No surprises here.
Top 25 (Explanation; Full Ranking):
The questions "who had the best overall season?" and "who are the conference champions?" grant, unsurprisingly, somewhat different answers. Below, in rank order, are teams listed according to network analysis of their overall body of work in the 2015 season.
The conference championship week provides an illustration of how second and third degrees of separation can affect overall ranking. For example, Oklahoma State, who had a bye week like most of the Big 12, experiences a decline due to Texas defeat of Baylor, who was one of the Cowboys' two losses. The Texas-Baylor game more firmly enmeshes Baylor in the network of losses, increasing the negative impact of OK St.'s loss to Baylor on their ranking.
Top 25 (Explanation; Full Ranking):
Notre Dame falls from number one, but they're still popular with the network. While the Stanford loss increases their loss value more than twenty fold, they still have the best value wins of any team in the FBS.
Understanding, of course, that their entrance into the playoff is politically unlikely with the CFPC, Ohio State in the 5 spot provides some interesting possibilities. If Florida beats Alabama, could Ohio State get into the playoff? Should be interesting.
With Navy's loss, Bama is now in.
Top 25 (Explanation; Full Ranking)
Still some football to played - though the sudden appearance of OU (WHO LOST TO TEXAS!!!) in the CFPC's ranking is a touch peculiar. With the game against Stanford on the horizon, a good opportunity is still available for Notre Dame to climb into the playoff, combined with the possibility of an OU loss in Bedlam.
With Ohio State's fall from grace, Navy sneaks into the top 4, highlighting that the Midshipmen have really been ignored. With their one loss to Notre Dame, however, they're under-ranking is very closely tied to the under-ranking of the Irish. Politically, however, it would be difficult for the committee to agree to have Clemson, Notre Dame, and Navy, given the Tiger beat the Irish who beat the Midshipmen, all in the playoff. Unfortunately, political necessity and evidence based merit are at times incompatible.
However, Bama has snuck up on the top 4. With a win in the Iron Bowl and the SEC Championship, the Network Ranking's evidence based perspective and the CFPC's expectation/political perspective may align for the Tide.
Top 25 (Explanation; Full Ranking)
Committee desperately wants Bama in the top 4, but with that loss to Ole Miss looking progressively worse, its a hard empirical case to make. Bama climbs three spots with their win over Mississippi State, but are in a distant #10.
Mike Leech's fighting cougars have cracked the top 25.
Top 25 (Explanation; Full Ranking):
So falls Michigan State. In the wake of Michigan State and LSU's losses, Notre Dame rises to the top. Even when discounting their win quality by the Clemson loss, their total body of work surpasses every other team in the country. The Big 12, however, despite two undefeated teams (#6 Oklahoma State & #11 Baylor), can't break into the top 4 due to their overly squishy schedules. Put simply, the combined quality of Oklahoma State's wins are not even as good as that of Memphis, even after subtracting the effect of the Navy loss.
Now, Oklahoma State/Baylor have more room to grow than Memphis or Houston (as do the leading teams in the SEC and PAC 12), but the dissatisfying position of the Big 12 in the overall CFB Network is persistent. As for the reason, we can compare their out of conference opponents which strongly influence the positioning of the conference relative to other clusters in the network.
Baylor - #102 SMU and #103 Rice
Oklahoma State - #71 Central Michigan and #122 UTSA
vs.
Memphis - #24 Ole Miss, #36 Bowling Green, #126 Kansas
Houston - #61 Louisville, #111 Texas State, #70 Vanderbilt
Just as a matter of logistics, it would be pretty difficult to play worse non-conference opponents than Baylor. Art Briles really went out of his way to schedule the worst teams he could short of scheduling nothing but FCS opponents.
As for the SEC, the Ole Miss loss to Arkansas punished Alabama, even if they beat LSU. Why? Because Bama lost to Ole Miss and Ole Miss lost to Arkansas, Alabama is now subsumed in the loss networks of such teams as Toledo and Texas Tech. Not good SEC, not good at all. But, still much better than the train wreck that is the PAC 12.
Remember, the network ranking is not arguing Memphis is "better" than Oklahoma State, just that they have a more impressive total body of work at this point in the season. Also keep in mind much of this will necessarily change. While Clemson should be alright (unless they go Clemson on us), Notre Dame still has to play Stanford, Iowa and Ohio State will eventually come to blows (assuming Ohio State makes it past Michigan and Sparty), Memphis and Houston will square off, and Oklahoma State will face Baylor, who still has to play TCU, and both of them have yet to play Oklahoma.
