Data scientist, physicist, and fantasy football champion

Week 8 DEF Predictions

Model H did very well last week because there were so many really high-scoring teams. Model F has been slightly better throughout the season, and Yahoo’s experts slightly better than that.

Let’s see some predictions for this week:

Model Type Score OppScore Opp HomeAway DataYears PointSystem
A LM X X X X 2017 IOUH/Standard
C LM X X X X 2015-2017 IOUH/Standard
F LM X X X X 2015-2017 TruScor
G Gibbs Sampler X X X X 2015-2017 TruScor
H LM X X X X 2017 TruScor
I Many LMs X X X X 2015-2017 IOUH/Standard

 

Model A:

.Model A-1.png


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 2017
Standard points
Did well in 2016, but didn’t stand out

Model C:

.Model C-1.png


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 2015-2017
Standard scoring
Did well in 2016, but didn’t stand out

Model F:

.Model F-1.png


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 2015-2017
Truscor ignores defense TDs but weights interceptions and fumbles to try to account for it
I really liked this one in 2016. It was marginally better than A and C

Model G:

.Model G-1.png


Model G 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 G Notes:

Uses data from 2015-2017
Truscor ignores defense TDs but weights interceptions and fumbles to try to account for it
Uses Bayesian statistics (Gibbs sampler) to create a model
Would have done very well in 2016 based on simulation

Model H:

.Model H-1.png


Model H 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 H Notes:

Uses only data from 2017
Uses TruScor points to reduce the effect of TDs

Model I:

.Model I-1.png


Model I parameters may include:

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 I Notes:

Data from 2015-2017
This model tries to predict all of the components of scoring separately (interceptions, sacks, etc.). Each component comes from a linear model containing only the significant terms (p < 0.05) from the list of possible terms above.

Week 8 model summary:
 

Rank A C F G H I
1 SEA MIN MIN MIN SEA MIN
2 PIT KC KC KC BAL KC
3 BAL TB TB CIN MIN CIN
4 DET PHI PHI PHI KC PHI
5 NO CIN CIN TB PIT TB
6 KC SEA BAL BAL TB ATL
7 TB ATL SEA SEA IND BAL
8 CIN BAL ATL ATL DAL SEA
9 DAL NO CAR NO NO NO
10 MIN MIA NO PIT DET MIA
11 CHI CAR PIT CAR PHI PIT
12 IND PIT MIA NE MIA CAR
13 HOU HOU NE MIA CIN NE
14 WAS DEN HOU WAS ATL DEN
15 MIA WAS WAS DAL WAS WAS
16 ATL NE DET HOU HOU DAL
17 PHI DAL DAL BUF CAR HOU
18 BUF DET DEN DET BUF DET
19 DEN LAC OAK OAK SF BUF
20 SF BUF BUF DEN NE LAC
21 LAC IND IND IND DEN IND
22 CAR OAK LAC NYJ LAC OAK
23 NYJ CLE NYJ LAC CHI NYJ
24 NE CHI CLE CLE NYJ CLE
25 OAK NYJ SF SF OAK CHI
26 CLE SF CHI CHI CLE SF

Conclusions

Model F has 5 top Defenses for the week: MIN, KC, TB, PHI, and CIN. Of those, PHI and CIN are likely available. Model H puts Sea, PIT, and IND up there as well. Let’s take a quick look at potential streamers for the week:

  • TB (home against CAR): Great matchup, but TB hasn’t done much this year. They’re allowing a ton of points each game, and I don’t see this week much different. Not sure why my models love TB so much.
  • CIN (home against IND): GREAT matchup! IND gave up 10 sacks last week. CIN had a tough week last week, but the start of their season was solid.
  • NO (home vs. CHI): Trubisky hasn’t been asked to throw very much, but he may have to against a dynamic NO offense. NO’s defense has donve very well in plus-matchups this year, and against a rookie QB they might find some success if the game is going their way.
  • PIT (at DET): If Pittsburgh is still available, please go get them. They’ve been one of the best fantasy defenses this year and Detroit has allowed 5 or 6 sacks each of the last 3 games.

And one caution:

  • ATL (at NYJ): Why are they even in the top-12 of my models? The Jets are in the middle of the pack for points to defenses, and Atlanta hasn’t done anything to stand out this year. I’m always surprised by at least one of the model choices each week, and I guess this week it’s Atlanta.

Don’t forget to check for teams coming off bye weeks. This week that’s Detroit and Houston. Both are in OK matchups this week, but have been good overall. Unless the dummies in you leagues stashed them (like they did in mine. WHY?), there’s a chance to grab them and stream them through a tough matchup as there will be easier ones coming.

Week 8 DEF Results

Week 8 QB Predictions