Monday, February 18, 2013

A bias-free look at the NCAA tournament

It's that time of the year again.

Yeah, THAT time. The time when everyone with even a cursory interest in college basketball becomes the foremost expert on who belongs in the NCAA tournament and who belongs in the NIT.

It's both the best and worst time of the year. The anticipation of the next month leading up to Selection Sunday, the feeling that your team might be out if they don't win this game, or if this other team wins their game ... it's definitely part of the fun.

But it's also fairly silly. The idea that one game out of 30 is truly meaningful because a team NEEDS a road win for its resume, or NEEDS a win over a top-50 RPI team, and so on, is just ridiculous. They're all statistics that a committee of people who really don't know what they're doing can point to as a form of revisionist history to explain what really happened in the season.

As an example of how stupid this is, here's a typical example you'll see at this time of year to compare two teams:

TEAM A: 24-7 record, 0-3 vs RPI Top 50, 1-5 vs RPI Top 100
TEAM B: 19-10 record, 3-5 vs RPI Top 50, 5-8 vs RPI Top 100

It is, of course, immediately obvious that one team is a big conference team with a lot of chances for those so-called QUALITY WINS, and the other team is a small conference team that didn't beat anyone of import all season.

What's frustrating, of course, is that anyone with any sort of knowledge about the game at all knows that the way these teams played these games is extremely important. TEAM A may have lost those three games against the RPI Top 50 by a combined 7 points, and those three teams they lost to may all be Top 10 teams. Meanwhile, TEAM B may have won some nailbiters vs. fairly average times based on what amounts to luck, while getting blown out in many of their other games against quality competition.

So, obviously, looking at so-called resumes is really a poor way to go about selecting teams for a tournament. You need to look at the full picture.

---

As with many things in life, you have to know what question you're trying to answer before you spew something out.

One of the main problems with the NCAA tournament selection process over the years is that the question the committee is trying to answer keeps on changing. One year, we'll hear that the committee liked to see a team play a tough schedule even though it didn't necessarily pull out the big wins. Another year, we'll hear the committee placed a lot of value in a few select road wins despite an otherwise unimpressive "body of work."

In other words, the committee often answers one of two questions, only you're never exactly sure which:

1. Which team had the best season?
2. Which team is the better team?

It's a truly important distinction. Take the season we're going through right now, and take two teams in the Big 10: Ohio State and Illinois. Ohio State is 18-7, Illinois is 19-8. You would be pretty hard-pressed to say Ohio State is having a better season. Other than wins over Michigan and Wisconsin at home, they really haven't beaten anyone. On the other hand, Illinois has won AT Gonzaga, on a neutral floor against Butler, and has knocked off both Ohio State and Indiana at home.

But Ohio State is also pretty indisputably a better team. They lost to Duke by 5, Kansas by 8, Michigan State by 3 and Michigan by 2. As mentioned previously, if you have any concept of how to evaluate the game of basketball, you know that such results are a lot more meaningful and predictive of true future performance than wins and losses.

So, to sum this part of the conversation up, the goal of the NCAA tournament is to determine the best team. Therefore, I think the question that needs to be answered is not which team had the best season; rather, which teams are better than the others?

---

There's one final point of contention: How much do you weight recent results? And how much do you penalize a team for an injury that might really hurt them?

Kentucky is the obvious example this season. Although the Wildcats have only played one game without Nerlens Noel, they certainly look like a completely different, non-tourney-bid-deserving squad without him. So, if Kentucky flounders down the stretch, how do you reconcile their recent results without Noel with their results with him? Or what if another team that has been completely ordinary to date -- let's use 13-12 Clemson as an example -- goes on a tear and wins four of their final five regular season games, then makes the finals of the ACC tournament?

Simple: You have to look at the full season.

You might think my answer contradicts with my "best team, not best season" answer above, but it really doesn't: A sample size of a few weeks is pretty non-representative of a team's true skill. It's similar to the notion of a "hot hand" in basketball -- just because a guy makes 4 of his last 5 shots doesn't suddenly make him an 80% shooter. The Noel and other injury situations are more interesting in that you can argue that they're totally different teams without their best players, but I'd rather stick to the full season as a true example of a team's ability than a short sample and a lot of guessing at how good a team "really" is.

