Sunday, January 27, 2008

W-7 BP65

1. KANSAS (BIG 12)
1. UCLA (PAC 10)

2. Duke

3. Texas
3. Washington State
3. Indiana
3. Marquette

4. Texas A&M
4. Wisconsin
4. Vanderbilt

5. Pittsburgh
5. Clemson
5. Mississippi

6. USC
6. UConn
6. Notre Dame

7. Ohio State
7. Oklahoma
7. Stanford
7. Kansas State

8. Arizona
8. Oregon
8. Florida

9. Villanova
9. Florida State
9. Rhode Island
9. BYU (MWC)

10. Creighton
10. West Virginia
10. Boston College
10. Purdue

11. Saint Mary's
11. Louisville
11. Baylor
11. Virginia Tech

12. Missouri
12. Virginia



15. YALE (IVY)


Other teams considered, but that just missed the cut:
Miami (Fl), NC State, Wake Forest, Dayton, Massachusetts, Providence, Syracuse, Minnesota, Texas Tech, Illinois State, Southern Illinois, UNLV, Arizona State, Arkansas, Mississippi State, South Alabama

Good resumes, but need a little bit more:
Georgia Tech, Maryland, Duquesne, St. Joseph's, Seton Hall, VCU, Houston, Indiana State, San Diego State, California, Washington, Alabama, Georgia

Other teams with a decent shot, but that really need to improve their resume:
Charlotte, Temple, Cincinnati, Depaul, Illinois, Iowa State, Oklahoma State, UAB, Cleveland State, Akron, Ohio, New Mexico, Utah, Auburn, Utah State

Still alive, but pretty much need a miracle:
St. Louis, Iowa, Penn State, Nebraska, Pacific, UCSB, Tulane, UCF, UTEP, Valparaiso, Wright State, Miami (OH), Bradley, Missouri State, Northern Iowa, Kentucky, South Carolina, San Diego, Santa Clara, Boise State


Jameson said...

Come one... one team from the A10 really? What is your deal with hating the A10. I would take half of those teams over Creighton.

Also give it up with Villanova. They suck...flat out, they suck. They were terrible against Rutgers and pretty bad against Notre Dame. The Wildcats are going to struggle to get to 9 wins in the Big East which won't get them in the tournament because they don't have a strong non-conference schedule.

You obviously need to watch some of these teams play to be able to accurately project where they are going to finish.

Jeff said...

Before going on such an angry rant, I suggest you learn which teams are in which conferences. I have two Atlantic Ten teams in - not one. I've said for a while that I had been projecting Dayton to finish ahead of Rhode Island, and now I'm moving Rhode Island ahead. I still think that conference is getting two teams.

As for Villanova, just look at their remaining schedule. I built this week's BP65 by going through the schedules of all of these teams and trying to project records. I project Nova to finish 20-10 and 9-9 in the Big East. They should then go at least 1-1 in the Big East Tournament, which should put the RPI up into the Top 40. They are going to end up with good numbers against top teams (they're already 4-2 against the RPI Top 100). I think that gets them in.

Anonymous said...

Dayton lost Christ Wright in December with a Broken Foot (at that time he was 1st on the team in rebounding and 2nd in scoring).
He's projected to return towards the end of the season - no later than the A10 tournament

In the UMass game, Charles Little, who plays the same position as CW and last years won the 6th man of the year award in the A10, went out also with a broken foot bone. At that time he was 2nd on the team in rebounding and 4th in scoring.

For the past few weeks Brian Roberts, the senior leader (20+ pts a game) has been suffering with a bad sinus infection. He had to leave the Xavier game becasue he had trouble breathing and has really struggled the past 2weeks - he is expected to be near 100% by the next game

Becasue of injuries and illnesses, dayton has been playing the past 3 games without their top 2 rebounders and 3 of their top 4 scorers.

Once Roberts is healthy and if the team can adjust to other players being out they'll be OK as the NCAA selection committee will take into account that during this losing streak they were not at 100%

Jeff said...

Rather than repeat myself, I will direct you to where I have already responded to the Dayton issues here.

Jameson said...

My bad on the A-10 I missed Xavier because i didn't notice the (A-10) in parenthesis. I'll give you that.

But on Villanova, let's look at their remaining schedule, realistically they should lose at Pitt, Georgetown, and Louisville. It's a toss up at Providence, St. Joes, home against Syracuse, Marquette, Seton Hall, and West Virginia.

Their only gimmes are South Florida and St. John's. So for them to go .500 in the Big East they would need to win four of those six games which at most I can see them winning four.

Go to Villanova blogs and you'll see almost everyone there agrees, this team isn't that good.

They have a poor non-conference schedule which will hurt them a lot.

Evilmonkeycma said...

Was this done before or after IU lost to UConn at home?

Jeff said...

After. It's not a "bad" loss to lose to a team with an RPI and Sagarin Rating inside the Top 25, which is improving weekly.

Anonymous said...

