Analyzing the Rankings – Week 8

Tom here with commentary on the rankings changes from Week 8.

Stability & Variance

Two big movers this week, Philadelphia and Arizona, each moving over ten ranks. The overall ranking changes have fallen again, demonstrating more overall stability in the rankings after the shakeup in Week 6.


One point of comparison with Week 8 is Week 5. Approximately the same overall number of rank changes (86 vs. 84), but a higher variance this week. The difference in variance means that back in Week 5 lots of teams moved an average amount, whereas in Week 8 most teams had a small change in rank while a few teams had major shifts.

Pick Confidence

Last week the pick confidence didn’t do well. The top confidence picks were less reliable than the low confidence picks. We’ll see if the shuffle in the ranks gives us greater accuracy in making picks.

Matchup
(Winner-Loser)
“Confidence”
(out of 100)
EdgePower comparison
(predicted winner – predicted loser)
Result
Atlanta-Washington 40.25 60.98-20.73
Green Bay-Tampa Bay 38.42 54.27-15.85
New Orleans-Carolina 38.41 71.95-33.54
New England-Miami 15.85 65.24-49.39
Jacksonville-Kansas City 15.24 48.78-33.54
Indianapolis-Houston 12.19 65.24-53.05
Dallas-Philadelphia 10.97 65.24-54.27
San Francisco-Tennessee 10.36 47.56-37.20
Denver-Pittsburgh 8.53 70.73-62.20
Cincinnati-Baltimore 4.87 66.46-61.59
San Diego-NY Giants 4.27 56.71-52.44
Seattle-Detroit 1.22 43.29-42.07
Arizona-Chicago 0 51.83-51.83

As usual, I’ll return to fill in the results as they roll in.

One Response to Analyzing the Rankings – Week 8

  1. […] Last week, pick confidence worked pretty well, getting only one high confidence pick wrong (Green Bay losing to Tampa Bay), while getting two lower confidence picks incorrect. As the season progresses, and the rankings of the teams become more accurate according to Beatpaths, we should see this pattern continue: about 2/3rds of the incorrect picks consigned to the low confidence games. […]

Leave a Reply

Your email address will not be published. Required fields are marked *