In short: You want the best teams over the course of the season.

---

With all of that said, I went through and used my proprietary set of numbers to select the 31 conference champions and 37 at-large teams based solely on the numbers. There was no evaluation of road wins or RPI Top 50 wins, etc. It was simply taking the top team in my system to win each conference, and then the top 37 teams that didn't win their conferences. Without further ado, here goes:


Conference winners:

A10 VCU
ACC DUKE
AEAST STONY BROOK
ASUN FL GULF COAST
B10 INDIANA
B12 KANSAS
BE LOUISVILLE
BSKY WEBER
BSTH NC ASHEVILLE
BW PACIFIC
CAA GEORGE MASON
CUSA SOUTHERN MISS
HORZ DETROIT
IVY PRINCETON
MAAC IONA
MAC AKRON
MEAC NC CENTRAL
MVC CREIGHTON
MWC SAN DIEGO ST
NEC BRYANT
OVC BELMONT
P12 ARIZONA
PAT BUCKNELL
SB MIDDLE TENNESSEE
SC DAVIDSON
SEC FLORIDA
SIND SF AUSTIN
SUM ND STATE
SWAC SOUTHERN
WAC DENVER
WCC GONZAGA

At-large bids (Nos. 1-37):

B10 MICHIGAN
B10 MICHIGAN ST
ACC MIAMI FL
B10 WISCONSIN
BE SYRACUSE
B10 MINNESOTA
BE PITTSBURGH
B10 OHIO ST
MWC COLORADO ST
B12 BAYLOR
SEC KENTUCKY
BE CINCINNATI
B10 ILLINOIS
ACC NORTH CAROLINA
B12 KANSAS ST
ACC NORTH CAROLINA ST
B12 OKLAHOMA ST
MWC BOISE ST
P12 COLORADO
MWC UNLV
MWC NEW MEXICO
BE GEORGETOWN
BE MARQUETTE
P12 STANFORD
WCC ST. MARY'S
MVC NORTHERN IOWA
MVC WICHITA ST
SEC MISSOURI
ACC VIRGINIA
A10 BUTLER
MWC WYOMING
SEC MISSISSIPPI
B12 OKLAHOMA
P12 UCLA
B10 IOWA
B12 IOWA ST
A10 LA SALLE

First four out:
BE VILLANOVA
P12 OREGON
P12 CALIFORNIA
BE NOTRE DAME

Next four out:
A10 ST LOUIS
WCC BYU
BE PROVIDENCE
MVC ILLINOIS ST.

Breakdown of bids by conference:

Big 10 (8): Indiana, Michigan, Michigan St., Wisconsin, Minnesota, Ohio St., Illinois
Big 12 (6): Kansas, Baylor, Kansas St., Oklahoma St., Oklahoma, Iowa St.
Big East (6): Louisville, Syracuse, Pittsburgh, Cincinnati, Georgetown, Marquette

MWC (6): San Diego St., Colorado St., Boise St., UNLV, New Mexico, Wyoming
ACC (5) : Duke, Miami, UNC, NC State, Virginia
Pac 12 (4): Arizona, Colorado, Stanford, UCLA
SEC (4): Florida, Kentucky, Missouri, Mississippi
A10 (3): VCU, Butler, La Salle

MVC (3): Creighton, Northern Iowa, Wichita St.
WCC (2): Gonzaga, St. Mary's

American East (1) : Stony Brook
Atlantic Sun (1): Florida Gulf Coast
Big Sky (1): Weber St.
Big South (1): NC Asheville
Big West (1): Pacific
Colonial (1): George Mason
Conference USA (1): Southern Miss
Horizon (1): Detroit
Ivy (1): Princeton
MAAC (1): Iona
MAC (1): Akron
MEAC (1): NC Central
NEC (1): Bryant
Ohio Valley (1): Belmont
Patriot (1): Bucknell
Sun Belt (1): Middle Tennessee
SoCon (1): Davidson
Southland (1): SF Austin
Summit (1): North Dakota St.
SWAC (1): Southern
WAC (1): Denver


A few surprises, obviously:

* Notre Dame is a No. 6 seed in Joe Lunardi's breakdown, but I have them missing the tournament currently. Of course, this is the bias-free world ... in the real world, a 20-6 Notre Dame team is solidly in, no matter how they played to get there. I think they will stumble repeatedly down the stretch and actually turn into the bubble team I currently have them as.

