Created computer polls wanted feedback! Topic

You have to remember this is a computer poll not an eye test human poll.  Do I agree Iowa is the number 2 team in the nation, not really, but statistically they are, I don't think my model clearly benefits Iowa in any special way that would make it biased towards Iowa.
10/29/2015 5:42 PM
moranis has OSU, in small letters, tattooed on his *****.    Any poll that doesn't have them #1 is obviously wrong. 
10/29/2015 7:34 PM
Posted by MikeT23 on 10/29/2015 7:34:00 PM (view original):
moranis has OSU, in small letters, tattooed on his *****.    Any poll that doesn't have them #1 is obviously wrong. 
haha I'm a student there, guess I'm not a total 100% homer(I do think with JT we are the #1 team, but with Cardale we were barely top 15)
10/29/2015 8:09 PM (edited)

Nah.  moranis is just a cheerleader for whatever team he roots for.    He doesn't let facts get in the way.   See: "Flacco - greatest road QB ever" thread in NFL forum.

10/29/2015 8:10 PM
Now my post makes less sense due to your edit.
10/29/2015 8:11 PM
I have no issue with Clemson or LSU ahead of OSU as this point in the season.  Iowa is the one that doesn't make any sense.  They do have victories over Pitt, NW, and Wiscy, but they also have wins over 0-7 North Texas, 2-5 Iowa State (1 of the wins is a FCS school), and a FCS school (albeit a good, but still a FCS school). 
10/29/2015 8:39 PM
Pitt being down 20-3 at the half (if it holds) will probably help correct the Iowa factor a bit.
10/29/2015 8:40 PM
And don't get me wrong, I know computers sometimes yield weird results, but if the results venture too far from reality then it may be time to adjust the methodology.  That is something colonels couldn't quite understand for a long time.  That is why I mentioned a few outliers in each season to see if they were outliers because of the same methodology as it might need to be adjusted.  Having the correct #1 team at the end of the year isn't that difficult, but if you keep having outliers in your top 5, that might be indicative of a larger problem with the methodology.
10/29/2015 8:43 PM
Yep, I'm hoping adding the following fixes some outliers, I probably think the current method does have too many of those weird outliers, Sunday I'm free and will be changes methods a bit.  Probably
  • Weight Rushing 55-45 over passing in yardage offense and defense(get rid of the pass heavy teams that aren't great[rush offense has better correlation to success] and get rid of the teams that are so bad at defending the run that noone bothers passing against them like San Jose St.
  • Include Yards/play fand 3rd down conversion rate for total offense and defense equation maybe something like 60% scoring, 2% yardage, 10% 3rd down rate, and 10% yds/play.
  • Create a "special teams factor" which is short field drives to measure field position(as well as kickoff and punt coverage/return ability and turnover margin probably something like 60% turnover margin and 40% short field drives.(this is the one I'm most unsure of atm of what weight to use)  not sure what else is needed to measure "ST play", maybe FG kicking if there is a decent correlation.
I think those factors and then playing with the weights will help with the current outliers and provide a better outlook for ranking teams.

10/29/2015 9:01 PM
The fact that you have a bigger problem of Iowa-2/OSU-4 than Memphis-5 speaks volumes.    You just can't look at CFB without OSU glasses.
10/29/2015 9:02 PM
Memphis shouldn't be top 5, and probably shouldn't be top 10 at this point in the season.  Ole Miss is a nice win, but the second best win to this point is Cincinnati or South Florida and that doesn't cut it.  And it isn't like Memphis is rolling people.  That said they have Houston and Temple and if they win those probably get Temple again in the AAC title game.  They win out you could maybe make a case for them as top 5, especially if Temple beats ND. 
10/29/2015 9:17 PM
And I have no issue with OSU at 4, just can't see how Iowa is 2.  The OSU schedule hasn't been very good to this point, and the Buckeyes didn't exactly roll through the early schedule, but the last three games are MSU, UM, and Iowa.  That is where it gets meaty and where if OSU keeps winning will rise in the computer rankings. 

BTW, the old BCS computer guys have the top 5 as LSU, Clemson, Alabama, Ohio State, Michigan State followed by Iowa, TCU, Baylor, ND, and Florida.  Memphis is 11 ahead of Stanford.  Wolfe has Iowa at 1 and Colley has them at 5, none of the other 3 have Iowa in the top 10.  Sagarin has Memphis at 26, the other four rankings are 5, 6, 14, and 15.  BTW, it is Wolfe that has Memphis at 5 so perhaps these rankings here use a similar methodology as Wolfe.

http://www.usatoday.com/story/sports/ncaaf/rankings/2015/10/27/sec-big-ten-college-football-computer-composite-rankings-lsu-alabama/74637862/


10/29/2015 9:25 PM
Yeah, if Wolfe has Iowa 1 and Memphis 5, I suspect it's almost identical.      So much for inventing a new wheel.
10/29/2015 9:41 PM
Posted by MikeT23 on 10/29/2015 9:41:00 PM (view original):
Yeah, if Wolfe has Iowa 1 and Memphis 5, I suspect it's almost identical.      So much for inventing a new wheel.
well yeah with the advent of the personal computer, pretty much anyone can create a ranking system.  Bound to be lots of overlap.
10/30/2015 9:07 AM
BTW, colonels poll which he posts here and I finally just looked at has Iowa at 3 and Memphis at 6, though he still has Utah at 5 and OSU all the way back at 12.  All four of those seem really off.  Perhaps colonels needs to scrap his rankings again and start over.
10/30/2015 9:09 AM
◂ Prev 1|2|3|4...7 Next ▸
Created computer polls wanted feedback! Topic

Search Criteria

Terms of Use Customer Support Privacy Statement

© 1999-2024 WhatIfSports.com, Inc. All rights reserved. WhatIfSports is a trademark of WhatIfSports.com, Inc. SimLeague, SimMatchup and iSimNow are trademarks or registered trademarks of Electronic Arts, Inc. Used under license. The names of actual companies and products mentioned herein may be the trademarks of their respective owners.