Tuesday, February 23, 2010

Last 11+, What’s Left, Why the Outrage?

Bubble talk is ubiquitous. So rather than add to the monotony, I want to point to a few links, do a few unique things, and then get out of the way. First, Andy Cox’s team sheets are available again this year. These replicate the type of information that is presented to the NCAA committee when comparing teams. If you want to compare two teams head-to-head, click on the at-large tab, then click on the individual team name to see the team sheet. If you want a higher level summary of the Nitty Gritty numbers for free, I recommend Warren Nolan.com. He has a few discrepancies from the official NCAA feed. For example, as of Monday he listed UTEP with only one neutral site game, while the NCAA lists two. (These annoyances are why Ken Pomeroy got out of the RPI business.) But Nolan updates extremely frequently and he’s a must-reference for your RPI top 50 and RPI top 100 stats. I add two handy references to this list:

1) Last 11+

My tracker came about because I find it dumb that most web sites list the last 12 as of today. What I care about is what the last 12 is going to look like at the end of the season. My tracker lists the order of wins in the last 11 regular season games, assuming you’ll play at least one game in the conference tournament. If you lose your first conference tournament game, tack an L on the end. If you win your first conference tournament game, erase the first letter and keep adding to the end.

So let’s say you are a fan of Arizona St. (And if you are, sorry the Pac-10 sucks this year.) Your case for a bid is that you’ve improved as the season has worn on. Even though you started the last 11+ with an L, if you win a game in the conference tournament, you can erase that L. By winning some in the Pac-10 tournament, Arizona St. could be in the at-large discussion with a 10-2 season finish.

On the other hand, Arizona can’t remove its losses from the last 12. At best if they lose in the conference tournament, Arizona will be 7-5 in the last 12. That’s just not going to cut it in a weak Pac-10 this year. Arizona probably had too horrible a profile anyhow, but those recent losses make it impossible.

(One final tangent, I intended to write about USC which has two L’s at the start of the last 12. But USC isn’t playing in the Pac-10 tournament or the NCAA tournament this year. Doh! That’s too bad because USC could make an interesting test case. USC has a crazy 8-3 record against the top 100, but they have 7 bad losses. The product of no offense and fabulous defense is tremendous inconsistency. I was just warming up to the possibility that USC was on the bubble when I remembered the post-season ban.)

2) Best Three RPI Wins

This is a handy reference if you want to argue with people. Why is Georgetown getting the love? They beat RPI #2, 8, and 9. But Pitt should get similar love after beating #4, 5, and 8. On the other hand, Mississippi St.’s best win at this point is #41 Old Dominion. That’s why they were out of Joe Lunardi’s bracket on Monday.

This also helps to remind us of the key non-conference victories. Did you remember that Purdue beat West Virginia and Tennessee? What about St. John’s wins over Temple and Siena?

The chart below is through Sunday’s games. (So you can add UConn’s win over West Virginia manually if you want.) Scroll to the right to see all the columns. I’ll update this at the start of championship week.

What’s Left?

With two weeks left in the regular season, who can still help themselves and hurt themselves? The next table lists the number of RPI top 50 and RPI top 100 opponents left for each team, along with the complete remaining schedule.


As discussed here previously, Northwestern and Ole Miss are effectively done. Neither team has a RPI top 100 opponent left. And neither team has a good enough profile to earn an at-large without serious work in the conference tournament. The team I might add to this list is Seton Hall. Two games against Rutgers and a game against Providence might get them to .500 in Big East play, but probably won’t do enough for the overall profile.

UNLV also has zero games left against the RPI top 100, but at least they are currently projected in the field. Things are a little more precarious in the A10 where Rhode Island has just one big game left at home against Charlotte. Rhode Island’s no lock and that is a must win game.

On the flip side, what do you make of St. Louis? Left for dead at the end of January, they’ve won 6 games in a row. They are a long shot at best at this point, but with four top 100 opponents left, and three of those games at home, I don’t think you can count Rick Majerus out yet.

Louisville also has 4 RPI top 100 opponents left. They face Georgetown and Syracuse at home and Marquette and Connecticut on the road. It is time to find out how good Louisville really is. Rick Pitino’s teams usually play their best at the end of the year, and if that is true again this season, they will improve their seed substantially.

