conference rankings per bell curve analysis

[beorach]

I don't have the time to watch every game of every team in the "Power 5" conferences so, to get a decent idea of how they compare, I made bell curves of the stats (for all 128 FBS teams) within the categories of passing defense, passing offense, rushing defense, rushing offense, scoring defense, scoring offense, total defense, total offense, and turnover margin. I then calculated an overall ranking for teams based on their stats posted within games (against only FBS teams - no stats versus FCS teams were included) relative to those curves.

 

E.g., Nebraska posted a stat of 5.8 yards per attempt in the passing defense category. The national average value for that stat is 7.08 with a standard deviation of 0.83. Considering 5.8 is less than the average minus one standard deviation, I rate that stat as excellent. Had the posted value just been less than the standard deviation minus half a standard deviation, I'd have rated that as good. Had it been within the range set by the average plus or minus half a standard deviation, I'd have rated it as average. I do this for every stat it makes sense to across those categories I listed above and keep a running tally of performance to issue a final grade.

 

I'm going to share the rankings below with and without a strength of schedule factor applied. The data I used was from before conference championship games were played.

 

WITH SOS

 

The average performance of B1G teams (relative to these bell curves) puts them at 4th out of the 5 "power" conferences. Only the ACC's stats are weaker overall. The Big XII is practically equal with the Pac-12 and the SEC takes the top spot.

 

Looking a little deeper, I took the averages and standard deviations of my ratings in order to determine which "P5" teams have bad or great stats (in terms of statistically significant figures within these categories) overall.

 

The B1G has three teams with great stats (Michigan State, Ohio State, and Wisconsin in descending order), three with poor stats (Indiana, Rutgers, and Illinois in descending order), and eight with average stats (Nebraska, Iowa, Minnesota, Michigan, Penn State, Northwestern, Purdue, and Maryland in descending order).

 

The SEC has five teams with great stats (Alabama, Georgia, Auburn, Mississippi State, and Ole Miss in descending order), one with poor stats (Vanderbilt), and eight with average stats (LSU, Arkansas, Florida, Missouri, Tennessee, Kentucky, South Carolina, and Texas A&M in descending order).

 

The ACC has no team with great stats, two with poor stats (North Carolina and Wake Forest in descending order), and twelve with average stats (Clemson, Georgia Tech, Louisville, Miami, Boston College, Florida State, Virginia, Virginia Tech, Pittsburgh, Duke, Syracuse, and North Carolina State in descending order).

 

The Big XII has four teams with great stats (TCU, Oklahoma, Kansas State, and Baylor in descending order), four with poor stats (Oklahoma State, Texas Tech, Iowa State, and Kansas in descending order), and two with average stats (West Virginia and Texas in descending order).

 

The Pac-12 has no team with great stats, one team with poor stats (Washington State), and eleven with average stats (Oregon, USC, UCLA, Stanford, Arizona, Washington, Arizona State, Utah, Cal, Colorado, and Oregon State in descending order).

 

WITHOUT SOS

 

The SEC still tops the list by this rationale but the B1G is second with the ACC third, Pac-12 fourth, and Big XII last (if I just averaged all the rankings for their respective teams).

 

The B1G has three teams with great stats (Michigan State, Ohio State, and Wisconsin in descending order), four with poor stats (Indiana/Purdue (tie), Rutgers, and Illinois in descending order), and seven with average stats (Nebraska, Iowa, Minnesota, Penn State, Michigan, Northwestern, and Maryland in descending order).

 

The SEC has three teams with great stats (Georgia, Mississippi State, and Alabama in descending order), one with poor stats (Vanderbilt), and ten with average stats (Ole Miss, Auburn, Florida, LSU, Missouri, Arkansas, Tennessee, Kentucky, South Carolina, and Texas A&M in descending order).

 

The ACC has no team with great stats, two with poor stats (North Carolina and Wake Forest in descending order), and twelve with average stats (Clemson, Louisville, Georgia Tech, Boston College, Miami, Florida State, Duke, Pittsburgh, Virginia, Virginia Tech, North Carolina State, and Syracuse in descending order).

 

The Big XII has three teams with great stats (TCU, Baylor, and Kansas State in descending order), four with poor stats (Oklahoma State, Texas Tech, Iowa State, and Kansas in descending order), and three with average stats (Oklahoma, West Virginia, and Texas in descending order).

 

The Pac-12 has no team with great stats, two with poor stats (Colorado and Washington State in descending order), and ten with average stats (Oregon, Stanford, USC, Arizona, Washington, UCLA, Arizona State, Utah, Cal and Oregon State in descending order).

