Ray vs the World, part II

Apologies for not getting this out sooner, but I was struggling with how to present the rankings differences in a manner anyone could understand. Let's jump in, and if you're interested in how ND is #1 and Cincy is #15 I'm going to dive deeeeeep.

Ray vs the AP

I'm going to tell you up front that nearly every poll is just ignoring body of work in favor of recency and and ancient thinking. I have actually seen people advocating for all undefeated to be ranked ahead of any 1-loss team. The AP appears to buy into this thought process (though how much thought happens is unclear).


AP
RRDiff
1GeorgiaSEC32
2CincinnatiAmerican1513
3AlabamaSEC41
4OklahomaBig 121410
5OregonPac-121712
6Ohio StateBig Ten71
7Notre DameIndependent1-6
8Michigan StateBig Ten135
9MichiganBig Ten6-3
10Oklahoma StateBig 125-5
11Texas A&MSEC121
12MississippiSEC219
13Wake ForestACC185
14Brigham YoungIndependent11-3
15Texas-San AntonioC-USA3520
16AuburnSEC248
17HoustonAmerican5336
18BaylorBig 1216-2
19IowaBig Ten10-9
20WisconsinBig Ten2-18
21North Carolina StateACC3413
22Coastal CarolinaSun Belt8563
23Penn StateBig Ten9-14
24LouisianaSun Belt6945
25PittsburghACC316

Funny how the top teams are further from my ranks than the middle. That's very unusual.

Mean7.60
Variance322.92
SD17.97


SEC4.20
Big Ten-6.33
Pac-1212.00
Big 121.00
ACC8.00
American24.50

And there you go. The SEC with their usual favoritism in place, but look at the Pac 12. And the B1G is just being disrespected. 

Ray vs the CFI

The College Football Invitational committee seems to flip coins, or maybe they...


Anyway, the one thing we can count on is the inconsistency of the committee to follow its own guidelines.


CFP
RRDiff
1GeorgiaSEC32
2AlabamaSEC42
3OregonPac-121714
4Ohio StateBig Ten73
5CincinnatiAmerican1510
6MichiganBig Ten60
7Michigan StateBig Ten136
8OklahomaBig 12146
9Notre DameIndependent1-8
10Oklahoma StateBig 125-5
11Texas A&MSEC121
12Wake ForestACC186
13BaylorBig 12163
14Brigham YoungIndependent11-3
15MississippiSEC216
16NC StateACC3418
17AuburnSEC247
18WisconsinBig Ten2-16
19PurdueBig Ten8-11
20IowaBig Ten10-10
21PittsburghACC3110
22 SDSU
Mtn West308
23UTSAC-USA3512
24UtahPac-1223-1
25ArkansasSEC250

Strangely enough, they seem to be closer to what I have. Set aside Oregon and Cincinnati and it's not super ridiculous. I read somewhere that the committee selections most closely mirror the ESPN FPI, which is a retrodictive measure of performance just like mine is!

Mean2.400
Variance65.833
SD8.114


SEC3.00
Big Ten-4.67
Pac-126.50
Big 121.33
ACC11.33
American10.00
Mtn West8.00

Like I said, not as bad, but the B1G still being disrespected. And it's still early in the process for the committee.

Ray and Massey's Composite


Massey's
RRDiff
1GeorgiaSEC32
2AlabamaSEC42
3Ohio StateBig Ten74
4MichiganBig Ten62
5CincinnatiAmerican1510
6Notre DameIndependent1-5
7OklahomaBig 12147
8Oklahoma StateBig 125-3
9Michigan StateBig Ten134
10WisconsinBig Ten2-8
11Texas A&MSEC121
12MississippiSEC219
13IowaBig Ten10-3
14Wake ForestACC184
15OregonPac-12172
16Penn StateBig Ten9-7
17BaylorBig 1216-1
18AuburnSEC246
19Iowa StateBig 12278
20UTSAC-USA3515
21Brigham YoungIndependent11-10
22PurdueBig Ten8-14
23NC StateACC3411
24PittsburghACC317
25UtahPac-1223-2

Massey's has 82 ranking systems included this week. I'm starting to wonder if I should try to contact him and ask to be added? Anyway, we have relatively small differences, but bigger than I'd like considering the number of systems included.

Mean1.64
Variance49.16
SD7.01


SEC4.00
Big Ten-3.14
Pac-120.00
Big 122.75
ACC7.33
American10.00

The average difference is only 1.64, but we have an almost perfect offset between the SEC and B1G. Yet when we include my ranks, there is still enough variance to move 2 teams!


Massey
w/RRDiff
1GeorgiaSEC10
2AlabamaSEC20
3Ohio StateBig Ten30
4MichiganBig Ten40
5CincinnatiAmerican50
6Notre DameIndependent60
7OklahomaBig 1270
8Oklahoma StateBig 1280
9Michigan StateBig Ten101
10WisconsinBig Ten9-1
11Texas A&MSEC110
12MississippiSEC120
13IowaBig Ten130
14Wake ForestACC140
15OregonPac-12150
16Penn StateBig Ten160
17BaylorBig 12170
18AuburnSEC180
19Iowa StateBig 12190
20UTSAC-USA200
21Brigham YoungIndependent210
22PurdueBig Ten220
23NC StateACC230
24PittsburghACC240
25UtahPac-12250

Irish vs Manbearpigs, part II

I caught grief today from some character who doesn't read. "How can ND be #1 and UC be #15?" First, we shall revisit my sandbag analogy. Each team in FBS football has a stack of empty sandbags. For every win, they fill a sandbag with a specific amount of sand they get from the team they beat. The better the team, the more sand goes into your sandbag. If you lose, the sandbag for the team you played stays empty. Also, for each team you beat you get some sand from the teams they beat and some from the teams they beat. So there are 3 win/loss sources of sand, plus 1 source from scoring differential.