* Oregon is another team Lunardi has pegged as a No. 6 seed that I have on the outside looking in. This may seem surprising, but when you look at their results -- and I don't want to get into this type of breakdown too much because it's the very thing I'm trying to avoid -- you'll see that they have three wins that might warrant some positive impressions: AT UNLV, AT UCLA, and at home against Arizona. Beyond that, they've had some very narrow (read: potentially lucky) wins over the likes of Arizona State, USC, Washington and Washington State. The Ducks are overrated thanks to a good record in an exceedingly poor Pac 12.

* The Mountain West gets six teams in. Boise State is a surprise but widely presumed to be on the bubble ... Wyoming is the real surprise. But you have to remember ... this is BIAS-FREE. This team started 13-0 while beating Colorado, Illinois State and Denver, and has since notched a nice win over San Diego State. They've also narrowly lost to New Mexico and Boise State (twice). This team is 18-7! Let's put it this way: Give Notre Dame a resume like Wyoming's and they're solidly in. This is the problem with the current process.

---

From here, let's just go straight to matchups. I'm ignoring geography because that's a total joke, too -- just pairing the top team with the first four winners, and so on. I'll also predict scores as we go along. I should note that just because one team was rated above another in my overall listing does not necessarily mean that team will win -- my system accounts for a bunch of factors that may predict some (mostly mild) upsets. Here goes:

FIRST FOUR:

#16 George Mason def. #16 Bryant 68-65
#16 Florida Gulf Coast def. #16 Southern 69-67
#12 La Salle def. #12 Iowa St. 75-74
#12 UCLA def. #12 Iowa 73-72

That would be a fun first four, except for the state of Iowa. Let's hope the games are even remotely that close.

SECOND ROUND:

#1 Florida def. #16 Florida Gulf Coast 80-55
#9 Marquette def. #8 Georgetown 61-60
#5 Colorado St. def. #12 La Salle 72-65
#4 San Diego St. def. #13 North Dakota St. 61-57
#6 VCU def. #11 Butler 68-67
#3 Syracuse def. #14 NC Central 73-60
#7 Kansas St. def. #10 Virginia 60-58
#2 Wisconsin def. #15 Weber St. 64-56

#1 Louisville def. #16 Iona 86-71
#9 St. Mary's def. #8 Colorado 70-67
#5 Cincinnati def. #12 UCLA 72-70
#4 Ohio St. def. #13 SF Austin 58-48
#6 Illinois def. #11 Oklahoma 70-65
#3 Gonzaga def. #14 Denver 69-60
#7 Boise St. def. #10 Northern Iowa 67-66
#2 Duke def. #15 Davidson 77-63

#1 Michigan def. #16 NC Asheville 80-60
#8 UNLV def. #9 Stanford 68-67
#5 Kentucky def. #12 Belmont 70-67
#4 Kansas def. #13 Detroit 77-70
#6 UNC def. #11 Ole Miss 79-75
#3 Pittsburgh def. #14 Bucknell 63-55
#7 Oklahoma St. def. #10 Wichita St. 64-62
#2 Michigan St. def. #15 Stony Brook 65-56

#1 Indiana def. #16 George Mason 80-59
#8 New Mexico def. #9 Southern Miss 67-63
#5 Baylor def. #12 Middle Tennessee 71-67
#4 Arizona def. #13 Akron 71-64
#6 Creighton def. #11 Wyoming 63-62
#3 Minnesota def. #14 Pacific 66-58
#10 Missouri def. #7 NC State 79-78
#2 Miami def. #15 Princeton 66-57