Top 25 (Explanation; Full Ranking)
With Notre Dame continuing to chalk up impressive performances, would the College Football Playoff Committee allow a Clemson v. Notre Dame rematch? Could a one loss Notre Dame leapfrog an undefeated TCU or Baylor?
Given the politics involved, the answer to both is probably no. However, given how close the loss to Clemson was (22-24), a rematch has a lot of potential to be a great game. As for the latter, given their soft Big 12 schedules, Notre Dame has a much more impressive body of work thus far this season than undefeated TCU (#12) or Baylor (#18 and the worst ranked undefeated). Indeed, in the Network Ranking, they have the most impressive body of wins of any team, and are only second due to the negative impact of losing to Clemson. If Clemson remains undefeated and Notre Dame wins out, I would expect that to continue to be the case regardless of the final result out of the Big 12. Right now, the most probable outcome out of the College Football Playoff Committee is Michigan State, Iowa, or Ohio State out of the Big 10, TCU or Baylor out of the Big 12, and if they win out, LSU and Clemson. Unfortunately, there will be good arguments to be made by Notre Dame and a possibly undefeated Memphis or Houston team, highlighting the problems of selecting four teams for a playoff on the basis of anything other than an evidence based system. Right now, Notre Dame fans should consider themselves in the running, but they'll need some help elsewhere to survive the politics of the playoff committee.
This coming weekend, we'll get what amounts to the de facto SEC championship when LSU faces Alabama in Tuscaloosa. At minimum, the absence of an undefeated champion from two or more power five conferences is what Notre Dame needs. If Bama can take down the tigers, the Irish are much closer to being in the conversation.
Top 25 (Explanation; Full Ranking):
Not a lot of surprises here given the big performances this week. And, looking forward to next week, no big games within the top 4. One big difference between this ranking, as an evidence based system, vs. the AP Poll, which is a speculation based system, is Ohio State and Baylor. For fun, let's compare Baylor's wins to LSU's.
Quality of Baylor wins as determined by the losing team's Network Rank:
84. SMU
71. Rice
34. Texas Tech
85. Kansas
51. West Virginia
Quality of LSU wins:
28. Mississippi State
33. Auburn
79. Syracuse
125. Eastern Michigan
59. South Carolina
8. Florida
The variance in team quality is higher for LSU, with both the best and least impressive wins, but the average opponent ranking for Baylor is 65, while for LSU it's 55, with wins over Florida, Mississippi State, and Auburn all more impressive than any win by Baylor. Baylor has a back-loaded schedule, with their, at present, four most difficult games in the four final weeks of the season. If they win they will very likely climb into the top four, but they haven't earned the right to be in the conversation yet.
Top 25 (Explanation; Full Ranking):
Big shake ups this week driven by a few big games. Florida gets leaped by Utah, Michigan, and the Aggies. Utah jumps (and jumps dramatically) for two reasons: (1) They beat Cal, who beat Texas, who just upset Oklahoma, as well as Washington, who upset USC. The consequences of those two upsets reverberate pretty strongly throughout the network. (2) They beat Michigan, who just beat Northwestern. The value of both Utah and Michigan's wins are really quite impressive, with Michigan only receiving a slight negative for losing to undefeated Utah. The Aggies and the Gators are pretty close, and though the Aggies were on a bye, the value of their wins increased this week by slightly more than those of the Gators.
While Utah should take care of business at home against Arizona State, Michigan gets another test with #23 Sparty in the Big House. In the SEC, we'll find out if the Aggies and Gators are for real against #14 Bama and #13 LSU.
Finally, where's Baylor? While the Bears are undefeated, their best (and only half-decent) win is against Texas Tech. While their schedule improves with West Virginia next week, it doesn't get interesting until mid-October, where in a three week period they play OU, Oklahoma State, and TCU. If they make it out of that, they'll have a case to make.