 

TL;DR? I rated the body of work of all 128 FBS teams and then focused on just the "P5" teams in order to discern a pecking order among the conferences. It would seem the ACC and Pac-12 have some real parity and the SEC is fairly loaded. The other conferences' teams vary greatly in terms of quality...with the Big XII striking me as the greatest wild card in terms of figuring out how good the teams in that conference are. To simplify it all once more, here are the percentages of teams, in each respective conference, that have above average cumulative rankings among just the "P5" using the data described above (w/ and w/o SOS, respectively):

 

B1G - 41.67% and 50.00%

SEC - 71.43% and 64.29%

ACC - 50.00% and 57.14%

Big XII - 60.00% and 50.00%

Pac-12 - 50.00% and 58.33%

 

By that rationale, you can see that the B1G is the only conference with less than half its teams posting above average numbers overall when strength of schedule is factored into the calculation.

 

Whoop-tee-doo - what's it all mean, Basil? I think we need to simplify the college football playoff system because even the objective data can be misleading. Take the conference champions and let the so-called experts, in concert with computers, pick another one to three team(s) from the rest of the FBS.

[Glendower]

I kind of have to agree with your Whoop-tee-doo. While I appreciate the time that you must have sunk into this analysis, with little meaningful interconference play, the stats all have to be interpreted in isolation on a conference level. All they really tell us are the performance of teams relative to their groups with no real quantitative means to compare on a group or between groups level because they were not randomly assigned to groups, so we cannot make any assumption about equality of the groups. In other words, for a team's within group performance to be extrapolated to a comparison with another team in another group, the groups need to be equal or at least assigned in such a way where a reasonable assumption of equality can be made. That's going to be the real hurdle of any attempt at an objective method of ranking between groups. There is so much "seems-like" thought at work and the profoundly flawed method of applying illogical common sense (e.g. "everyone knows that the SEC is the best...") necessarily at play to come up with national rankings.

 

Whew. As an aside, I was trying to compare statistics on week 5 (which was waaaayyy too early) by looking at the team of interest's offensive and defensive scoring data, then comparing that to the opponents' offensive and defensive scoring data. I was curious about NU v MSU, so I looked at the average number of points NU's opponents scored against their opponents. I created an average point total (not including NU) and compared that against what they scored against Nebraska (and did the same thing for their defenses). So instead of seeing if Nebraska held their opponents to some such number of points, I was looking to see if the defense held them to a score below their average. That would let us know if a team was simply, on average, low-scoring and if what looked like a defensive success could be attributed to what was a low-scoring offense. Anyway, the result of all of that was that NU tended to hold teams to below their average, MSU held teams to about average. Based on that, and that NU was lower scoring than MSU, I predicted a close game that would, optimistically, go NU's way. Well, I had the close game thing right (in the end!). Anyway, I thought that might be a fun system to try to work out in the off season, but there just aren't enough data points early in the season to do any analyses that are very meaningful. However, I did wonder if something like that could be used to make reasonable predictions across conferences.

 

My TL;DR-- this only further demonstrates that just trying to come up with even a moderately objective system should leave anyone with the realization that 1) Only within conference rankings are possible and 2) we need a larger playoff. I would LOVE it if official rankings were only released based on teams' standings in their respective conferences.

[Hunter94]

the four team playoff is a miserable failure......going forward there will be so much bitching and complaining that it will be scraped and something different will emerge..this year was only an example of the first of a flawed system based on the political nature of the "experts" seated on the committee...not saying they didn't deserve it, but no one was going to say no to Nick Saban!

[beorach]

I appreciate the discussion. The early stats are tough to deal with for certain. For what it's worth, I came up with my own spread for the Michigan State game and had the Spartans as a -1/2 point favorite but I was giving them 3.5 points for being in East Lansing. All I did to calculate that was consider a theoretical best team (TBT), with all the best rankings for each category, playing against a theoretical worst team, with all the worst rankings per my method, and assign a spread of -59.5 points - figuring the TBT was also the home team. When I do this for the NFL, I go with a lower score (-31.5 or so). Cfbstats.com has a nice site that allows you to take the games against FCS teams out but it'd be great if you could only do P5 stats, too. I'm not doing much but copying, pasting, and using pulldown menus aside from changing my coefficient to set that initial, theoretical spread. If I was having great success picking games, maybe I'd be up for working with refining data more. I suppose I should also work some more categories, like special teams and tackles for loss, into the mix as well. p.s. - I just looked and they do have a P5 only link.