Third round:

#1 Florida def. #9 Marquette 76-64
#5 Colorado St. def. #4 San Diego St. 65-64
#3 Syracuse def. #6 VCU 73-69
#2 Wisconsin def. #7 Kansas St. 63-58

#1 Louisville def. #9 St. Mary's 71-66
#4 Ohio St. def. #5 Cincinnati 67-63
#3 Gonzaga def. #6 Illinois 72-70
#2 Duke def. #7 Boise St. 76-69

#1 Michigan def. #8 UNLV 74-66
#4 Kansas def. #5 Kentucky 69-66
#3 Pittsburgh def. #6 UNC 69-65
#2 Michigan St. def. #7 Oklahoma St. 65-62

#1 Indiana def. #8 New Mexico 71-64
#4 Arizona def. #5 Baylor 72-69
#3 Minnesota def. #6 Creighton 72-67
#2 Miami def. #10 Missouri 72-68

Sweet 16 and Elite 8:

#1 Florida def. #5 Colorado St. 69-61
#2 Wisconsin def. #3 Syracuse 66-63
#1 Florida def. No. 2 Wisconsin 62-58

#1 Louisville def. #4 Ohio St. 67-66
#2 Duke def. #3 Gonzaga 73-72
#2 Duke def. #1 Louisville 72-71

#1 Michigan def. #4 Kansas 70-65
#3 Pittsburgh def. #2 Michigan St. 61-60
#1 Michigan def. #3 Pittsburgh 65-62

#1 Indiana def. #4 Arizona 73-69
#3 Minnesota def. #2 Miami 65-64

#1 Indiana def. #3 Minnesota 73-68

Final Four:

#1 Florida def. #2 Duke 70-65
#1 Michigan def. #1 Indiana 73-72

#1 Florida def. #1 Michigan 69-64

(As mentioned, there are some SLIGHT variations in my formula ... for instance, Indiana would be predicted to fare slightly better against Florida, losing 71-67, despite being predicted to lose against Michigan)

Anyway, the takeaways are as follows:

1. Take a look at the predicted scores again. This tournament is going to be TIGHT. I mean, REALLY TIGHT. Everyone says this is going to be a really fun March, and for once at least, the general narrative is pretty much spot-on. Except for ...

2. Florida is really, really good. I absolutely cannot understand how this team isn't considered a shoo-in for a No. 1 seed. There's really only one thing they don't do all that well: Get to the FT line. Beyond that, they are FAR AND AWAY the best team in the nation. They beat a very good Wisconsin team by 18. Marquette by 33. Missouri by 21. Ole Miss by 14. Kentucky by 17. And let's look at their losses. Arizona pulled a rabbit out of the hat by storming back for a 65-64 win on Dec. 15. Florida led 64-58 with less than a minute remaining, and somehow Arizona pulled it off. Florida's win probability was around 98% at that point, so that's just bad luck. Kansas State beat Florida the following Saturday, Dec. 22, 67-61. The Wildcats shot 29 FTs in that game as Florida was whistled for 20 personal fouls. Finally, there was the one real stinker put up against Arkansas where the Gators trailed by 27 at one point before making it look respectable down the stretch in an 80-69 loss.

Know how many other games they've played that have been decided by less than 10? ZERO. Yes, the SEC is down, but they're just destroying everyone. They're obviously very capable of losing with a poor effort -- we saw that against Arkansas -- but how anyone can see this team as anything other than the best in the country is beyond me.

3. Don't get too worked up about the lack of upsets. Obviously when the games are expected to be this close, it doesn't take much for a game to swing in either direction. Even seemingly clear-cut games take a surprising direction more often than you think they should, so a lot of these one or two-point games could be anyone's to take. In fact, there were a surprising number of games where the margin I came up with was LESS than 1, so take heart that I don't honestly expect 90% of higher seeds to win. It just doesn't make a whole lot of sense to arbitrarily pick and choose which slight underdog I expect to win when the numbers say they're only going to win 49.4% of the time.