Top 25 (Explanation; Full Ranking):
Florida hangs on to the top spot after crushing Ole Miss, but faces yet another test against #25 Missouri in Columbia. USC shoots into the top 25 despite being on a bye week thanks to Arizona State's (who USC defeated last week) victory over UCLA. Northwestern is looking strong with a defeat of Stanford (who beat USC, who beat UCLA, etc.). But, Saturday they face #6 Michigan in The Big House. Utah, likewise, has a test in their home game against a decent Cal team, while Ohio State should take care of business at home against Maryland, who Michigan shut out 28-0 this weekend. Other than Ohio State's conference cupcake, great weekend of football ahead.
Top 25 (Explanation; Full Rankings):
Inaugural 2015 rankings have arrived (a bit late in the week)! The Gators, with pretty good wins against a high performing Kentucky and a ECU team that took down Virginia Tech, take the first spot. For a comparison of the difference between a post-dictive, evidence based model and the popular survey approach, I've placed each team's ranking in the AP poll in parentheses below. The difference between the two provides some indicator of the degree to which teams at this point in the season are under or over rated.
Top 25 (Explanation; Full Ranking)
The Network:
The State of World Liberty Index was a combination of indices compiled originally in 2006 by Nick Wilson. Since then, I've periodically updated the index. My methodology is similar to the original with the following exceptions: I use only the civil liberties portion of the Freedom House, economic liberty (Fraser/Heritage) is weighted evenly with social liberty, and the Reporters without Borders Press Freedom Scores are of equal weight to each of the Freedom House Civil Liberties sub-components. All changes are done with the purpose of creating the most accurate relative ranking of countries by the degree of liberty (economic and social) enjoyed by the average individual within the state or territory. Note, the ranking is not reflective of, nor does it include, political or electoral rights. Where data is available, territories or colonies (e.g. Hong Kong) are included.
2015 Liberty by Quintile. Colors represent countries grouped by quintiles in order of blue (top 20%), green (20-40%), yellow (40-60%), orange (60-80%), and red (80-100%).
2015 Changes in Liberty. Blue is an increase in rank by >11 and green >5. Orange is a fall in rank of >5 and red >11.
2015 Most Free Countries:*
1. Bahamas
1. Hong Kong
3. Dominica
3. St. Lucia
5. Barbados
5. New Zealand
5. St. Vincent and the Grenadines
8. Australia
8. Canada
8. Switzerland
*note data for Liechtenstein on economic freedom, tied in 2014 with Bahamas for overall most free, was not available in the most recent time period.
2015 Most Tyrannical States
1. North Korea
2. Venezuela
2. Central African Republic
4. Zimbabwe
5. Iran
5. Cuba
7. Eritrea
8. Ethiopia
8. Equatorial Guinea
8. Republic of Congo
8. Chad
2015 Greatest Increases in Liberty:
1. Madagascar
2. United Arab Emirates
3. Democratic Republic of Congo
3. Myanmar
5. Tunisia
6. Fiji
6. Qatar
6. Mozambique
9. Uzbekistan
10. Costa Rica
10. Sao Tome y Principe
10. Georgia
2015 Greatest Increases in Tyranny:
1. Thailand
2. Argentina
3. Turkey
4. Malawi
5. Ecuador
6. Egypt
6. Hungary
8. Lebanon
9. Iran
9. Slovenia
Colored maps modified from images at http://www.outline-world-map.com/.
As with previous years, attempting to answer "who had the best season?" based upon their full body of evidence doesn't always line up with who the eventual college football champion may be. In a retrodictive ranking system, you seek to evaluate the total quality of evidence, not predict who wins games (or preoccupy ourselves with who won the championship game above all else). Treating playoff games equally with the regular season games, Ohio State's victories over Alabama (who drops to 6) and Oregon (who drops to 4) catapult them up the rankings by subsuming the win networks of both these impressive teams. In a ranking that evaluates all the evidence, however, we can't simply overlook the fact they lost to Virginia Tech (ranked 43). In the end, it's just not enough to get them into the number 1 spot, with Florida State despite a somewhat embarrassing loss to Oregon, holding on to number one (though, it's certainly close!).
