Don't worry about the "trend" numbers this week, as they are heavily influenced by the removal of over 40 teams are not playing this season. There are 77 total teams.

Sun Belt

27. Appalachian State
41. Louisiana
53. Coastal Carolina
55. Georgia Southern
59. Troy
61. Arkansas State
66. South Alabama
68. Georgia State
73. ULM
75. Texas State

Others

9. UCF
17. Iowa State
20. Kansas State
25. Memphis
30. West Virginia
35. Marshall
38. Arkansas
43. Tulsa
44. Army
49. BYU
50. UAB
52. SMU
63. ECU
67. Southern Miss
69. Liberty
70. Charlotte
76. MTSU

https://www.espn.com/college-football/fpi

:(

Love being one place higher than a team that got shutout by Army.

(09-06-2020 12:23 PM)slycat Wrote:  :(

Love being one place higher than a team that got shutout by Army.

ESPN has always placed more emphasis on prior games / seasons. It takes a while to influence the algorithm in a positive way. This is true in the reverse as well, but it's not quite as strong. Two or three bad games will plummet you downstream faster than two or three good games will allow you to climb the ladder.

Solid showing by Texas State for sure, but wins have 3x the impact that competitive losses do, even if being competitive is a step up from year's passed.

(09-06-2020 12:59 PM)TroyFootball05 Wrote:  

(09-06-2020 12:23 PM)slycat Wrote:  :(

Love being one place higher than a team that got shutout by Army.

ESPN has always placed more emphasis on prior games / seasons. It takes a while to influence the algorithm in a positive way. This is true in the reverse as well, but it's not quite as strong. Two or three bad games will plummet you downstream faster than two or three good games will allow you to climb the ladder.

Solid showing by Texas State for sure, but wins have 3x the impact that competitive losses do, even if being competitive is a step up from year's passed.

ANY stat based ranking is going to have previous season data included for the early part of the season. Otherwise there's no data to base on. They all filter it out/reduce weighting on it as the season progresses though. SP+ (done by Bill Connelly now at ESPN) uses last season data until about midseason when there's a much better picture of what the current team looks like.

You're going to see some highly reactive moves by these rankings after games like Army and Marshall where they were just so lopsided, at least in FPI. I know SP+ has been modified to be less impacted by these as early season outcomes tend to be a bit all over the place.

(09-06-2020 01:39 PM)eaglewraith Wrote:  

(09-06-2020 12:59 PM)TroyFootball05 Wrote:  

(09-06-2020 12:23 PM)slycat Wrote:  :(

Love being one place higher than a team that got shutout by Army.

ESPN has always placed more emphasis on prior games / seasons. It takes a while to influence the algorithm in a positive way. This is true in the reverse as well, but it's not quite as strong. Two or three bad games will plummet you downstream faster than two or three good games will allow you to climb the ladder.

Solid showing by Texas State for sure, but wins have 3x the impact that competitive losses do, even if being competitive is a step up from year's passed.

ANY stat based ranking is going to have previous season data included for the early part of the season. Otherwise there's no data to base on. They all filter it out/reduce weighting on it as the season progresses though. SP+ (done by Bill Connelly now at ESPN) uses last season data until about midseason when there's a much better picture of what the current team looks like.

You're going to see some highly reactive moves by these rankings after games like Army and Marshall where they were just so lopsided, at least in FPI. I know SP+ has been modified to be less impacted by these as early season outcomes tend to be a bit all over the place.

My point was that ESPN FPI is one of the least likely to give the benefit of the doubt, and tend to be slower moving than other rating systems. There is also more of P5/G5 separation in their ratings than others, which makes it harder for G5 teams to move up the ladder.

I hate computer rankings or power rankings that are too knee-jerk after week 1, but that also means not tanking teams like SMU that don't come close to covering. And it looks like the FPI did that accounting for the 53 (?) teams that dropped from the FPI.