 

The Holiday Bowl matchup is one that my calc's put at USC +2 after rounding to the nearest point. This spread, as was the case with the MSU one, doesn't have SOS factored into it. The site I like for SOS calc's is footballperspective.com and they had the Trojans' SOS as greater than ours. Between that and the turnover margin, I wouldn't bet the Corn but what do I know? I only won by taking them (and the points) against MSU because of DPE and that missed field goal.

 

Back to the playoffs, I suppose it was always going to be an evolving process because of the bottom line. Incremental progress is still progress at the end of the day.

[Moiraine]

the four team playoff is a miserable failure......going forward there will be so much bitching and complaining that it will be scraped and something different will emerge..this year was only an example of the first of a flawed system based on the political nature of the "experts" seated on the committee...not saying they didn't deserve it, but no one was going to say no to Nick Saban!

Compared to the BCS it was a huge success.

[skersfan]

I heard the other day on XM, the same 4 teams would have finished in the BCS, but it would be Florida State, Bama, Oregon, Ohio State. So it is not that much better. Great to see four get a chance though.

 

I am not ever going to agree with conference champions being a lock.

 

8 best, play the first game at the higher ranked teams home.

 

The prices for these games are going to hurt the playoff.

 

Packer was saying it will turn it into another super bowl deal. Most boring game to attend in the world.

 

1681 bucks for the average ticket at the National Championship.

 

Not going to be a lot of real fans showing up there, mostly corporate ticket sales. That will be a sad day for college football.

[Moiraine]

So it is not that much better. Great to see four get a chance though.

 

It's a lot better for that reason.

[Glendower]

I heard the other day on XM, the same 4 teams would have finished in the BCS, but it would be Florida State, Bama, Oregon, Ohio State. So it is not that much better. Great to see four get a chance though.

 

I am not ever going to agree with conference champions being a lock.

 

8 best, play the first game at the higher ranked teams home.

 

The prices for these games are going to hurt the playoff.

 

Packer was saying it will turn it into another super bowl deal. Most boring game to attend in the world.

 

1681 bucks for the average ticket at the National Championship.

 

Not going to be a lot of real fans showing up there, mostly corporate ticket sales. That will be a sad day for college football.

What would you base the "8 best" on? More people voting? Another committee arguing about it instead of letting the teams that proved they were the best in their conference in? How do you argue that a non-champion of a conference was better than the champion who had a better record in the conference? Why even have anything in place that allows for that kind of foolishness? I don't want to ever have to hear another explanation for why a team that LOST in a head-to-head match up is better than the team they lost to.

[beorach]

That's exactly it, Glendower. I hear this on talk radio all the time, too (that the best teams should play). I understand people fear the inevitability of a team with several losses winning a CCG and getting a chance to play for the NC but improvement and timing are all part of what makes sports fun to watch. If a conference can't figure out a fair way to determine its champion, that's on the conference instead of some writers and/or a magic formula. I think most people want the regular season to remain special and this is the easiest way along with having higher seeds get home games in the playoffs (so I agree with Skersfan on that part). I figure the two best arguments against the old system were: 1) a team that can't win its conference shouldn't be playing for the NC; and 2) the smaller (Division 1/FBS) programs shouldn't be shut out automatically. I know Marshall lost to the Hilltoppers but would guess they'd have been on the outside looking in regardless. The rationale behind that would have been that other teams had proven more despite having lost a game? I'd rather an undefeated team of kids was given a chance.

[Count 'Bility]

If we want the best teams to play, then Florida St shouldnt be one of the four teams this year. That's my whole deal with it. Of the top 6 teams, Florida st is sixth. Maybe even behind some of the teams lower than that, like Miss St. it's obvious to me. Just my opinion. But they went unbeaten and won their conference. And I have always clamored for high importance on conference championships. So in the end I'm cool with it. I just wish they would just do what they say. "We're not guaranteeing conference champs and we're picking the 4 best regardless. using the eye test. etc. etc. etc.". Well if all that's the case, Florida St really shouldntve been in the conversation.

 

Make it 8 teams, guarantee the power 5 conference champs, cuz that is played out on the field. You take a lot of guessing and interpretation out of it. Hand out 3 wildcards. Gives a chance for the every-now-and-then elite non P5 Boise/Cinci/PreBig12 TCU elite types a shot. Just my opinion folks. And I really believe this is what we're headed too. Obviously i would think the P5 conferences would want this. Can you imagine what kind of ratings the CCG's would get knowing theyre having a straight up play-in game for the playoffs?

[louisianared]

I really think we are headed to and need an 8 team playoff.

 

I'd have automatic bids to the power 5 conferences

highest rated mid major champion

2 at large

 

Highest seed hosts 1st round

Bowl rotation for semi-final and final

 

I'd also institute a cap price for the tickets allocated to the schools to make sure fans actually get into the game.