I Created a CFB Ranking System

#177
#177
WEEK 8 Top 25 (with AP/Coaches):

1. Alabama (8-0) - 185.0365 (1/1)
2. Penn St (7-0) - 183.5305 (2/2)
3. Georgia (7-0) - 178.9749 (3/3)
4. UCF (6-0) - 178.2233 (18/17)
5. Ohio St (6-1) - 172.0976 (6/6)
6. TCU (7-0) - 170.9337 (4/4)
7. Clemson (6-1) - 170.3092 (7/7)
8. Notre Dame (6-1) - 169.3296 (9/10)
9. Wisconsin (7-0) - 166.7159 (5/5)
10. Miami (6-0) - 165.9202 (8/8)
11. Oklahoma St (6-1) - 155.5470 (11/12)
12. Michigan St (6-1) - 152.2889 (16/18)
13. Washington (6-1) - 151.1555 (12/11)
14. Memphis (6-1) - 150.6664 (24/UR)
15. NC State (6-1) - 149.6168 (14/15)
16. Oklahoma (6-1) - 148.5437 (10/9)
17. Virginia Tech (6-1) - 146.4401 (13/13)
18. Mississippi St (5-2) - 145.8997 (UR/UR)
19. USC (6-2) - 145.8054 (21/21)
20. Stanford (5-2) - 142.8059 (20/20)
21. Washington St (7-1) - 141.2867 (15/16)
22. Auburn (6-2) - 139.8544 (19/19)
23. South Florida (7-0) - 139.7347 (17/14)
24. Michigan (5-2) - 138.9193 (UR/25)
25. Boise St (5-2) - 138.6915 (UR/UR)
 
#178
#178
The Rest of the SEC:

33. Texas A&M (5-2) - 131.9008 (UR/24)
34. LSU (6-2) - 131.5133 (23/23)
36. South Carolina (5-2) - 128.7729
44. Kentucky (5-2) - 121.6205
58. Florida (3-3) - 111.5586
77. Tennessee (3-4) - 100.3500
78. Ole Miss (3-4) - 98.3096
83. Arkansas (2-5) - 95.6873
89. Vanderbilt (3-4) - 91.8757
100. Missouri (2-5) - 83.3634
 
#179
#179
Despite the big loss, UT jumped in both score and ranking because their already strong SOS got even stronger by adding Bama, coupled with wins by Georgia Tech and UMass.
 
#181
#181
BW, I know you set this up to prove that the system doesn't work, but to me it seems like it does. Any given week it may be off, but over a whole season it seems to ring pretty true.
 
#182
#182
LouderVol said:
BW, I know you set this up to prove that the system doesn't work, but to me it seems like it does. Any given week it may be off, but over a whole season it seems to ring pretty true.
Click to expand...

It does ebb and flow. But just using this week as an example, I wouldn't have OSU or UCF in my top 5. I've watched the games, and I can reasonably judge each team's strength of schedule. Heck, I'm not sure UCF would be in my top 20. Memphis certainly wouldn't be in my top 15.
 
Last edited:
#183
#183
bamawriter said:
It does ebb and flow. But just using this week as an example, I wouldn't have OSU or UCF in my top 5. I've watched the games, and I can reasonably judge each team's strength of schedule. Heck, I'm not sure UCF would be in my top 20. Memphis certainly wouldn't be in my top 15.
Click to expand...

I guess my point is that a computer based system is just as valid as another system. they are all going to be flawed, there will always be upsets and people that come out of no where. So I don't think your system proves that computers are bad, just that the game of football is impossible to predict.
 
#184
#184
LouderVol said:
I guess my point is that a computer based system is just as valid as another system. they are all going to be flawed, there will always be upsets and people that come out of no where. So I don't think your system proves that computers are bad, just that the game of football is impossible to predict.
Click to expand...

Here's why computer and math systems are terrible: they don't have the ability to account for attrition.

For example: Say Saquon Barkley suffers a tear in practice this week and misses the rest of the season. If Ohio State beats PSU without Barkley, they will receive the same credit for that PSU game that the previous 7 teams did. But we all know that Penn State cannot possibly be the same team without Barkley as they were with him. A human being can process that information and rank teams accordingly. Formulas can't.
 
#185
#185
bamawriter said:
Here's why computer and math systems are terrible: they don't have the ability to account for attrition.

For example: Say Saquon Barkley suffers a tear in practice this week and misses the rest of the season. If Ohio State beats PSU without Barkley, they will receive the same credit for that PSU game that the previous 7 teams did. But we all know that Penn State cannot possibly be the same team without Barkley as they were with him. A human being can process that information and rank teams accordingly. Formulas can't.
Click to expand...

OSU is still going to get credit for beating a good team in the human polls as well. and in the computer world at the end of the day that will work out as PSU starts losing more games or not winning by the same amount. so by the end OSU's victory looks not as good as it did.

as I was saying any given weak yes computer = BS. but over the season it should all play out.
 
#186
#186
LouderVol said:
OSU is still going to get credit for beating a good team in the human polls as well. and in the computer world at the end of the day that will work out as PSU starts losing more games or not winning by the same amount. so by the end OSU's victory looks not as good as it did.

as I was saying any given weak yes computer = BS. but over the season it should all play out.
Click to expand...

I see what you're saying, but I simply don't agree with that last part.

Let's take it a step further. Let's say Barkley, McSorley, and a couple of offensive linemen go down for the season. PSU proceeds to lose to Ohio St, Michigan State, and Nebraska. They don't make the Big 10 title game at 9-3.