Top 25 (Explanation; Full Rankings):
Last week I compiled the rankings I could find done by professors with appropriate methodological skills. More about those rankings here. I've repeated that process for the end of the season, adding to the list the Simple Rating System by Dan Drinen. Taking the median ranking (and then the mean to break ties) the results are that the professors don't like Florida State. Poor TCU remains in the top 4, with Ohio State beating out the Seminoles for the fourth spot. If you look at the rating methods, the ones including win margins are those that really punish Florida State, whereas the more simple win-loss methods, such as my own, have them much higher. The 8 professors are:
Jay Coleman, University of North Florida - PhD Industrial Management, Clemson University (http://www.unf.edu/~jcoleman/minv.htm)
Wesley N. Colley, University of Alabama in Huntsville, former BCS contributor - PhD Astrophysical Sciences, Princeton University (http://www.colleyrankings.com/)
Doug Drinen, University of the South - PhD Mathematics, Arizona State (http://www.sports-reference.com/cfb/years/2014-ratings.html)
Cody Kirkpatrick, Indiana University - PhD Atmospheric Sciences, University of Alabama in Huntsville (http://talismanred.com/ratings/cf/)
Kenneth Massey, Carson-Newman College, former BCS contributor - PhD Mathematics, Virginia Tech (http://www.masseyratings.com/rate.php?lg=cf&sub=FBS)
Peter J. Mucha, University of North Carolina - PhD Applied and Computational Mathematics, Princeton University (http://rwrankings.blogspot.com/)
Patrick Rhamey, Virginia Military Institute - PhD Political Science, University of Arizona
Matthew Rissler, Loras College - PhD Mathematics, Notre Dame (http://discnerd.wordpress.com/)
As I talked about last week - human beings have a propensity to overly bias the most recent observation and ignore broader evidence. This week, the playoff committee engaged in the most extreme and obvious example of this illogical propensity. From the perspective of the Network Ranking, Ohio State, ranked 11, has no business being in the playoff.
I'll create a Professors' ranking similar to what I did yesterday as soon as the other 6 rankings are available. Perhaps some of the rankings that include win margins will push Ohio State into the top 4 in the composite. But, in the meantime, here's the Network Ranking:
Top 25 (Explanation; Full Ranking):
In the midst of watching today's college football, and incurring ever rising angst to the pontificating of just about every sports caster and pundit, I had a thought - forget the playoff committee and the punditry, what if you created a ranking based purely on the different rigorous mathematical approaches created by individuals with the appropriate expertise. So I combed through the College Football rankings that exist with two critiera: 1) the ranking is mathematically rigorous, unbiased, and non-arbitary in its construction and 2) is created by a professor with an advanced degree in a field where that professor is employing similar mathematical methods in his own substantive research. The first is a necessity in retroactive based rankings (meaning they evaluate existing evidence rather than attempt to predict), the second grants the ranking credibility as the author has related skills that have resulted in academic publication.
This quest resulted in the following 7 rankers:
Jay Coleman, University of North Florida - PhD Industrial Management, Clemson University (http://www.unf.edu/~jcoleman/minv.htm)
Wesley N. Colley, University of Alabama in Huntsville, former BCS contributor - PhD Astrophysical Sciences, Princeton University (http://www.colleyrankings.com/)
Cody Kirkpatrick, Indiana University - PhD Atmospheric Sciences, University of Alabama in Huntsville (http://talismanred.com/ratings/cf/)
Kenneth Massey, Carson-Newman College, former BCS contributor - PhD Mathematics, Virginia Tech (http://www.masseyratings.com/rate.php?lg=cf&sub=FBS)
Peter J. Mucha, University of North Carolina - PhD Applied and Computational Mathematics, Princeton University (http://rwrankings.blogspot.com/)
Patrick Rhamey, Virginia Military Institute - PhD Political Science, University of Arizona
Matthew Rissler, Loras College - PhD Mathematics, Notre Dame (http://discnerd.wordpress.com/)
To calculate the rankings, I took the median for each team (rather than the weird drop the high low and take the average method of the BCS). The mean was used to break ties. Incidentally, the Professor Ranking and the Playoff Committee give the exact same top 4 in the exact same order. However, note there is no SEC bias - if anything after you get past the top 4, the polls and the playoff committee are actually unfairly biased against the sec.
The dramatic outcomes of this week result in some pretty significant movement in the rankings this late in the season, with many teams experiencing double digit moves - but one observation does not inherently negate an entire body of work. One of the problems of human evaluation, true of everything from ranking college football teams to generally evaluating risks or making financial decisions, is a propensity to dramatically outweigh small pieces of evidence and outliers while ignoring dominant trends - sort of like denying the existence of global warming just because it's snowing at your house. Mississippi State did lose to Ole Miss, which caused them to fall a spot in the ranking, but it doesn't (and shouldn't) matter much. For starters, Ole Miss was underrated following an overreaction to the LSU and Auburn losses, both of which are top 10 teams. State losing to Ole Miss is just another redundancy in the SEC West - evidence of a lack of dominance among a group of excellent teams rather than evidence of the group's overall lack of quality. The SEC East, however, does take a hit, with Georgia, UF, South Carolina, and Kentucky all experiencing significant drops following losses to ACC opponents, which all likewise make gains.