I started laying out the win tree for these teams and realized that it's a huge list of teams to go 3 deep like my system does (around 720 teams for ND and Cincy both, so nearly 1500 total!). I'm sure given enough time I could automate the process, but I can't right now. I'll just do one team to 3 levels and you can extrapolate from there. Remember that each team you beat contributes their own sand, sand from the teams those teams beat, and sand from the teams those teams beat.

Notre Dame

    • Florida State 
    • Toledo  
    • Purdue
    • Virginia Tech 
    • USC  
    • North Carolina 
    • Navy
    • Wisconsin
        • Eastern Michigan
        • Illinois
        • Army
        • Iowa
        • Rutgers
        • Purdue
            • Oregon State
            • UConn
            • Illinois
            • Iowa 
            • Nebraska
            • MSU

Cincinnati

    • Miami (OH)
    • Murray State 
    • Indiana 
    • Temple 
    • UCF 
    • Navy 
    • Tulane 
    • Tulsa
    • Notre Dame
        • Florida State
        • Toledo  
        • Purdue 
        • Va Tech 
        • USC  
        • UNC
        • Navy
        • Wisconsin
            • Eastern Michigan
            • Illinois
            • Army
            • Iowa
            • Rutgers
            • Purdue

Each team in the win tree contributes some sand. 

  • Cincinnati has 0.8170 tons of sand in 9 sandbags currently. 1 sandbag is empty because FCS schools have no sand to give and UC played Murray State.
    • 0.3564 tons directly from the wins over Miami (OH) through Notre Dame
      • 0.1983 tons from the next layer (like the Florida through Wisconsin sand ND collected)
        • 0.1902 tons from the 3rd layer (like Eastern Michigan through Purdue sand Wisconsin collected)
    • .0721 tons from their scoring differential against all teams played
  • Notre Dame has 1.2424 tons of sand
    • 0.5496 tons from FSU through Wisconsin wins
      • 0.3154 tons from the next layer like the Eastern Michigan through Purdue sand Wisconsin collected
        • 0.2945 tons from the next layer like the Oregon State through MSU sand Purdue collected.
    • 0.0830 tons from their scoring differential against all teams played

So, each team downline is worth a specific amount of sand. My system changes that amount on a weekly basis based on game outcomes. When Wisky was 1-3 they were worth less than they are now at 6-3, which pretty obvious. What isn't obvious is how much sand the Badgers have collected over the past 5 weeks, and that keeps getting added to the sandbags owned by Notre Dame, PSU, and Michigan (the three teams to beat Wisky).

Intuitively this all makes sense if you just look at schedules. Notre Dame has one less win than Cincinnati, but the total amount of sand in those sandbags weighs more than the sandbags Cincinnati has even with the full bag from the Irish. ND's opponents have 11 more wins than Cincy's. 

Notre Dame 1st lvl pts W L
Florida State  0.0402 1 8
Navy 0.0949 2 7
North Carolina  0.3197 5 4
Purdue 0.3923 6 3
Toledo   0.1231 4 5
USC   0.1470 4 5
Virginia Tech  0.1650 4 5
Wisconsin 0.4701 6 3
Cincinnati 0.0000 9 0
TOTAL 1.7523 41 40


Cincinnati 1st lvl pts W L
Indiana  0.0359 3 7
Miami (OH) 0.1248 4 5
Murray State  0.0000 0 0 FCS,  no W/L
Navy  0.0949 2 7
Notre Dame 0.5496 8 1
Temple  0.0795 3 6
Tulane  0.0000 1 8
Tulsa -0.0231 3 6 Lost to UC Davis
UCF  0.2402 6 3
TOTAL 1.1018 30 43


I feel like this is a pretty stark picture of why UC hasn't earned a top 4 (or top 10 for that matter) spot in my system. And to drive the point home, here's 3-loss Wisconsin.

Wisconsin 1st lvl pts W L
Penn State 0.0000 6 3
ND 0.0000 8 1
Michigan 0.0000 8 1
Eastern MI 0.1496 6 3
Rutgers 0.1701 4 5
Army 0.2192 5 3
Illinois 0.2292 4 6
Iowa 0.3983 7 2
Purdue 0.3923 6 3
TOTAL 1.5587 54 27

If you would like to think of it differently, how about a glass of water?

  • Each team has 2 glasses
    • 1 big empty glass that gets filled by wins
    • 1 small with water equal to its ranking, it is much smaller than the empty glass
      • #1's glass is FULL 
  • After 4 weeks, we can start distributing water.
    • UGA
      • Big glass gets some water from Clemson (0.71oz), UAB (0.28oz), South Carolina (0.36oz), and Vandy (a droplet).
      • Small glass is nearly full at #2.
    • The only team whose big glass has more water is Michigan (remember, week 4).
      • Big glass gets water from Western MI (0.54oz), Washington (0.23oz), NIU (0.67oz), and Rutgers (0.55oz).
      • Small glass is full @ #1
  • So if someone beat UGA in week 5, they would get about 1.0oz of water from their small glass, and Michigan would have to give up a little bit more from their small glass.
  • Your win glass level changes as the teams you beat perform throughout the season. The teams who beat Wisky early are getting extra water added every time the Badgers win.
I swear, I need to start a YouTube channel just so I can explain this with visual aids. 
 
Anyway, sorry for the length and the redundancy of the explanation. I'm looking forward to this weekend. As we get down to the wire, teams will get exposed. Just a matter of time! 

No comments:

Post a Comment

Latest Rankings

Latest Rankings

How Ray’s Rankings are achieved: My system is based on the Ohio State High School Football playoff calculator (and I'm sure other state...