All seven teams that Penn State has beaten will finish the year with Penn State's 9-3 record factored into their scores, even though the Penn State they lost to was considerably better than the Penn State OSU, MSU, and Nebraska beat. You can't program an objective formula to account for losing critical players.

So it's not a one week problem; it's an entire season problem. If a team stays relatively healthy, then yeah, it's no big deal to give their opponents the same credit from Week 1 to Week 13. But if a team becomes a MASH unit by season's end, then you have to place a subjective standard that credits their Week 1 opponent more than Week 13.
 
#187
#187
bamawriter said:
I see what you're saying, but I simply don't agree with that last part.

Let's take it a step further. Let's say Barkley, McSorley, and a couple of offensive linemen go down for the season. PSU proceeds to lose to Ohio St, Michigan State, and Nebraska. They don't make the Big 10 title game at 9-3.

All seven teams that Penn State has beaten will finish the year with Penn State's 9-3 record factored into their scores, even though the Penn State they lost to was considerably better than the Penn State OSU, MSU, and Nebraska beat. You can't program an objective formula to account for losing critical players.

So it's not a one week problem; it's an entire season problem. If a team stays relatively healthy, then yeah, it's no big deal to give their opponents the same credit from Week 1 to Week 13. But if a team becomes a MASH unit by season's end, then you have to place a subjective standard that credits their Week 1 opponent more than Week 13.
Click to expand...

Do you think that ever happens in the human world?

TN last year is the perfect example of the MASH unit you speak of. UGA, Florida, Kentucky & Mizzou losses to us all counted the same. and we dropped (rightly so) for losing to SC & VU even though we were running out of players when we lost to them.

so I am not seeing the human system as any better.
 
#188
#188
LouderVol said:
Do you think that ever happens in the human world?

TN last year is the perfect example of the MASH unit you speak of. UGA, Florida, Kentucky & Mizzou losses to us all counted the same. and we dropped (rightly so) for losing to SC & VU even though we were running out of players when we lost to them.

so I am not seeing the human system as any better.
Click to expand...

Vandy's win was subjectively worse
than Florida's loss.
 
#189
#189
which is kinda my point. the Computer is objective to the point where it seems subjective to us. especially since the games never play out like they should on paper.

the human side applies subjectivity all over the place. look at the playoff voters. it will be really interesting to see what happens if Bama and Georgia both go undefeated until the SEC, one loses a tight match, I am willing to bet the committee leaves them off. otherwise you will offend two of the P5 conferences by leaving them out over a team that just lost.
 
#190
#190
LouderVol said:
which is kinda my point. the Computer is objective to the point where it seems subjective to us. especially since the games never play out like they should on paper.
Click to expand...

When each of 130 teams plays only 12 or 13 games, subjectivity is preferrable to objectivity. You can't objectively compare that many teams with such a small sample.
 
#191
#191
bamawriter said:
When each of 130 teams plays only 12 or 13 games, subjectivity is preferrable to objectivity. You can't objectively compare that many teams with such a small sample.
Click to expand...

I gave you hell a year ago or so. Mocked you in fact. I was wrong. I am sorry. I appreciate your contributions to the board.
 
#193
#193
bamawriter said:
When each of 130 teams plays only 12 or 13 games, subjectivity is preferrable to objectivity. You can't objectively compare that many teams with such a small sample.
Click to expand...

and yet the objective approach falls remarkably close to the subjective.

Especially as you are comparing one computer model to averaged human rankings. AP, coaches, CFP are averages of individual votes. average some computer ones together and you would increase the similarities.
 
#194
#194
LouderVol said:
and yet the objective approach falls remarkably close to the subjective.

Especially as you are comparing one computer model to averaged human rankings. AP, coaches, CFP are averages of individual votes. average some computer ones together and you would increase the similarities.
Click to expand...

All true. But, as I've said before, one need look no further than the BCS. The worst pairings were always a result of the computers (2001, 2003 being the most obvious). That's why they reduced the computers in the overall formula.
 
#195
#195
bamawriter said:
All true. But, as I've said before, one need look no further than the BCS. The worst pairings were always a result of the computers (2001, 2003 being the most obvious). That's why they reduced the computers in the overall formula.
Click to expand...

Not a fan of Oklahoma I guess?
 
#197
#197
bamawriter said:
Don't have any real issue with them, but they shouldn't have been in that game in '03. Thanks to the computers, we got #s 2 and 3 in both polls playing for the right to be #1.
Click to expand...

because a more recent loss matters more than one against a worse team from a while ago?
 
#198
#198
LouderVol said:
because a more recent loss matters more than one against a worse team from a while ago?
Click to expand...

True, USC's loss was to a worse team in the form of 8-6 Cal. But it was by a field goal in 3 OTs.

Oklahoma didn't even compete with a 3-loss Kansas State team.

So USC might have lost to a "worse team," but OU wound up with a worse loss.
 
#199
#199
so USC had more time to beat Cal and couldn't.

so we are down to the Dobbs4Heisman "eye test" to make or break. :)
 
#200
#200
LouderVol said:
so USC had more time to beat Cal and couldn't.
Click to expand...

Okay. Compare apples-to-apples, then. After 60 minutes, USC was tied with Cal, and Oklahoma was down 4 touchdowns to KSU.

so we are down to the Dobbs4Heisman "eye test" to make or break. :)
Click to expand...

I've never had a problem with the "eye test" argument, except that D4H's eye seems to be watching events that only he can see.

In the case of 2003, I know that the eye test showed Oklahoma getting drilled by a good, but by no means great Kansas State team. I don't know what my formula would have returned for that year as I didn't go back that far. But if it were to return LSU and Oklahoma, it would be wrong.
 
New posts
You must log in or register to reply here.

VN Store



Back
Top