Clemson faces four RPI top 50 opponents the rest of the way. Georgia Tech and Florida St. probably aren’t worth as much as their lofty RPI rating, but with most of these games on the road, Clemson is either going to win some key road games or fall apart down the stretch.

Finally, Florida faces a murders row, facing every team in the SEC East except South Carolina down the stretch. I thought they were done after losing at home to Xavier, but they bounced back nicely winning at Ole Miss, and they have a great chance to prove they deserve a bid by winning some of the upcoming games.

Defending the Selection Process

I’ve long believed college basketball has the best post-season selection system of any sport for several reasons. First, real people evaluate the teams. Unlike a formula which can’t anticipate unusual events, (see Evan Turner’s injury), people can digest information and attempt to give it proper weight.

Second, 34 at-large teams is more than adequate to include all the national title contenders. If your team has ever been Final Four good, you know what a Final Four team looks like, and you know when your team just doesn’t have it. Sure, it is nice to be one of the last teams in, but the 11 and 12 seeds in the tournament are rarely a threat for a deep run.

Third, the committee system is superior to a “pure records” system because there is no point at which everything is clinched. By emphasizing the last 12 in seeding, college basketball teams have to fight hard in every game and play their best at the end of the season. You don’t see good teams resting their players late in the year like you do in the NFL.

Finally, the committee system is also able to reward tough non-conference scheduling. Unlike College Football where Texas can choose to play zero non-conference games against BCS opponents and slide into the national title game, NCAA basketball teams frequently get hammered for weak non-conference schedules. Non-conference schedule has nothing to do with picking the best teams, but it does increase my enjoyment of the entire season by creating more marquee games along the way.

That’s why I find posts, like that by Ken Pomeroy last Thursday, so puzzling. How can he work up genuine disgust and angst at the selection process at this point? Let’s start with his main question? Would anyone use this process to pick the 34 best teams?

The answer is clearly no. Even if the stated goal is to pick the 34 best teams, that’s obviously not what the NCAA committee is trying to do. If it was, non-conference SOS wouldn’t be a real reason to exclude a team from the tournament. Every year, you’ll see one good team whining about getting left out, and the committee chair will say non-conference schedule was a key factor why they were left out.

(If you want to get all crazy, you could argue the NCAA’s goal is to pick the 34 best teams in this year and future years. In other words, they are solving a dynamic problem and not a static problem. And with limited non-conference games it is almost impossible to compare teams across conferences. Thus if the committee has to sacrifice one or two teams each year to ensure better non-conference matchups, that still might result in a better overall evaluation of teams in the long-run.)

But Ken’s complaint isn’t really about non-conference SOS. He’s complaining that record vs the RPI top 50 or 100 doesn’t make any sense. But that calculation has evolved out of a historical process. At one time, record was the key criteria for getting into the tournament. But then people realized “who you play” has a big impact on your record. So they developed the RPI. Should they have moved on to a superior system by now? Of course. But with a committee of real people making the final decisions, I don’t think the ultimate product is that egregious.

But then we get down to the heart of the complaint, which I’ve rambled on about before. The NCAA’s big mistake is that it does not want margin-of-victory to be a criteria in any of its sports. The NCAA does this in order to promote good sportsmanship. But without margin-of-victory, no ranking system can properly evaluate teams. Ken Pomeroy recently decried New Mexico’s high seeding. But it is hard to make a case against New Mexico without the margin-of-victory stats. Look at Jeff Sagarin’s ELO CHESS ranking compared to his PREDICTOR ranking. The former uses outcomes only and the later uses margin-of-victory. New Mexico is 12th in ELO CHESS and 39th in PREDICTOR. Margin-of-victory is clearly useful, but the NCAA does not want to touch it.

So the NCAA ends up with something that is not about selecting the best teams. What you get in the post-season is the best teams and the teams that won their close games. But is that different from any other sport? Do the best 12 teams make the playoffs in NFL? Or do the best teams plus the luckiest teams make the post-season? You play to win the games, and so even if the NCAA ignores margin-of-victory, I’m comfortable with that.

And most importantly, the bracket mistakes are the only reason I can do halfway decent job filling out an NCAA tournament bracket. There are few things I know this year, but here are two of them: Whoever gets New Mexico in their bracket is going to be very happy. And whoever gets Maryland in their bracket is going to be very sad.