Sunday, March 15, 2009

Flawed Bracketology

Selection Sunday, one of the best sporting days of the year. Selection Sunday, the only reason Joe Lunardi makes more money than you do.

His Last Four In (Pomeroy Ranking):

Minnesota (45th)- The Big Ten is ridiculously overrated this year, but the Golden Gophers get a pass from me because they aren't Penn State. They beat Louisville on a neutral site, a good win even if it was at the beginning of the year. I don't get how 9-9 is clearly good enough in the Big Ten (not listed in the "Last Three In") when the Irish are well off the bubble at 8-10 in a much better conference and ranked eight spots higher by Pomeroy.

Maryland (54th)- 7-9 puts the Terrapins in but the Irish are well out at 8-10? Not that I think Notre Dame deserves to be in the Tournament (they don't), but this is the best example I know of how erratic basketball seeding is. Beat UNC and Miami at home, but was absolutely pounded by Georgetown on a neutral site. Ranked 28 places below the Hoyas by Pomeroy.

Creighton (68th)- Talking heads have to be careful talking about mid-majors because they don't want to offend anyone. However, there's a reason high-major teams take almost all of the at-large bids. They're better, it's not even close. With the exception of George Mason in 2006, a team that was actually seeded far too low at 11, mid-major schools just don't stack up against ACC and Big East programs. The Patriots' numbers ranked them 23rd that year. If there's anything I've come to learn about Ken Pomeroy's computer rankings, it's that they are the single best predictors of future success that I have found in college basketball. 23rd is good, 68th is not.

St. Mary's (60th)- Speaking of overrated small school teams, I give you the Gaels. Much is being made about how they lost Paddy Mills in the middle of the season and could have contended for a WCC title with him healthy. But who have they beaten? San Diego State? Utah State? Providence is decent, too, but no other wins over the top 100. The game of basketball is about winning, not trying really hard with your leader out.

Teams Who Should Be In-

Arizona (39th), San Diego State (34th), New Mexico (38th)- The Wildcats have fallen down the stretch, but they have some really big wins to their credit. Beating Gonzaga, Kansas, Washington, and UCLA as part of your nine top 100 wins is much better than St. Mary's joke of a resume. The Gaels did beat San Diego State head-to-head, but the difference between strengths of schedule is enormous. The Aztecs played a non-conference schedule ranked about 160 spots above Creaighton. If you're looking for a mid-major to support, why not one that's actually battle-hardened? New Mexico is another good option, having beaten Creighton by seven on the road. Their quality wins have come in conference, but BYU and Utah are ranked much higher than anybody the last two teams have defeated.

Team Who Should Be Really Really Far Out of the NCAA Tournament-

Penn State (78th)- 10-8 puts you on the bubble in the Big Ten, but Big East team Providence is sniffing a 3 or 4 seed in the NIT. In the words of Dick Vitale, "Are you kidding me?" Sure, they beat Purdue by three at home, but have since been clearly out of their element against the Boilermakers. Michigan State is a good win, same with two against Illinois, but 38-33 is an incredible load of filth. Their defense, one of the few remaining excuses for Big Ten apologists, is worse than Notre Dame's (117th).


Craig said...

The issue begins and ends with RPI. If you want to know why ND is out (way out) and Maryland is likely to be in, look at their RPIs.

As much as I like Pomeroy, the NCAA is never going to favor it because it's production-based rather than results-based.

BlackandGreen said...

The Committee has said it will not take RPI into account this year. I don't know how true that is.

My problem with the RPI is twofold: it is easily manipulated and poorly used. Leagues like the Missouri Valley made their names in the middle of this decade by beating teams in November and December that would help their RPI. Not necessarily scheduling competitive teams, but other mid-major schools ranked in the top 100. You think Dayton was the 13th best team in the country in 2003? Should Drake have been #10 last year? Both were the victims of big first round upsets. Southern Illinois (7th) and UNLV (10th) both made the Sweet 16 in 2007, but the Selection Committee obviously though Kansas (11th) was much better team than either of those two.

ESPN likes to post RPI, SOS, and record vs. top 50 teams in their bubble profiles. It's a nice snapshot, but gives the wrong impression. A team like Notre Dame played very different top 50 opponents than anybody in the Big Ten. It's very different to beat #50 Virginia Commonwealth than #1 Pittsburgh. Just looking at one's record against top 50 opponents does not tell the whole story.

Craig said...

Part of the problem with RPI (contributing to the MVC gaming it, and it becoming marginalized) is the way they built social engineering into it to try to discourage teams from playing all their non-conference games at home. Be that as it may, if they went back to the original formulation (or at least cut back a bit on the home/road split), it would still be highly flawed, but it would be considerably more useful than it is now.

While I prefer Pomeroy, there's something to be said for preferring the results to how you achieve them. In particular, as far as I can tell, Pomeroy tends to give equal weight to offense and defense, but historically defensive teams have performed better.

I agree with you about how misleading the top 50 breakdown is. Top 50 is still too broad of a brush; you really need to cut out the top 10 or 20 at the top end of that.

BlackandGreen said...

You make a good point about gauging productions vs. results. West Virginia, a 6th seed, was the 8th best team in the country by virtue of Pomeroy's calculations. Georgetown didn't deserve a bid even though they outranked almost half the at-large bids.

Pomeroy's formula is great for predicting future success, but in the end you have to get it done.

MJenks said...

I think San Diego State got the shaft.

I'm a big supporter of the small schools, but I don't think St. Mary's should have gotten in this year. Unfortunately, the whole RPI thing is a problem because a lot of mid-majors don't get that opportunity to play the larger schools, especially not in a home-and-home. Of course, St. Mary's and SDSU both played Arizona this year (and both lost to Arizona), so that could punch some holes in my argument.

I'm always dismayed when they show the at-large numbers and the mid-major schools keep getting fewer and fewer in each year. Of course, with such a glut of lousy bubble teams this year, it makes more sense that the bigger schools win out on Selection Sunday. I wish there was a way to bring more parity to the schedule, but it's always going to feature the big schools padding their pre-conference schedules with tiny schools and then letting their conference schedule bring their RPI back up (Boston College, I'm looking at you).