The Knish’s Model NFL Week 6: Projections, Picks, and Analysis, Part 2

43
The Knish’s Model NFL Week 6- Projections, Picks, and Analysis, Part 2

UPDATE: The total in the Texans v Colts game has moved back to 43. Just under 3.5% of NFL games land on a total of 43. For that reason, 43 is an important number as we now capture 3.5% of push or tie value. With the added value, the model is taking OVER 43.

Houston Texans v Indianapolis Colts (-10, 43.5)

Texans 18.40 – Colts 23.20

The model’s raw expected spread makes the Colts -7.16. So the line at -10 seems within the realm of reason. The simulations tell a different story, the model has the Colts winning by only 4.8 points. 

Wentz looked great against Baltimore, and it is easy to see the path for a blowout. These are two bad teams playing each other, but the Texans may just be next-level bad. It is also hard to know what to make of Davis Mills’ performance last week. Not only did he look sensational, but he also did it against Bill Belichek. I know I proved through the numbers that Belichick is simply an average coach, but he is an exceptional defensive mind that usually has his way with rookies. 

Davis Mills, in addition to having two last names, has the legitimate arm talent and is smart. Was it a fluke? We will see, but as Kevin from The Office will remind you:

The model is not taking a position in this game.


Green Bay Packers v Chicago Bears (4.5, 46)

Packers 21.29 – Bears 17.79

The model’s raw expected spread makes the Packers a -5.93 favorite on a neutral field. Thus, I would expect a line around -3.5. The simulations have the Packers winning by 3.5 exactly. 

This is one of those medium favorites that the model believes could be vulnerable this week. It would not be shocked to see the Bears win this game. The Bears will be able to run the ball. And Fields has played far better than people understand. The PAckers will be without J’aire Alexander at cornerback. The combination of Bill Lazor calling the right type of offense for Fields, and this defense puts the Bears firmly in this game.  

The model is not taking a side in this game. 


Cincinnati Bengals v Detroit Lions (3.5, 48.5)

Bengals 28.00 – Lions 15.09

The model’s raw expected spread makes the Bengals -7.61 on a neutral field. Thus, the line is close but shaded to Detroit. The model would expect this line to be either -4.5, or -4. The model sees an easy win for the Bengals by 12.01 points. 

The shallow line is a reaction to two concurrent events. First, the Lions are playing hard and have lost in an improbable, heartbreaking fashion. There is a building sense that the Lions are somehow “due.” They are not. There is no such thing as due. Five black numbers on a roulette table do not mean that red is due. Second, we have a history with the Bengals that makes them hard to trust. The Bengals have the better quarterback and the better coach. 

Oddly specific prediction: Chidobe Auwzie picks off Jared Goff. 

The model is taking the Bengals -3.5


Los Angeles Rams v New York Giants (10.5, 48)

Rams 31.44 – Giants 17.31

The model’s raw expected spread makes the Rams -7.38 on a neutral field. This line is maxed toward the Giants due to the injuries to Jones and Barkley. The model’s simulations do not believe it is juiced enough as it sees the Rams -14.13. 

This is a great pick for a survivor pool. Otherwise, I am simply staying away from this game. This is the classic bad beat, back-door cover game. 

The model is not taking a position in this game. 


Arizona Cardinals v Cleveland Browns (-3, 54)

Cardinals 26.03 – Browns 29.38

The model’s raw expected spread makes this game Cardinals -.71 on a neutral field. The line opened at Browns -2.5 which seems nearly spot on what the model would expect. The model simulations have the Browns winning by 3.35 points. 

I am not as high on the Cardinals as everyone else seems to be and I am higher on the Browns than most. The Cardinals schedule:

  • Two rookie quarterbacks (Lawrence & Lance)
  • Struggled mightily in the first half against the Jags
  • Were lucky to win against the Vikings
  • Were lucky to win against the 49ers
  • Beat a bad Tennessee team
  • Beat a good Rams team

This team could very easily be 3-2 or 2-3. Let’s pump the brakes on Arizona. I think they are in the middle of a hot run and I look for them to regress overall, especially on the defensive side of the ball. Cleveland should have zero trouble running the ball and eating the clock against this team.  

Full disclosure: I teased the Cardinals to +8.5 and the Vikings to +7.5 earlier in the week. I backed the Browns at -2.5, but am not interested at -3. 

The model is not taking a position in this game. 


Las Vegas Raiders v Denver Broncos (-4, 44.5)

Raiders 21.21 – Broncos 22.94

The model’s raw expected spread (RES) makes the Broncos – .14 on a neutral field. So I would expect the spread to be -3 at the most.  The simulations make the Broncos -1.73. 

Gruden is out, so what do we expect now? He had the Raiders offense performing over expectation, so perhaps a little long-term regression. However, they may also get a short-term boost as a change in coaching tends to refocus everyone on the team in the short run. 

Now that the number has crossed to 4, I am taking the Raiders. 

The model is taking the Raiders +4


Seattle Seahawks v Pittsburgh Steelers (-5, 42.5)

Seattle 23.62 – Pittsburgh 19.14

The model’s raw expected spread (RES) makes Seattle -2.56 on a neutral field. So I would expect the lien to be Pittsburgh -1. The model simulations see Seattle winning by 4.48.

The line is driven by two factors. First, Wilson’s injury. Second, how “well” Pittsburgh played last week against Denver. As great as Wilson is, he does not make a nine-point difference (the 4.48 projected win, plus the line of 5). 

Of course, confidence is quite low because we have such a small sample size for Geno Smith. Yes, he played really well against the Rams and they should have covered my +8.5 teaser (Newman!) but he played against a Rams prevent defense. The pick was not his fault as his receiver was tripped on the play. 

If this line gets to -6, it is at -5.5 at places, I will jump on the Seahawks. 

Waiting to see if the line moves to -6


Buffalo Bills v Tennessee Titans (5.5, 54.5)

Bills 30.56 – Titans 15.91

The model makes the raw expected spread (RES) Bills -5.56 on a neutral field. So It is understandable why the books opened this spread at -3.5. The simulations see an absolute slaughter, Bills by 14.65 points. 

Whenever there is a massive disparity between the RES and the simulations is a bit of a red flag. If you read my preseason previews, you know that I think very little of the AFC South. The model thinks less of them. The only difficulty I see is if Allen struggles with his accuracy. Badly. 

The model is taking the Bills -5.5.

About the author:

+ posts

I write about data and sports. I created my first model in 1997 using nothing more than Excel. Currently, I have data-driven models for the NFL, NBA, and World Cup Soccer.

Mathematics is the music of reason.
— James Joseph Sylvester, English mathematician

Leave a Reply