Did they really try Charmin’ out on a bear's ass? I don't think it could wipe a bear's ass, it has problems with mine. They like to lie.

On to week 4 and our second week of trying to predict who will cover the NFL lines.  Last week the model did surprising well by correctly predicting 10 of 16 games (62.5%).  Honestly, I don’t expect this run of good predictions to continue but I can still be hopeful.  For those who don’t remember or new readers (again I can be hopeful), the models are created using multiple regression to predict points scored for the week.  The difference in points scored between the two opponents is used to decide which team will cover against the line.  This week I will be introducing a competing model to the one used last week.

Model 1:  This is the same model as last week.  It uses the following variables to predict points scored:

1.      Rushing attempts

2.      Pass Completion Percentage

3.      Rushing Touchdowns

4.      Passing Touchdowns

5.      Turn-overs

6.      Spread

However, I did become a little concerned that the adjusted r-square dropped from .82 to .79.   This is more than likely due to the fact that we are ignoring some variables that affect the amount of points scored.  My answer to that:  No shit.  The opponent is not being modeled at all as well as the effect of injuries.

Model 2:  This model is the result of straight data mining.  I plugged all the offensive variables into a model selection procedure that spits out the model with highest adjusted R squared.  Since the top 20 results varied in their adjusted R squared by less than .01, I selected the most parsimonious model (which still had 8 predictors).

1.      Rushing 1^st downs

2.      Red Zone Efficiency

3.      Rushing Attempts

4.      Rushing Yards

5.      Rushing Touchdowns

6.      Pass Completions

7.      Passing Touchdowns

8.      Other Touchdowns

Because this is data mining, we get some weird results.  For instance increasing red zone efficiency actually decreases points.  Same with 1^st downs obtained by running the ball.  Now this could be explained away theoretically, but it is probably the result of inter-correlations among the variables. 

Anyways on to the games.  This week we will be comparing both models to predictions made by me Wojay and Kyle Kelly.

·         Panthers at Bears

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Bears by 6	Panthers	Panthers	Panthers	Bears	Bears

Wojay Wonders:  It’s hard to pick a winner when you know at least 6 combined interceptions will be thrown.

·         Bills at Bengals

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Bills by 3.5	Bills	Bills	Bills	Bills	Bills

Wojay Wonders:  Circle the Wagons!!!

·         Titans at Browns

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Pick 'em	Browns	Titans	Browns	Browns	Titans

Wojay Wonders:  Didn’t Chris Johnson get paid all that money to be like, you know, good?

·         Lions at Cowboys

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Cowboys by 1.5	Lions	Lions	Lions	Lions	Lions

Wojay Wonders:  The Lions have a receiver named Megatron.  Have you seen Transformers?!?!

·         Steelers at Texans

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Texans by 4	Texans	Texans	Steelers	Texans	Steelers

Wojay Wonders:  How can you pick the Steelers when they needed a last second FG to beat the Colts?

·         Saints at Jaguars

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Saints by 7	Saints	Saints	Saints	Saints	Saints

Wojay Wonders:  How much longer until they’re the LA Jags?

·         Vikings at Chiefs

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Vikings by 2	Vikings	Vikings	Vikings	Vikings	Chiefs

Wojay Wonders:  The Vikings will finally avenge Super Bowl 6.

·         49ers at Eagles

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Eagles by 9	49ers	49ers	49ers	Eagles	Eagles

Wojay Wonders:  Alex Smith has only been in the league 6 years.  Give the kid time to develop.

·         Redskins at Rams

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Redskins by 2	Redskins	Redskins	Rams	Rams	Redskins

Wojay Wonders:  Are Redskin’s fans even excited for this game?

·         Giants at Cardinals

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Giants by 1	Giants	Giants	Giants	Giants	Giants

Wojay Wonders:  Hard to take the Cardinals seriously, when they lose to the fightin’ Tavaris Jacksons.

·         Falcons at Seahawks

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Falcons by 4.5	Falcons	Falcons	Falcons	Falcons	Falcons

Wojay Wonders:  See above for my opinions on the fightin’ Tavaris Jacksons.

·         Broncos at Packers

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Packers by 12.5	Packers	Packers	Broncos	Packers	Packers

Wojay Wonders:  All of Tebow’s faith won’t help the Broncos in Green Bay.

·         Patriots at Raiders

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Patriots by 4	Patriots	Patriots	Patriots	Patriots	Raiders

Wojay Wonders:  There may be 70 combined points in this game.

·         Dolphins at Chargers

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Chargers by 7	Dolphins	Chargers	Dolphins	Dolphins	Chargers

Mike Z Thoughts:  Don’t ask…

·         Jets at Ravens

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Ravens by 3.5	Jets	Ravens	Jets	Ravens	Jets

Wojay Wonders:  Ray Lewis killed two people with a knife.  Why isn’t that talked about more?

·         Colts at Buccaneers

Spread	M1 Prediction	M2 Prediction	Mike_Z	Wojay	Kyle Kelly
Bucs by 10	Colts	Colts	Colts	Colts	Buccaneers

Wojay Wonders:  The betting Gods have been smiling on the Colts

So there you have it.  The two models differ on three games, so we should be able to evaluate which model did better at the end of the week.  As always thanks to Wojay and Kyle Kelly for collecting data, providing input, and calculating the money.