Thursday, February 13, 2014

W-4.5 BP68

Starting five weeks out from Selection Sunday I switch to twice weekly bracket projections… so here we are. The next bracket will be out after Saturday night's games, and then again after next Wednesday's games.

The biggest change at the top of the bracket is Wichita State surviving their toughest remaining regular season games. They suddenly are at close to 50% odds of making Selection Sunday undefeated (which would almost certainly earn them a 1 seed). At the bottom of the bracket, none of the bubble teams just barely out of the bracket really made much noise. The biggest mover near the bubble is Oklahoma State continuing to tank.

Two changes were made to the Field of 68, though both came among the automatic bids. Iona is the new MAAC favorite, replacing Manhattan. And Vermont is the new America East favorite, replacing Stony Brook.

Five more teams were eliminated from at-large contention: UCF, Colorado State, Illinois State, Rhode Island and Towson. That leaves 56 teams not in my bracket that still have a chance at an at-large bid. That number will be reduced to around 10-12 on Selection Sunday.

Please remember that the following is a projection of Selection Sunday and not a list of where teams would be seeded if the season ended now. The teams in all-caps are the projected automatic bid winners for their respective conference.


1. FLORIDA (SEC)
1. MICHIGAN STATE (BIG TEN)
1. KANSAS (BIG 12)
1. ARIZONA (PAC-12)

2. WICHITA STATE (MVC)
2. DUKE (ACC)
2. Syracuse
2. VILLANOVA (BIG EAST)

3. Iowa
3. SAN DIEGO STATE (MWC)
3. Michigan
3. Creighton

4. LOUISVILLE (AAC)
4. Kentucky
4. SAINT LOUIS (ATLANTIC TEN)
4. Virginia

5. Cincinnati
5. Iowa State
5. Wisconsin
5. VCU

6. Memphis
6. North Carolina
6. Ohio State
6. Texas

7. Pittsburgh
7. SMU
7. UCLA
7. GONZAGA (WCC)

8. UConn
8. Kansas State
8. Arizona State
8. Xavier

9. New Mexico
9. UMass
9. Oklahoma
9. Tennessee

10. Minnesota
10. California
10. George Washington
10. Georgetown

11. Clemson
11. HARVARD (IVY)
11. Oklahoma State
11. Southern Miss

12. Florida State
12. Stanford
12. LOUISIANA TECH (CONFERENCE USA)
12. TOLEDO (MAC)
12. Missouri
12. Oregon

13. NORTH DAKOTA STATE (SUMMIT)
13. WISCONSIN-GREEN BAY (HORIZON)
13. DELAWARE (COLONIAL)
13. STEPHEN F AUSTIN (SOUTHLAND)

14. UC SANTA BARBARA (BIG WEST)
14. IONA (MAAC)
14. NEW MEXICO STATE (WAC)
14. MERCER (ATLANTIC SUN)

15. BELMONT (OVC)
15. GEORGIA STATE (SUN BELT)
15. AMERICAN UNIVERSITY (PATRIOT)
15. WEBER STATE (BIG SKY)

16. NC CENTRAL (MEAC)
16. VERMONT (AMERICA EAST)
16. DAVIDSON (SOCON)
16. ROBERT MORRIS (NEC)
16. VMI (BIG SOUTH)
16. SOUTHERN (SWAC)

Teams seriously considered that just missed the cut:
Saint Joseph's, Marquette, Providence, St. John's, Baylor, West Virginia, Colorado

Decent resumes, but not good enough:
Maryland, Dayton, Richmond, Indiana, UTEP, Indiana State, Boise State, Utah, Arkansas, LSU, BYU, Saint Mary's

Long shots, but still in the at-large discussion:
NC State, St. Bonaventure, Nebraska, Texas Tech, UNLV, Wyoming, Washington, Mississippi

Still alive, but pretty much need a miracle:
Houston, USF, Georgia Tech, Miami (Fl), Notre Dame, Wake Forest, La Salle, Butler, Seton Hall, Illinois, Northwestern, Penn State, Purdue, UAB, Middle Tennessee, Akron, Buffalo, Ohio, Missouri State, Nevada, Utah State, Oregon State, USC, Alabama, Auburn, Georgia, Mississippi State, Texas A&M, Vanderbilt

25 comments:

Anonymous said...