No matter how you slice it, the Atlantic 10 is the best of the non-BCS conferences. The A-10 will be getting AT LEAST three bids. And for someone who is so quick to discredit the RPI, you use it like a crutch when backed into a corner. I find it mighty curious, then, that you refuse to give Dayton any respect for the #17 ranking in the RPI.

Jeff said...

The Atlantic Ten is probably the second best non-BCS conference this season, narrowly behind the Missouri Valley. And because the Atlantic Ten's standings are more spread out (as opposed to the Valley where everyone beats everyone), it is the most likely non-BCS conference to get a third bid.

But, again, you're really overweighing the might of the A-10. I understand that you're a fan of the conference, but to say that they'll get "atleast three" bids is preposterous. I'd say at most three bids. I've said this all year - the Atlantic Ten is closer to getting a third bid than any other non-BCS conference. But I'm only going to give it a third bid if three teams really run away with things.

As for the RPI, it's a worthwhile tool, but it is generally overrated by bracketologists around the country. I understand that the #17 RPI seems great, but I can give you all sorts of examples of teams whose Selection Sunday results don't come close to where the RPI would suggest. Last year's Arizona team finished 14th and got a #8 seed. The year before Missouri State finished 21st and missed the Tournament altogether. The Selection Committee is at the point now where they almost don't even consider the RPI in their deliberations. When they do look at computer numbers they prefer the Sagarin ratings, which are much more complex and accurate tools to rate teams.

If Dayton finishes with an RPI of 17 then they probably make the Tournament (like I said, they'd be in the Tournament with something like a 10 or 11 seed if the season ended now). But the fact is that they're not as good as you think they are. They are going to lose at least three more games in conference, and that will drop the RPI out of the Top 25. If the RPI slips to 30th they're out for sure. If they're in the 25-30 range then we'll have to look at the Sagarin, as well as the other bubble teams.

As a final note, as I said in a previous post, keep an eye on the Sagarin ratings. It's rare for a team with an overall rating outside the Top 50 to get an at-large bid. Right now Dayton is 51st. They need to get that rating inside the Top 50 if they want to get into the Tournament.

Anonymous said...

You are right about Dayton, like how they are ranked in the top 20 in RPI, but in Pomeroy's rankings, their all the way down to 84.

Since you have said before that you are "predicting" the bracket, rather than making a bracket as if the season ended today, Pomeroy's ratings would be pretty beneficial to you since they are purely predictive.

Jeff said...

Yes, you are right that there are other rankings out there. Pomeroy is certainly a good one. I choose Sagarin because I think it's the best, and because it seems to hold the most weight on Selection Sunday. But you're right that I shouldn't be giving the impression that Sagarin is all there is.

The point is to wean people off the RPI, which is just too simple and inaccurate. It has a bias that can be taken advantage of by smart schedule-makers. Not so for Sagarin or Pomeroy.

Anonymous said...

Are you familiar with the pythagorean expectation formula used by baseball statisticians? Is that similar to Sagarin's predictor measure? Just curious.

I still think you're mostly wrong about Dayton, but I can see the glaring weaknesses on their tournament resume. Based on what you see on ESPN (which is what I've been fed for years), a team's RPI is very important. Dayton looks good there. They have a few signature wins, which also helps (also according to ESPN). Looking at Warren Nolan's Nitty Gritty Report, the only thing that he has highlighted as a problem for Dayton's profile is their 2-3 conference record. I would bet my house that the Flyers won't have a losing conference record, and in fact will be no worse than 10-6. So that shouldn't be a problem, either. But that's just my biased opinion.

You know what, I just thought of this, but if you could provide some sort of evidence that the committee prefers Sagarin ratings to RPIs, I'd love to read more about it. Because before I started reading your blog, I'd heard no such thing. Thanks...

Jeff said...

Okay, well I mean you can read people from the Selection Committee talking about how the RPI rarely even comes up in discussions there anymore. But you can also look at how many teams with bad RPIs or Sagarin ratings made the Tournament, and how many with good ones missed. Now, from kenpom I can get end-of-season RPI numbers before Tournament results, but Sagarin includes in Tournament results so the numbers are bit like comparing apples to oranges - but it's a worthwhile endeavor nonetheless.

So, here are the number of teams with an RPI of 50 or worse that got an at-large bid:

2007: 4
2006: 5

And Sagarin Ratings of 50 or worse:

2007: 2
2006: 4

Now how about teams with an RPI of 40 or better than missed a bid:

2007: 3
2006: 4

And Sagarin Ratings of 40 or better that missed:

2007: 5
2006: 4

At first glance, you say that the Sagarin Ratings aren't a much better predictor, especially when you look at Sagarin Ratings 40 or better. But like I said, that INCLUDES Tournament results. So, West Virginia ended up with a Sagarin Rating of 17th last year, but only because they won the NIT. I'm sure their Sagarin Rating wasn't even close to that on Selection Sunday or else they would have gotten a bid.

You can keep going along with other levels of analysis, but the simple fact is that the Selection Committee is not stupid. These more complex computer rankings are simply BETTER than the RPI. So they're used more frequently, and they carry more weight. The RPI is practically being phased out in favor of these better systems.

