Tuesday, December 06, 2011

The Best Team Wins the Championship? What Kind of Communist Notion is That?

It’s too bad that this article resides behind ESPN’s paywall and is full of untrustworthy notions like “computer rankings,” because Brian Fremeau makes a very good, basic point about college football’s structure: it is more likely to crown the most deserving team.  Fremeau notes that according to the Massey Rating composite, the BCS has consistently crowned the team that the computer consensus has placed as the best team in the country (or at least the team with the best resume, if those two concepts are distinct).  In fact, the Massey composite goes back to 1996 and reflects that the BCS (and the Bowl Alliance before it) is a perfect 15 for 15 in awarding the national title to the most deserving team.  The BCS is not as good at matching the best team against the second-best, but as Fremeau notes, playoff systems don’t really achieve that aim, either:

According to the Football Outsiders' DVOA ratings, the Green Bay Packers were only the third-best team in the NFL last year and didn't face the best team, the New England Patriots, in the playoffs. (Note: They did lose head-to-head to the Patriots in the regular season). In college basketball, the Connecticut Huskies played the best in the month of March and claimed the championship, but they finished only 10th in Ken Pomeroy's opponent-adjusted ratings. The Huskies didn't face any of the tournament's No. 1 seeds and faced only two teams in March Madness ranked in Pomeroy's end-of-year top 10.

Fremeau then notes the drawback of a playoff system: as the playoff increases in size, the odds that the best team will win goes down:

What is accommodated in a playoff system is the opportunity for weaker teams to have a shot at winning the championship. Opening up the field certainly has its rewards, potentially assuring every possible championship-caliber team a chance. But it changes the nature of what the championship means, as well. The Packers and Huskies were terrific teams down the stretch in those seasons, but advanced metrics and many fans would agree that the tournament title didn't suddenly mean they were the best team over the course of the whole year. A smaller playoff field, even as small as two teams, allows the best overall season to be rewarded with a championship.

Our college football FEI ratings project the Tigers with a 68 percent chance of beating the Crimson Tide in the BCS title game rematch. Certainly not a guarantee, but it presents a better than 50 percent likelihood that the clear No. 1 team in the nation will claim the crown. If we had a four-team playoff and LSU was required to beat both Stanford and Alabama on a neutral field, that likelihood drops below 50 percent. With an eight-team playoff built from the final BCS standings, LSU's likelihood of defeating Kansas State, Stanford and Alabama drops to 41 percent.

This is a great way of framing the playoff discussion and it helps me tighten my own beliefs.  There is a happy medium between the playoff structures of American professional sports,* which strike me as too big and not slanted enough in favor of the most deserving teams, and college football’s current structure, which is too small to accommodate teams like 2004 Auburn.  Brian Cook does a nice job of finding that medium with his annual description of a six-team playoff system that incorporates byes and homefield advantage to reward the best teams and is also small enough that any team that wins the tournament will end up with the best resume.  That said, Fremeau’s point about the advantage of the current two-team playoff cuts against Cook’s (and my) complaints about the BCS.  We all love to complain about the impossible task of picking the right #2, but the advantage of the BCS is that it is more likely to pick the right #1.

* – The NCAA Men’s Basketball Tournament is just a complete abomination to me.  If you need additional evidence that the Big Dance has killed college basketball, look at Saturday’s sports coverage.  Were you aware that North Carolina and Kentucky played?  Two top five teams, loaded with NBA prospects wearing the jerseys of the two winningest programs in college basketball history, played a one-point game that was decided in the final seconds and the game was a total afterthought on the day of the college football conference championship games.  That’s what happens when a sport becomes a four-month lead-in for a three-week tournament.


Anonymous said...

When the Fab Five had its coming out party against Duke, how many teams were in the tournament?

When UNLV ran a train on Arkansas on national TV and got a major feature in SI, how many teams were in the tournament?

I'm sure there's a reasonable anti argument somewhere (though I haven't encountered it yet), but this persistent, idiotic assertion that the tournament has hurt college basketball isn't it. You don't back your claim up with *ANY FACTS* (is attendance down since the tournament expanded? No. Is regular season revenue down since the tournament expanded? No.), you fail to account for the fact that college basketball is an inferior product to the product you remember from the 70s and 80s (the talent level is lower and the top players don't stay around long enough to generate interest), and you selectively choose games (you didn't notice a noon game played on the final day of cfb's regular season? Wow. Why not talk about the MSU-UNC game, which scored strong ratings and won its night? Why not talk about the Duke-UNC regular season games that get significant buzz and decent ratings?).

