Sunday, March 14, 2010
Conference tournaments not a bad way to decide best team
In many ways, conference tournaments get a bad rap. When somebody like Houston wins the Conference USA tournament, we are forced to try and make sense of how a team that had a losing conference record could be declared its champion. However, in the vast majority of situations a tournament is not a bad way to decide a conference champ.
More specifically, it’s not much worse than the regular season in this regard. It might seem like a 16-18 game schedule should reliably determine which team is the best, but it doesn’t always work out that way. Take for instance the Atlantic Sun. Four teams tied for the top spot with another a game back. The A-Sun could have played an 82-game schedule and we still could not have felt comfortable about which team was truly best. I don’t think anyone would argue that a tournament was a bad thing for the A-Sun. You have to break that tie somehow.
In order to determine how well conference tourneys do in identifying the best team, I used my ratings to see where the conference tournament champ ranked in its conference. I am taking the ranking after the conference tournament because that is one true benefit of playing a tournament – you get extra data points. I then compared these numbers to where the regular season champ ranked. The table below summarizes the findings.
Conference champs by kenpom ranking
Conf Rank (kenpom) Reg. season champ Conf. tourney champ
1 21 18
2 3 8
3 4 4
4 2 0
5 1 (Lispcomb) 0
6 0 1 (Houston)
For instance, for 21 of the 31 conference with a tournament, the regular season champ was the conference’s best team. In 18 of 31 cases, the tourney champ was the best team. As expected, the regular season was a truer test of conference supremacy. But that difference is small, especially when you consider that in only five tourneys did one of the two best teams not win. Compare that to the regular season where eight teams outside the top two were able to snag the tourney’s top seed.
Why is the conference tournament nearly as good of a test as the regular season? Think about the probabilities we assigned before the conference tourney. For instance, we thought Kansas had a 65% chance to win the Big 12 tourney. Now think about the chance that Kansas was truly the best team in the conference based on what we saw during the season. Ignoring, the eye test and just using numbers, I’d guess around 90%. There’s a difference between 90% and 65% of course, but in leagues without a dominant champ, that difference shrinks, and the additional info we get from conference tourney games helps make up the difference. Even in leagues with a dominant champ, the structure of the tourney can help, too. Think about the Horizon, with the double-bye and home court advantage that Butler benefited from.
So the next time you hear someone like Bob Knight bemoan the use of a tournament to decide the conference champ, think about what they’re saying. The regular season isn’t exactly a pristine way to identify a conference’s best team, either. Giving up the excitement (and revenue) that a conference tourney brings is not nearly worth it considering that tourneys also do a good job crowning the best team.
Saturday, February 27, 2010
Saturday morning reading
I am throwing a little Saturday morning reading out there in response to the fine posts authored by Gasaway, Hanner, and Jen regarding whether the selection committee should include margin of victory in their deliberations. Let’s do this with bullet points. Consider it Pomeroy’s wish for a brighter bracketing future.
1) This should not be framed as an MOV vs. non-MOV argument. This is about considering how a team wins and loses as opposed to solely whether they win or lose. If this was the NBA and every team played a similar schedule, this would be unnecessary. But given the imbalances that exist in college hoops scheduling and the brevity of the season, this is a logical and necessary consideration if you want to invite the best teams to the tournament.
I absolutely want the committee to consider the difference between Cornell getting screwed on a last-second call at Kansas, or falling behind 25-2 and losing by 40 at Kansas. The results are the same, but there’s much more useful information to be gathered from each game than just a result. (For a real-life example see Davidson’s 2008 non-conference losses.)
2) I actually don’t particularly care whether considerations are made about how a team wins and loses. If you just want the committee to look at wins and losses and not select the best teams, I could live with that. But the thing is that right now the NCAA instructs its committee members to select the best teams. I am living within that reality. And you cannot evaluate which teams are best without considering how they won and lost.
3) The NCAA Men’s Basketball Committee currently looks at more than just wins and losses. Otherwise, there would be no need for committee members to watch teams play on TV or in person, as they tend to do. Furthermore, as I tweeted Thursday night, they have to look at MOV in the Purdue case. If the Boilermakers lost big against Michigan State, and needed late buckets to beat Indiana and Penn State, they are likely to have their seed affected. If they lost in the last seconds to Spartans and dominated the Hoosiers and Nittany Lions, that would be looked at differently, despite the fact that the outcomes in the two cases are the same.
Additionally, this year’s Butler and last season’s Memphis teams do not have appreciably different resumes in terms of quality wins/bad losses. Yet I’m guessing Butler is not being considered for a two-seed because of the difference in the way each team won its games. Memphis provided a completely different level of domination which the committee took into account during the seeding process.