If the season is about the body of evidence, rather than the media hyperbole following a single game, than TCU is currently on the bubble. Mississippi State's wins are a little bit better, their losses are a little bit worse, but on balance they come out slightly ahead of TCU. However, given the absurd rankings we've seen so far from them, the committee is unlikely to be so sensible and cautious in their evaluation of last week's games. While TCU in place of State is not a huge error (they are only narrowly beaten out by State in the Network), if Ohio State makes it into the playoff it will be the most egregious violation of a college football ranking since 2012 when the AP put Lane Kiffin's USC as the preseason number 1.
Top 25 (Explanation; Full Ranking):
Minimal changes this week, and the top 4 remain the same. Ole Miss is punished a little for losing to Arkansas, but not enough to matter. Why? Because the teams that Arkansas has lost to predominantly overlap with those that LSU and Auburn lost to, who are already in Ole Miss's loss network, so we're not getting much new information about the severity of Ole Miss's losses (and their wins continue to look good). Indeed, this is generally true of most teams this week: win or lose, the ties are redundant given the extent of conference play information available to us (that or they played irrelevant cupcakes, ala Alabama). Arkansas, however, gains hugely, rising to number 15. Next week, however, with many teams playing out of conference rivals (Florida State and Georgia in particular), we should expect more movement. It also means as we approach the end, if Bama wins the Iron Bowl and State wins the Egg Bowl, there really isn't much of an argument to keep either one out of the playoff. Obviously, Marshall will be the deserving team on the sideline, likely to be replaced by the playoff committee by the PAC 12 champion.
Top 25 (Explanation; Full Ranking)
What had been minimally harmful losses for the SEC West to this point has opened the door to a more serious problem, causing the entire group to fall. LSU's loss to Arkansas, as well as Auburn's loss to SEC East Georgia and the Texas A&M loss to Missouri, significantly increases the centrality of those top remaining teams (Alabama, State, Ole Miss) to the network of losses, dragging down their relative rank. Indeed, the result of these three losses is some of the most dramatic movement in the rankings for the past 5 weeks, leaving Florida State at the top. While the quality of their wins is by no means higher than Alabama, Ole Miss, or even UCLA, the Seminoles have no losses, and with the increase in the costliness of the SEC West's losses this weekend, it's enough to propel them into first.
But, Florida State isn't the only undefeated team remaining. First, the Network Ranking is not overly biased toward win loss record apart from the value of wins and losses (unlike some rankings) - as evidence look at the past few weeks. But for all the same reasons Florida State climbs to number 1, Marshall has climbed to number 2. If you look at the win centrality, Marshall's wins are less valuable than any other team in the Top 25, which explains why they are so often overlooked. However, they have no losses to detract from that win record, and with this past weekend's chaos making all losses more costly for most of the top teams, it's enough to propel Marshall into the number 2 spot.
This highlights, in part, another serious flaw to the college football playoff. If at the end of the season there are only two undefeated teams remaining, shouldn't the undefeated CUSA champion get a spot? Obviously, they won't in reality, but this oversight ignores an excellent season by the Thundering Herd.
For what it's worth, Marshall has not only won every game this season, but done so by 15 points or more. Their average win margin is 30.8.
Top 25 (Explanation; Full Ranking)
While SEC cannibalism results in SEC West teams failing to gain improved evidence of their "goodness" this past week (sometimes win or lose given its cyclical nature, e.g. Ole Miss->Bama->LSU->Ole Miss), it doesn't otherwise penalize them, as evidenced by the only mild relative decline in the ranking of Auburn and Ole Miss. However, Florida State, while they continue to win, is experiencing a decline in the relative value of those wins as those opponents falter elsewhere (Notre Dame, Oklahoma State, Syracuse).
While Oregon rose significantly, outside the SEC West and Florida State, Marshall, Notre Dame, and UCLA are all more deserving of playoff inclusion.
Top 25 (Explanation; Full Ranking):