Michael said...

Evidence that college basketball's popularity is declining? Take a look at conference expansion. Nobody wanted Kansas, which has one of the top five college basketball programs. When the ACC expanded, they added football programs and went to divisions, both of which were an anathema to the league's basketball coaches. Tickets for the ACC Tournament used to be impossible to come by; now, they're easy to get.

College basketball is now an afterthought. Why? Talent may be one reason, but the fact that the regular season is utterly meaningless has to be another. The fact that you have to cite regular season games from 1990 and 1991 proves my point. At that stage, the Tournament had been at 64 teams for 5-6 years. Given a longer timeframe, the Tournament has killed college basketball. Eventually, fans figure out that they have heard the same irrelevant debates about bubble teams that never do anything in the Tournament, or about whether Duke is going to play in Greensboro or Birmingham.

The fact that you also have to cite a game PLAYED ON A FUCKING AIRCRAFT CARRIER proves that you are on weak ground. If Alabama and Oklahoma played a game in September, do you think they would have to play in some ridiculous setting in order to get attention? No, because the game would be meaningful in and of itself.

Anonymous said...

Using indirect measures that are affected by factors other than the popularity of college basketball, like "preferences in conference realignment decisions" instead of just using direct evidence tells me that you know you're wrong. You know that the direct evidence (attendance, revenue) discredits your claim so you have to go somewhere else.

There have been many relevant, salient regular season games in the expanded tournament era ((e.g. Kentucky-UMass had huge ratings and a 3 page feature article in SI in 95). They have declined as the talent level has dropped during the "early entry" or high school-drafted era.

How many buzzworthy regular season games would cbb have had in the late 90s if Kobe Bryant, Tracy McGrady and Kevin Garnett went to school and played 3 or 4 years, and if AI and Marbury stayed 1 or 2 more years? How many buzzworthy games would there have been in the mid-00s if Tyson Chandler, Amar'e Stoudemire, Eddie Curry, and Kwame Brown enrolled in college and played for 3 or 4 years, or if Jason Richardson, Zach Randolph, Gerald Wallace, etc. played for 2 or 3 more years? It gets really crazy if you look at the late 00s, with LeBron, Carmelo, Perkins, Bosh, Howard, etc.

If John Wall were playing against UNC the other day, you would have known about it.

K. Medvedovsky said...

Anon - I'm curious: Is revenue in fact up/stable? Is this information publicly available somewhere? I'd love to see a chart, or failing that, make a chart, tracking regular season revenue. Michael's point that it took a while for people to see how meaningless it all was would then be sort of testable as well.

The MSU-UNC game, as mentioned, was played on an Aircraft carrier and kicked off the regular season in effect. But I'd also be interested in seeing historic Duke-UNC ratings; my sense is that interest has subsided, so I'd like to test that theory.

Anonymous said...

Looking back it was UMass-Wake Forest that got the buzz in 95, which reinforces my point (Jr. Camby vs. Jr. Duncan).

Anonymous said...

According to the Football Outsiders' DVOA ratings, the Green Bay Packers were only the third-best team in the NFL last year and didn't face the best team, the New England Patriots, in the playoffs.

Those numbers include a game where Rogers was injured early and the team stagnated (DET) and one game where Rogers didn't play at all (NE, where GB lost by 4 using Matt Flynn as QB). Seems a little silly to say they weren't the best (at least with their starting lineup healthy).

Michael said...

I'm not ignoring the TV ratings and attendance. I honestly don't know the answer to that. I suspect that they are both down (especially TV ratings), but I am not sure where I would go to find that information on a historical basis. If you can find that information, then I would be all ears. I can point you to a poll that shows that college basketball has declined in terms of the response to the "what is your favorite sport?" poll question:


Look, I agree with you that early entry is one factor, although it's not like the NBA doing away with players going straight from HS to the NBA has led to a rush of popularity for CBB. Additionally, it's not like players didn't go pro early before. Magic Johnson, Shaq, and Chris Webber all went pro after their sophomore seasons. That said, the frequency has increased.