4) The seeding process definitely needs to consider which teams are better. Referring to point (2), I am negotiable on whether to select teams solely on wins and losses. What really bothers me is that all of our nation’s finest bracketologists will do their best to guess what the committee is going to do, and by and large, they’ll do an accurate job. However, when the brackets are announced, the same people that correctly predicted that Team Awesome would get a six-seed and Team Crappy would get a three-seed will then pick Team Awesome to beat Team Crappy in their second-round “upset special”. Likewise, people with a financial stake in the game will also side with Team Awesome.
All of this because it’s understood that Team Awesome is better than Team Crappy, yet Team Crappy got the higher seed because of a flimsy RPI-style resume analysis. This part of the bracketing process creates unfair brackets, and in this case, it’s unfair that Team Awesome had to draw a tougher first-round opponent because of a process that is less robust than it could be.
5) Mainly, I’m opposed to using a single ranking (and bastardization of it) as the sole objective measure used in the committee’s deliberations. Regardless of how you wish to select the 34 at-large teams, it is foolish in 2010 to rely on one system to provide a foundation to do this. I realize the committee gets more information than just the RPI, but data on the “team sheets” are completely based on RPI and if you read any bracket projection, the sole focus is RPI. Hey, I wouldn’t even want the committee to use my ratings exclusively. I currently have BYU higher than anybody and Villanova about as low as anybody. They are outliers that are corrected by the other systems out there.
And if the NCAA insists on using RPI, then use the team’s ranking directly. If New Mexico is getting credit for beating the 23rd best team when they have a win over Cal, it doesn’t follow that you aren’t allowed to assume that Cal is the 23rd best team when you evaluate the Bears.
Tuesday, February 23, 2010
A belated answer
Thanks for all of the responses to last Thursday’s brain teaser. What bothers me is this – the NCAA instructs committee members to select the 34 best at-large teams. (It says so right here!) Yet the method currently used doesn’t accomplish that. And if the tournament were invented today, I’m fairly sure we’d devise a more intelligent way to select the field.
Let’s take an example from the latest brackets posted yesterday. If you haven’t familiarized yourself with the national bracket, you should do so. It essentially neutralizes the bias or errors of any one bracketeer and gives you a consensus look at who’s in and who’s out. Currently, our country’s bracketologists view Charlotte and Utah State as essentially equivalent teams. For full disclosure, I have not seen Charlotte play this season and I love Bobby Lutz, but they better have some serious extenuating circumstances, because any objective method you use would indicate the Aggies are the better – and a significantly better – team.
I shouldn’t say any objective method, because clearly people are using an objective method to fill a bracket – some sort of bastardization of the RPI. I think everyone would agree (though I’ll hear from Charlotte fans!) that we need some sort of objective input into a process that has to sort out so many teams. And I’m not here to sell you on my system right now, although I think it’s awesome. But you should check out Kenneth Massey’s compendium of rankings for a complete list of options. As I write this, there are 36 systems listed on his page. This will grow by a few in the next day, but right now 35 of them have Utah State ranked higher than Charlotte. The one that has Charlotte higher, has them higher by a single spot.
Now sure, there may be other circumstances to consider, mainly injuries, but in the absence of that, why would you override this vast collection of data and make any case at all that the better team should be out (if they lose in their conference tourney, which is possible)? You might say “quality wins”, “good losses”, “road wins” or “strength of schedule”, but don’t all of these formulas take that into account to some extent?
Just as you wouldn’t want one person to be responsible for selecting and seeding the field, you wouldn’t want one poll, or even one method to be used by committee members. Maybe with a switch to 96 teams, we’ll have a system that can also better identify which teams are truly the best.
If you’ve read my work over the years, you’ve noticed that I rarely use the term “mid-major”. It doesn’t interest me to label subsets of the college hoops universe. I don’t begrudge those who arrive at their interest in the game from cheering on “mid-majors”. But if you do, wouldn’t you want your best and brightest represented come tourney time? Certainly, Charlotte could take down a 5-seed. A win at Louisville basically proves that. But this isn’t about which teams can win a game. (If that were so, Green Bay would be on the board by virtue of their win at over Wisconsin. Clearly, performance over the rest of the season matters.) It’s about picking the best teams. And Old Dominion or Saint Mary’s or Utah State would be more likely to pull off a tourney win (or two).
And if you don’t care about which conferences get how many bids, wouldn’t you rather have a balanced bracket? There’s no need to give one 5-seed a relative break. It’s not fair to the other 5-seeds in the tourney.
