*Power Ratings - Week 11

[speculadora]

For any who missed the first week I did this, here's a quick run down of how the ratings are calculated and what they mean:

I wanted to be as scientific about this as I could be, and I remembered that Bill James - who is mostly famous for baseball data analysis - has a formula he uses to make NFL power rankings. I won't even bother trying to walk through it because Bill does so quite a bit more effectively than I could anyway. The only significant changes I made were 1) making home field advantage worth 4.5 points (based on some other math I did) and 2) what I describe in the following paragraph. Anyway, here's the link breaking down the methodology: https://www.billjamesonline.com/article808/

I made one small change from the link above. I kept the raw ratings, but I also wanted everything to expressed as a number above or below an average. So I divided all numbers by the league average rating and multiplied by 100 to get a "Rating+".

The two sets of rankings are displayed below. The raw ratings represent a better comparison if you're looking at two teams. As an example, Team A has a rating of 107 and Team B has a rating of 102. So if you wanted to compare those teams you'd basically say Team A is 5 points better than Team B. If you're looking at the scaled ratings, Team A sits at 103 versus Team B's 98. In this case Team A is about 3% better than league average whereas Team B is about 2% worse than league average. The differences between teams are fairly close to the raw system, but the scaled system is much cleaner in my opinion.

If you're paying attention you'll notice that every team saw a decrease in their "raw" rating from Week 10 to Week 11. Part of this is just an annoying function of the code. In essence it's trying to find a balance whereby all of the game margins line up. Using the Butchers as an example, they need to be roughly 1 point better than Baltimore, but 1 point worse than Colorado, who is roughly 33 cumulative points worse than Baltimore. So the loop iterates over and over and tries to find the ratings that best satisfy those conditions. So sometimes the total sum of ratings will be high for that to work out, and sometimes it will be low. In other words, the Wraiths aren't 5 points worse than they were in Week 10. They're 5.3 points better than the Otters as of Week 11, or 8.2 points better than the Outlaws as of Week 11, etc. These differences between teams are roughly more comparable from week-to-week than comparing a team to itself.

Raw Ratings
Ranking - Team - Rating
1. Yellowknife Wraiths - 111.5
2. Orange County Otters - 106.2
3. Arizona Outlaws - 103.3
4. Baltimore Hawks - 102.9
5. Austin Copperheads - 101.3
6. Chicago Butchers - 98.3
7. Philadelphia Liberty - 97.8
8. New Orleans Second Line - 96.4
9. Colorado Yeti - 94.1
10. San Jose Sabercats - 89.1

Now the scaled ratings we can compare. I think. They're scaled to the average raw rating for that week, so they should be pretty comparable.

Scaled Ratings
Ranking - Team - Rating+ (Change)
1. Yellowknife Wraiths - 111.4 (0.0)
2. Orange County Otters - 106.1 (+1.3)
3. Arizona Outlaws - 103.2 (+0.4)
4. Baltimore Hawks - 102.8 (+0.1)
5. Austin Copperheads - 100.5 (+0.7)
6. Chicago Butchers - 98.2 (-0.5)
7. Philadelphia Liberty - 97.7 (0.0)
8. New Orleans Second Line - 96.3 (-1.3)
9. Colorado Yeti - 94.0 (-1.2)
10. San Jose Sabercats - 89.0 (+0.5)

Predicting Week 12 Spreads
All lines are expressed through the home team and are just calculated based on the difference in raw rating with an adjustment for HFA, rounded to the nearest "half-integer". For any unfamiliar with what these mean, a -19 means the team is favored by 19, a +6 means the team is a six point underdog.

@ : Hawks -11
@ : Yeti +4.5
@ : Liberty -13
@ : Otters -12.5
@ : Copperheads +5.5

[shadyshoelace]

Ok sure but where are the over/unders mr stat man