NFL 2018 Model Tracker
 Share
The version of the browser you are using is no longer supported. Please upgrade to a supported browser.Dismiss

View only
 
ABCDEFGHIJKL
1
2
RISKNETROIWLPWIN %AVG PEDGE
3
Overall238.1643.1018.10%7547660.94%50.91%10.02%
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
What is this?
25
This is a Google Sheet that shows the results of all of the NFL model plays from 2018.
26
How were the model plays shared?
27
Model plays were shared for free on Reddit and Twitter.
28
How did it work?
29
The model pulled in various data points and gives each of the 32 teams category ratings for pass offense, pass defense, rush offense, rush defense, and special teams. Those ratings were then analyzed to see how performance in each of those categories correlated with overall team strength. Those correlations were then translated into unique correlation coefficients for each category, which were then used in a formula I created to estimate expected wins. The correlation coefficients were dynamic from week-to-week and recalculated when new data was entered.
30
How were model plays determined?
31
For each matchup, each team's expected win probability for a single game and home court advantage were combined to spit out a model spread which is then compared to the Vegas line. Any disagreement of two points or more became a model play. 2 points of disagreement was a 1 unit play; 3 points of disagreement was a 1.5 unit play; 5.4 points of disagreement was a 2.7 unit play etc. The units were part of the "to win amount".
32
What if someone had a different line?
33
If the line moved in favor of the model play, 0.5 units were meant to be added to the "to win" amount for each point the spread moved. If it moved against the model play, 0.5 units were meant to be subtracted for each point. If the process of subtracting units brought a play below 1 unit, it was no longer a model play.
34
What book did you use?
35
I used Nitrogen lines for the NFL model.
36
37
38
39
40
41
42
43
44
45
46
47
Loading...