Syracuse as a 7 on the S Curve? Ha!

Jeff said...

It's a projection. They'd be #1 on S-Curve if the season ended now.

Anonymous said...

Understood that it's a projection. You also projected Syracuse to be an 8 seed early in the season.

They'd probably have to finish the season with 4 losses to not get a 1 seed so basically you're "projecting" them to lose 50% of their games the rest of the way. Sure they have a tough stretch ahead, but when has it ever happened that a team is undefeated through 24 games and then loses 50% of their games at this point in the season?

In the mean time you have Kansas with 6 losses as a 1 seed. Even if you think Kansas is "better" than Syracuse (as an aside, Syracuse is #2 in Kenpom's rankings and Kansas is #10), their win/loss records at this point in the season are too disparate to ignore.

It would be very difficult for a 7-loss Kansas team to get a 1 seed. So basically they'd have to go undefeated the rest of the way to have a realistic chance.

Sure it could happen that Syracuse finishes with 4 or 5 losses and Kansas goes undefeated the rest of the way - there are a ton of things that COULD happen. But how likely is it to happen?

For a blog that relies so heavily on mathematical reasoning, those projections are just surprising.

Jeff said...

Most rating systems project 3 losses for Syracuse between now and Selection Sunday. Projecting 4 isn't that far out there.

LenBias said...

I agree with the other comments.

You say that projecting 4 losses for Syracuse "isn't that far out there" but is it not the case that you are attempting to make your projections based upon probability?

If not probability, what exactly, do you base your projections on? Is it just a gut feeling?

Anonymous said...

Using Sagarin's predictor (which seems to judge Syracuse most harshly), Syracuse's most likely record is 29-2.

Jeff said...

And the probability of Syracuse losing a game in the ACC tournament will be around 75%. So Sagarin projects 3 more losses for Syracuse.

My concern with Syracuse is what we often see with teams that get lucky in a bunch of close games and then have one or two of those close games go the other way... there's a tendency for a decline in play. These are 19-21 year old kids, not robots. We've seen it from a few teams this season, from Baylor to UMass to Providence and others.

Anonymous said...

So, you think a 30-3 or 31-3 Syracuse will not be a #1 seed?

LenBias said...

gut feeling, got it.

Anonymous said...

LenBias, there's nothing wrong with 'gut feeling' or 'projection'. I just ask you don't hold yourself out as some stone cole analytical and efficiency guru when using it.

Anonymous said...

What do you project as Kansas' final record?

Sorry Jeff, you're just digging yourself into a hole. It's not easy to admit that you're wrong but keep in mind...

Mistakes are always forgivable, if one has the courage to admit them.
- Bruce Lee

Jeff said...

If Syracuse loses three games and wins the ACC tournament, they'll probably get a 1 seed. If Syracuse loses three games but loses in the ACC tournament, they probably won't.

I get it, you're a huge Syracuse fan and believe that this entire blog exists to troll Syracuse fans. I'm sure that when Syracuse gets a 2 seed you'll be just as visible in these comments as the Baylor fans who were all convinced I was the world's biggest Baylor hater when I had them as an 8 seed when they were 7th in the polls.

Jeff said...

By the way, I'm pretty sure I "give Syracuse credit" for winning all of those close games. If they'd lost all of those close games I'd have them somewhere around a 7 seed right now.

Anonymous said...

You still haven't answered what you project Kansas' final record to be.

Two more questions for you:

1) What is the percentage likelihood that Kansas will be a 1 seed?

2) What is the percentage likelihood that Syracuse will be a 1 seed?

Anonymous said...

You still haven't answered what you project Kansas' final record to be.

Two more questions for you:

1) What is the percentage likelihood that Kansas will be a 1 seed?

2) What is the percentage likelihood that Syracuse will be a 1 seed?

Anonymous said...