However, you've offered no reason as to why that explanation is mutually exclusive with the expanded Tournament also being a negative factor. Sports requires a narrative in order to drive interest. I don't see how college basketball popularity couldn't decline when the vast majority of the season is played for seeding. What exactly did the UNC-Kentucky game mean on Saturday, other than a slightly better chance at a #1 seed and within that, a chance that the winner will have to travel a slightly shorter distance to its regionals?

Uncle Mike said...

The reason nobody knew that North Carolina and Kentucky, arguably the 2 most respected college basketball programs, were playing each other is that it happened during college football's rivalry week. Seriously, if the Falcons were to play a relatively nearby opponent -- say, Carolina, Jacksonville or Tennessee -- on the same day as Bulldogs-Jackets, which do you think would sell out?

Anonymous said...

Shaq went pro after playing at LSU for 3 seasons, and he actually *came back to school* after winning the AP poy in 91. Larry Johnson is similar; he could hav ebeen drafted near the top of the first round after his sophmore juco season, but instead he went to UNLV for *two years!* He returned for his senior season after being a consensus AA and being a likely top-5 pick.
That's the lifeblood of marketing in a sport where interest tracks stars, not teams (for the most part; Kentucky and UNC are a little different).

The expanded NFL playoffs don't seem to have deleteriously affected regular season revenue/attendance or popularity, as your poll shows.

4.0 Point Stance said...

Of course, this suggests the best way to choose a champion would be to play out the bowls in the old style, then vote. But this is exactly what the BCS was designed to replace. The most logical conclusion is that the American people *want* some level of arbitrariness in who they call a "champion." The idea that the best team probably won't win a multi-level playoff is considered a bug, not a feature.

I think it's rooted in the American love of a plucky underdog. We actually like that George Mason lost a lot of games in the regular season before inexplicably turning it on and streaking to the Final Four. If they had gone 35-1 and then gone to the Final Four, no one would have been interested. Maybe this makes no sense, but there it is.

Michael said...

Anon, good catch on Shaq. As for the NFL, the answer to your question is parity. :)

4.0, I agree that the American public wants an arbitrary declaration of a champion, but I also think that the American public doesn't quite understand what they would lose with a big playoff.

Anonymous said...

The conference tournaments need to take a large share of the blame for decreasing the relevancy of regular season cbb games. There are a lot of one-bid mid-majors conferences out there where even a perfect conference record is meaningless if you don't win the tournament as well. With the notable exception of the Ivies, most smaller leagues have basically sold the integrity of their whole season out for the cash and exposure of one conference title game on ESPN/ESPN2.

Also, aside from the size of the bracket, the NCAA's inability to make a proper S-curve also hurts the value of regular season wins. Last year, Duke's reward for grabbing the last 1 seed was to be shipped out to Anaheim, where the regional's 4 seed (Arizona) and 2 seed (San Diego State) both had de facto home court advantage over it. They probably would've had an easier time making the Final Four if they were the 2 seed in the Southeast regional instead, even discounting the rash of upsets that ended up occurring in that region.

Meanwhile, as the overall #1 seed Ohio State did get the closest regional in New Jersey, but they also had to face 4 (Kentucky) and 8 (George Mason) seeds that were both vastly underseeded going by advanced statistical measures, plus arguably the toughest 2 seed (UNC) considering how much they improved after Larry Drew's benching. Both Pittsburgh and Kansas got much easier draws despite having the second and third best resumes according to the NCAA. This wasn't a one-off fluke, either, as in 2010 top overall seed Kansas got a brutal draw with quality teams at the 2, 3, 4, 5, 6, and 9 seeds. Considering how little help that top overall seed designation has provided historically, it's not clear at all that beating UNC on Saturday actually helped Kentucky's chances in March at all.

Michael said...

Anon, I totally agree with you on the crappy work done by the seeding committee. I made a similar point in 2010:


Because the NCAA has trouble filling up regional sites, they try to place teams close to home, but in so doing, they often screw more deserving teams.

chg said...

Anon argues credibly that conference tournaments have made many regular season matchups less meaningful. How can one conclude that a CFB playoff would not have the same effect?

My completely anecdotal experience is that the more devoted one's college football fandom, the more likely the person is to oppose a playoff system.