Tangentially, here’s an outstanding piece written by Bill James regarding the BCS and how the computer polls used in that system really don’t measure who the best teams are (see section 3), to which I sign off wholeheartedly with respect to current methods. If your mission is to select the best teams, let’s develop a system that identifies the best teams.
Also, I might add that the committee probably should be expanded beyond ten members. I base this on Andy Katz’s column regarding the selection process from a week ago, which had a couple of amusing items.
He interviewed two former members. First there was this…
The committee member said he would look at the bid process as a horse race, putting certain teams one length or two lengths ahead in the process. And then as the conference tournament unfolded he would move the teams up and down his racecourse.
I can only hope the committee member in question was playing a practical joke on poor Andy. I can just see this eccentric individual sitting in the corner of the room with his replica of Churchill Downs, moving little horsies around as he talks to himself during championship week.
But this quote was more relevant to the topic at hand…
“I’ve always argued that you get in if you’re deserving…If we’re just going to say who we think are the best teams, then why do we need a committee? If you’re going to say that then you’re going to gravitate toward the bigger, stronger and faster teams.”
Yeah, because there’s an exact correlation between a team’s size, strength, and speed, and how good it is. Things like game plans and shooting ability are irrelevant. And by the way, anonymous former committee member, we have a committee because different people have different opinions about the quality of teams in contention for at-large bids. And I’d want the opinion of that guy to be minimized by others who are serious about carrying out the instructions established by the NCAA and filling the bracket with the best teams in the game.
Thursday, February 18, 2010
Quick question
Let’s say you were given the following task: From a list of 320 college basketball teams, pick the 34 best. Not the most deserving, not the most difficult to play against, not the ones with the best athletes or the cutest stories. You had to pick the best teams. How would you do it? Would you use the RPI? Except, you wouldn’t actually use a team’s RPI rank, but you’d look at the ranks of the opponents that each team had beaten and lost to. Who would do that? Please e-mail me if you are one of those people. Just wondering if I am nuts for thinking that this is not the best way to accomplish this task.
Wednesday, February 17, 2010
No, Kansas did not choke
It’s rare that I post e-mails in here anymore, mainly because most of the mail I get sucks. So I present this to you both as a thought-provoking piece and as a model for how to converse with me. Please address me as “Mr. Pomeroy” and keep it brief with something insightful for the subject matter. My time is valuable. Be sure to say how awesome I am, too. Also, use paragraphs and punctuation. Sign your work. Be proud of it.
The author, Chris, references one of the first posts I made on the blog, which wasn’t a very good post. But that was before I had even read Basketball on Paper.
As someone who has grown weary of the overuse of free throw shooting as a reason games are won and lost, I can sympathize with Chris’s reaction to old-Ken’s thoughts. But no more! New-Ken endorses the critical thinking behind this e-mail. And since there’s a non-zero chance of a rematch of the ‘03 title game occurring this season, it’s somewhat topical. Anyhow, I’ll let Chris take it away…
Mr. Pomeroy,
Huge, huge fan of your wesbite. I have a blast from the past for you tonight.
Re-watched Syracuse-Kansas 2003 title game tonight. Reminded me of your old blog post, “Did Kansas choke?” http://bit.ly/deST22
As a Syracuse fan, the idea that Kansas choked always bugged me. Not only does it feels like a swipe at the legitimacy Syracuse’s title, but the numbers don’t hold up. Surprising though it may be, Kansas was actually more efficient from the free throw line than Syracuse that night. Take a look at the play-by-play and box score here:
http://sports.espn.go.com/ncb/boxscore?gameId=234000063
Kansas went 12-30 from the line, and missed the front end of 2 one-and-ones. Effectively, that is 12-32. However, Kansas also scored 6 points via offensive rebounds on their missed free throws. So, effectively, Kansas produced 18 points from 32 free throw attempts.
Syracuse went 10-17 from the line. They also missed the front end of 1 one-and-one, and scored zero points from offensive rebounds on missed free throws. So, effectively, Syracuse produced 10 points from 18 free throw attempts.
Kansas: 18-32 for an efficiency rate of 0.5625 per free throw
Syracuse: 10-18 for an efficiency rate of 0.5556 per free throwThus, Kansas was actually slightly more efficient in terms of effective points per free throw attempt than Syracuse.
The 2003 national title game was actually won and lost at the three-point line, not the free throw line. Syracuse shot 11-18 beyond the arc, while Kansas went only 4-20. Whether or not you consider that to be luck might be another matter. However, as a Syracuse fan and a number cruncher, I feel a lot more comfortable discussing Syracuse’s timely three-point shooting than the illusion of Kansas choking from the line.
Keep up the great work. Thanks for reading my email.
Best,
Chris Bowers
Openleft.com