Anonymous said...

a) You didn't answer my question about the Pythagorean formula used in baseball Sabermetrics.

b) You didn't provide the source(s) I was looking for in regards to tournament committee members prefering Sagarin to RPI.

Jeff said...

Okay, well yes I am familiar with baseball's pythagorean formula. Sagarin's formula is far more complex that baseball's pythagorean formula - so they're not the same.

But one similarity is with the Sagarin PREDICTOR. Basically what you can get out of the pythagorean formula is whether a team is getting lucky. If they've won a lot of very close games. If their pythagorean record is far below their actual record, you can bet on the near-fact that they will lose a disproportionate amount of close games the rest of the way to narrow the gap.

Similarly, if the PREDICTOR is far worse than the ELO_CHESS then the team has won a lot of very close games, and they are likely to lose some close games as the year goes on.

Yes, some teams are more clutch than others (the Yankees always finished a couple games ahead of their pythagorean when Mariano Rivera was in his prime), but you can't beat the system too much. In the end, a lot of it is luck. And luck averages out over a full season.

As for the Selection Committee not using the RPI as much as other things... I'm surprised you didn't know that, honestly. You can do a google search and a thousand links can come up.

All the proof you need is that the Selection Committee is not dumb. They are trying their best to fight their own biases. And they know that the RPI is a blunt object. So they've been making it a smaller and smaller part of the proceedings each year.

Anonymous said...

My surprise has to do with the fact that the Sagarin, like all indicators, is imperfect. I'd be shocked to know that they happen to put more faith in that one, than say, Pomeroy (or anything else). I just always got the impression from major media outlets that the RPI was far and away the most important measure that the committee looks at.

As for the Sagarin PREDICTOR and the Pythagorean baseball formula, that's the original comparison that I made (I wasn't considering the entire Sagarin in the comparison). Anyways, I'd like to learn more about what goes in to his formula (specifically the PREDICTOR aspect), but I couldn't find a description in the few places that I looked. The way you described it, the first thing that jumped in to my head was that it was similar to the Pythagorean. Is it based on total points scored and total points allowed?

Anyways, just because a team has won a lot of close games does not mean that they lose a lot of close games down the stretch to "even things out." It's likely that that will happen, but certainly not a guarantee. A team can continue to get "lucky" (if that is indeed what is happening).

Jeff said...

No system is perfect. But Sagarin and Pomeroy are both far better than the RPI. The people who run the Selection Committee recognize that. You can find the quotes - some people on the Selection Committee actually say that the letters "RPI" sometimes don't come up the entire day.

As for the other issue: when a team has been lucky so far this year, it doesn't necessarily mean that they will be unlucky the rest of the way. But if you think of it statistically, it means that their wins/losses are better than they should be based on their ability level. So for any given time in the future, the most likely result is that they'll play to their expectation value - they will be neither lucky or unlucky (by definition of the words "lucky" and "unlucky"). And if we know that a team has been lucky thus far, it means that their expectation value for the future is inferior performance to what they've shown so far. So you can predict that their seed will drop a bit.

And, yes, the PREDICTOR is like the pythagorean in that it takes into account the scores of the games. Although it is much more complex than the pythagorean for baseball, which is just about some squaring and square rooting.

Anonymous said...

Oh, I understand completely what you're saying in regards to a team being "lucky" (or "unlucky"). You assume that they will regress back towards the mean if their record is better than what other indicators would predict. But the problem is that the sample size of 11-13 games (or whatever Dayton has left) is not big enough to guarantee that things will even out. Now if they were playing a longer schedule, like the NBA does, you'd be more likely to see it all even out. But even then there's no guarantee that it would happen.

If you know where I can find an explanation of the Sagarin PREDICTOR, I'd still love to see it. Even if I don't fully understand everything that goes in to it, I'd like to at least get an idea of what it's based upon.

Jeff said...

You are correct. There is no guarantee of anything. There is some possibility that Dayton will win the national title. If they play out the rest of the season a trillion times, at least one of those times you'll find that Dayton wins the national title. Even these long odds happen from time to time. Are the odds of Dayton winning the title any lower than those of the '05-'06 George Mason team making the Final Four? I doubt it.

Like you put it - with such a small sample size, none of these extreme outcomes are really too unlikely.

All I'm trying to do here is to find the most likely outcome. And mathematically and logically, the most likely outcome is that Dayton will "regress to the mean", as you correctly phrased it. The odds are that their seed will drop. Right now I'd already put Dayton in the 8-11 seed range (5 of the 6 brackets in the bracket matrix that were completed in the past two days agree). So if they drop a little bit, they will fall out of the Tournament.

Anything is possible - I'm just trying to find the most likely outcome.

As for a specific methodology for Sagarin's PREDICTOR... don't bother looking. Most of these guys guard their formulae like KFC's secret recipe. I remember that one of the guys whose numbers were used in the BCS actually was willing to open up and give his formula to the public, in the form of a (more or less) 50 page paper. You can bet that Sagarin's methodology is similarly complex.