is a homo. -Andy
No. I don't have a homo factor in the model.
It's mostly due to interceptions and fumbles. But the Cowboys' penalty rate (pen yds per play) is sky high. It's likely to come down in the future, but for now it's hurting them.
I doubt if the Redskins are truly the favorite, but they shouldn't be an 11-point underdog.
Brian,
I found the site about a month ago and really enjoy what you do here. I plowed through all of the archives and put together a little system of my own. I'm anxious to see how mine performs, along with yours.
The numbers I have right now seem to be a little more "correct" on the surface. I'm a Redskins fan, and even I know there is no way they are favored heading into Sunday. I have the Cowboys winning 77% of the time.
I'm putting my numbers up on my "site" if you care to take a look. Keep up the good work here.
For some reason, entering my URL wasn't correct, so here it is.
http://www.smarterthanexperts.com
For the guy who said that if washington is favored in the model they shouldn't be a dog....I've worked in the casino biz for 10 years and the game spread has NOTHING to do with the actual abilities of either team but it is instead designed to mirror the publics perception of the two teams in an effort to generate equal betting action on both sides thus assuring that the sportsbook will win it's 10% vig.
While I agree with you 100% about what the goal of the line is, I disagree with you 100% on the game spread having nothing to do with the actual abilities of the teams. The closing line is arguably the best indicator of the talent level of teams. The "wisdom of crowds" theory applies here. To say that the game spread has nothing to do with the actual abilities is, to me, foolish.
The line definitely takes in the public's perception, but vegas sets the line based on their rankings of teams. Right now they have Dallas as 8 point better than the skins on a neutral field, plus the usual 3 points that a home team gets, equals an 11 point favorite this week. Vegas has dallas as the best team in the league by far. I think they would be 3 or 4 point favorites over any team on a neutral field right now. I'm not making this up either, this is based off of the LVSC's rankings, I just don't have the link right now. So Vegas's preliminary spread most certainly is set by how good they think certain teams are.
brian
i forget, how you deal with HFA?
do you account for HFA in probabilities? If so is it a genric % for
all home teams or is it specific to each team?
thx
dan
Hi Mr. C,
HFA is accounted for by a dummy variable (1 or 0) in the model. The coefficient works out to give the home team a win % of 57% on average. The closer the two opponents are in efficiency performance, the stronger the effect of HFA.
wow so far this week these predictions are great
Whoah.
Congrats on a strong start! 9-2 so far!
Now why didn't I go with these picks like I said I would? Nice picks! I imagine if you knew Palmer wasn't starting you might not have picked Cincy.
Yeah, that's the Achilles heel of a purely statistical model--the injured QB.
I dont know how I stumbled on this site, but I made very small bets on each game this week based solely on your probabilities and did fairly well. I always used to read Troy Aikmans rankings actually, as he took into account many factors in a game as well, but this seems very well done.
I just wish I could have predicted Brett Favre throwing for 6 TDs for my fantasy team haha!
Wow, I just checked http://www.thepredictiontracker.com/prednfl.html and your system was THE ONLY one that correctly predicted the outcome of both the WAS @ DAL and PHI @ CHI games. That is, your system probably predicted the true favorite, whereas the spread reflected more of the public opinion. I find that very interesting. And again, I believe you already noted it, but your system is the only one that includes a "homo factor", thus correctly capturing the effect of Tono Romo on game outcomes. - AndyP
I am hoping that it is OK to post this here. I just posted my results from this week on my site, along with next week's probabilities, a spread and total for each game.
BTW Brian, nice pick on the 'Skins. Hopefully your system picking them to win is the catalyst they need to go 15-1.