I'm not a Syracuse fan at all, but Ohio State. But I am a fan of analytics, and that I think is where we have a major departure.

Anonymous said...

You also had Syracuse as an 8 seed when they were in the top 10 in the polls.

Anonymous said...

Up to this point, Syracuse has performed pretty much equal to Duke and Virginia in terms of average dominance. They lag behind Duke by about 4 total points scored, and trail Virginia by 9, as reflected by their +/- PPP in conference play. However, Syracuse's PPP standard deviation and variance have been MUCH smaller than Virginia and Duke. They've shown consistency.
Moreover, a 6 point Syracuse win is equivalent to an 8-10 win for most other teams, due to adjustments for pace. Because of the their consistency and pace-free relative, I wouldn't consider Syracuse to be overly "lucky". Most of their close wins aren't as close as they seem when just factoring raw score. It's probably why they are only borderline top third in the country in KenPom's luck metric.

LenBias said...

Anonymous poster at 1:37pm, who posted twice for emphasis, in case Jeff cannot or does not answer your question, Kansas has a 40% chance of being a 1 seed and Syracuse has a 61% chance.

http://www.teamrankings.com/ncaa-tournament/bracketology/

I guess we just have to accept the fact that these projections don't have any sort of mathematical formula or statistics underpinning them, but rather are based upon Jeff's review and interpretation of other projections which incorporates his cognitive biases.

Full disclosure: I am a Syracuse fan.

Jeff said...

Yep, you've figured it out. I'm well known on the internet for my staunch opposition to analytics.

Anonymous said...

I should apologize for the shot at you about not caring about analytics. I'm sorry. It's true that the basis of your projections are based on analytics, but then it seems they are adjusted for your 'feelings' and 'hunches' to the point where even projections by Sagarin and Pomeroy are disregarded. A lot of the talk has been about Syracuse, but I'm also not sure how Arizona is not the likeliest #1 seed.

Anonymous said...

It's just one person's opinion, why are you getting so worked up over it? I like to consult the Bracket Matrix, since it is an aggregate of a whole bunch of individual brackets, one of which is Jeff's

DMoore said...

OK, fine. I'll explain why a prediction of 4 losses for Syracuse before the NCAAs makes complete sense.

First, some background info:
http://johngasaway.com/2014/02/11/tuesday-truths-promotional-synergies-edition/

Look at the analytics there for the ACC. You can see that Virginia, and then Duke, have the best performance in the ACC so far this season. It's even more pronounced than that, because their schedules were front loaded, while Syracuse's is backloaded. Since Syracuse has to go on the road to both of those teams, you would predict two losses for Syracuse from that.

You would also predict that since Syracuse has a MUCH tougher schedule the rest of the regular season than Virginia, that Virginia will have the best record in the ACC going into the conference tournament. That would make Syracuse the 2 seed, which means they will possibly have to beat both Duke and Virginia to win the ACC. I mean the ACC tourney, of course, since it's the tourney winner who is considered the conference champion. So it's a reasonable prediction to expect Syracuse to have a third loss there.

Finally, even though Syracuse will be favored to win their regular season games outside of Virginia and Duke, they are likely to lose one. If you have an 80% chance of winning each remaining game, and you play 5 games, of course you're likely to come out of that 4-1. And, Syracuse will not be that big a favorite to win at Maryland and at Florida State.

I know the ACC seems easy to Syracuse fans so far, but you haven't played a single road game yet against a formidable traditional ACC team yet.

Anonymous said...

That was a decent explanation. However, the converse is also true about the Virginia and Duke games for Syracuse. Using, Sagarin's predictor, they will win at least 1 of those games (and maybe 2) 56.2% of the time. Thus, the most likely outcome of Syracuse playing UVA and Duke on the road is for them to go 1-1.

Anonymous said...

It's not statistically improbable that Syracuse will lose 4 games to close out the season - it's just somewhere south of a 25% probability. If I had to bet, I would unquestionably put my money on three losses. Historically, three-loss major conference teams have been 1 seeds. There has been only one exception this century.