Data scientist, physicist, and fantasy football champion

Week 13 DEF Prediction

It’s week 13 and for me the last week of the fantasy regular season. Hopefully you’re in a better position than me and you can take this time to look ahead with hope at the fantasy playoffs. I, however, will be praying that all of this modeling finally pays off big time and gets me that win since I need it to make the playoffs. Hopefully the DEF won’t be the deciding factor, but last week I lost by less than 5 points, and trusting a better model may have pushed me over the top.

This week I’m looking at six models. I want to eliminate them, but they’re all so hit-and-miss. Weeks 8 through 10 Models A and B did very well, but for the last 3 weeks Model C has been catching up. For the last two weeks the new model (F) has been the winner (with an asterisk: I didn’t put model F in my week 11 prediction article, but I did evaluate it after the week was done). Models D and E really haven’t done much positive, but last week they were definitely better than A or B. When the season is over I’ll want to write a script that will give me the accuracy for each week to see how much it varies from week to week, but for now I’m just trying to stay on top of these posts and put them up before the waiver deadline on Tuesdays.

No new models this week unless you haven’t read my Methodology article about TruScor (seriously, what is a trademark?) or the Week 12 Evaluation article. If you’re just here for the predictions, then the short version of Model F is that it ignores TDs but weights turnovers since they’re probably a little more predictable. Read this article, but then go read some of the Methodology articles and maybe check out some Data Exploration articles.

On to this week’s models!

Model A:

Model A parameters:

  • The team that’s playing
  • Who their opponent is
  • Their predicted score
  • Their opponent’s predicted score
  • Whether the team is home or away

Model A Notes:

  • Uses only data from 2016
  • Best model from weeks 8 and 9, 3rd best in week 10
  • Terrible in week 12

Model B:

Model B parameters:

  • The team that’s playing
  • Who their opponent is
  • Their opponent’s predicted score

Model B Notes:

  • Uses only data from 2016
  • Comparable results to Model A while reducing overfitting
  • Terrible in week 12

Model C:

Model C parameters:

  • The team that’s playing
  • Who their opponent is
  • Their predicted score
  • Their opponent’s predicted score
  • Whether the team is home or away

Model C Notes:

  • Uses data from 2015 and 2016
  • OK performance weeks 8 and 9, best model in week 10

Model D:

Model D parameters:

  • The team that’s playing
  • Who their opponent is
  • Their predicted score
  • Their opponent’s predicted score
  • Whether the team is home or away

Model D Notes:

  • Uses only data from the last 6 weeks

Model E:

Model E parameters:

  • The team that’s playing
  • Who their opponent is
  • Their predicted score
  • Their opponent’s predicted score
  • Whether the team is home or away
  • NEW: “Team:week” term (does the team have momentum?)
  • NEW: “Opponent:week” term (does the opponent have momentum?)
  • NEW: “week” term (basically a bookkeeping term if I want to use the other two)

Model E Notes:

  • Uses data for last 8 weeks
  • includes week + Team:week and Opponent:week terms for each team
  • New terms try to account for momentum

Model F:

Model F parameters:

  • The team that’s playing
  • Who their opponent is
  • Their predicted score
  • Their opponent’s predicted score
  • Whether the team is home or away

Model F Notes:

  • Uses data from 2015 and 2016
  • Best model in weeks 11* and 12
  • Ignores defense TDs but weights interceptions and fumbles to try to account for it

Week 13 model summary

Rank

Model.A

Model.B

Model.C

Model.D

Model.E

Model.F

1

TB

TB

DEN

TB

DEN

DEN

2

SEA

DEN

SEA

DEN

WAS

SEA

3

DEN

SD

PHI

IND

JAC

PHI

4

WAS

SEA

IND

SD

SD

WAS

5

SD

WAS

WAS

SEA

CAR

GB

6

PHI

PHI

TB

WAS

MIA

TB

7

IND

IND

GB

PHI

HOU

NE

8

BAL

BAL

NE

JAC

ARI

IND

9

GB

GB

CAR

MIA

DET

SD

10

NE

NE

ARI

CAR

NO

OAK

11

MIA

CHI

MIN

NE

TB

PIT

12

MIN

MIA

SD

DET

OAK

JAC

13

KC

MIN

KC

CHI

CIN

BAL

14

CAR

KC

BAL

BAL

IND

CAR

15

ARI

ARI

JAC

PIT

SF

ARI

16

CHI

CAR

MIA

GB

PIT

CIN

17

JAC

OAK

PIT

NYG

SEA

KC

18

HOU

ATL

OAK

CIN

LA

CHI

19

SF

SF

HOU

SF

PHI

HOU

20

CIN

HOU

ATL

ARI

NYJ

MIA

21

OAK

NYJ

SF

HOU

NYG

DAL

22

PIT

JAC

LA

NYJ

NE

NYG

23

NYJ

CIN

DAL

LA

GB

NYJ

24

ATL

DET

NYG

OAK

CHI

MIN

25

DET

PIT

CHI

DAL

BAL

NO

26

LA

BUF

CIN

NO

KC

ATL

27

NYG

NO

NYJ

KC

MIN

SF

28

DAL

DAL

NO

MIN

BUF

LA

29

BUF

NYG

BUF

ATL

DAL

BUF

30

NO

LA

DET

BUF

ATL

DET

I was shocked at how much these models like Tampa Bay until I looked at them more closely. They’ve been inconsistent, sure, but have scored more points overall than LA. They’re going up against San Diego who have given up the 7th most points to defenses.

Washington is going up against Arizona this week. ARI have given up the 3rd most points, but WAS has only hit double digits once this year. 

Green Bay is another weak team with an OK matchup that Model F likes. HOU have given up the 8th most points, but GB has only hit double digits twice and has 7 games with 5 points or fewer.

San Diego and Baltimore have both popped up in the top 10 in Models A and B and are a little lower but still viable options in models C and F. They have about the same number of points overall this year and going up against teams ranked in the middle of the pack for giving up points to defenses. They’re both a bit more consistent than any of the other choices above with 5 games over 10 points for SD and 6 for BAL.

This is the last week of the fantasy regular season in my league, and I need to win to make it into the playoffs. These are all good options this week. Start them with confidence.

Unless you’re Carly. Then start them with the knowledge of your inevitable loss.

Week 12 K Evaluation

Week 12 DEF Evaluation