Net Efficiency Margin offense (blue)/defense (red) splits (all charts here are 5-game running average NEM)

I've already shown the neutral effect of Kendall Marshall on UNC's defense compared the positive impact on their offense. The Heels have dropped off a little defensively since then, but are still very solid. There has still been inconsistent offensively, typical of a young team.

Kentucky's chart shows a roller coaster effect, with their offense dropping when their defense peaks and vice versa. Recently they're playing well on both ends of the floor.

The Heels and Wildcats have already played this season, but that game was nearly four months ago. For two teams that are among the least experienced in the country, using that game to assess today's matchup is not terribly useful. Not to mention the fact that Kendall Marshall, who has averaged more than 30 minutes per game since taking over the starting job, played only ten minutes in that game.

This game could be the ultimate test of the value of a good team on a hot streak against a consistently strong team.

Net Efficiency Margin offense (blue)/defense (red) splits (all charts here are 5-game running average NEM)

VCU has been playing out of their minds offensively in March, and the defense hasn't been too shabby either.

Kansas is now the overwhelming favorite to win the national championship, with Nate Silver's 538 bracket showing their title chances close to 50% going into today. The Jayhawks have played increasingly better in the tournament according to NEM (28.5 vs. Boston University, 42.9 vs. Illinois and 48.3 vs. Richmond).

The last CAA team in the Elite 8 (George Mason in 2006) played that game against UConn a short drive from campus. A VCU win today would be the most surprising upset of this year's tournament, especially with the decided crowd advantage for Kansas, and would be a bigger upset than the last CAA team to make the Final Four.

Connecticut and Arizona can credit their surge to the Elite 8 on their offenses peaking at the right time.

Net Efficiency Margin offense (blue)/defense (red) splits (all charts here are 5-game running average NEM)

While Arizona had a mid-season offensive mini-peak in February, Connecticut's offense was above average but relatively stagnant after their fast start in Maui. Beginning with their loss to Notre Dame just before the Big East tourney, the Huskies have averaged 23.8 offensive NEM with no game below 12. For some perspective, a team averaging 12 offense NEM would rank around 25th nationally.

Connecticut has outpaced the Wildcats defensively all season and so far in the tournament, so I'd give the slight edge to the Huskies. Of course, the way Arizona shredded Duke's top ten defense in the second half Thursday, their own defense may not matter.

Since their February 3 loss at Youngstown State, Butler has won 12 straight games and found themselves one more win from consecutive Final Four trips. As this chart shows, the Bulldogs have done it by playing, well, bulldog defense.

Net Efficiency Margin offense (blue)/defense (red) splits (all charts here are 5-game running average NEM)

The offense has stayed about the same, but the defense has improved significantly since that game. They actually played good D at the beginning of the season, but several bad games in January caused the poor spike in the middle. The last nine games in particular have been solid; their average -13.7 defensive NEM over that stretch is roughly top ten quality.

During the tournament, Butler has relied on their defense to carry them this far. Their 11.8 offensive NEM average during the tournament is the lowest of the eight teams remaining. Of course, defense was also how the Butler did it last season, and that worked out pretty well. They may need the offense to pick it up to beat a more well-balanced Florida team.

Here are the update combined net efficiency margin (NEM) ratings for the teams that played in the Sweet 16. Results for all teams through the first two rounds can be found here.

Teams are ranked by combined NEM for all games played. Larger numbers are better for total and offensive NEM, and smaller numbers are better for defensive NEM. Shaded teams are still alive going into the Elite 8.

The Sweet Sixteen games demonstrated just how fickle tournament performance can be. The four lowest teams in total NEM through the first two games (Butler, Kentucky, North Carolina, and Arizona) are still alive, while four of the top seven have gone home.

Also, at this time last year three of the four regions were won by the team with worse NEM tournament performance to date. So Connecticut, VCU, Florida, and North Carolina shouldn't count on their tournament performance so far to carry them to Houston.

Here is a complete list of all NCAA Tournament teams with their combined net efficiency margin (NEM) through the first week. (I excluded the First Four games to avoid skewing the totals.) As a reminder, net efficiency margin measures a team's performance in a game in points per 100 possessions (the "efficiency" part) compared to how a hypothetical average team would have performed against the same opponent (the "net" part). Factoring out the tempo and opponent allows comparison of teams' performance from the same baseline. I posted reviews of the best performances from what I like to call rounds one and two earlier.

Teams are grouped by the number of games played, then ranked by their total NEM for the tournament. The numbers are totals across all games. Larger numbers are better for total and offensive NEM, and smaller numbers are better for defensive NEM. Shaded teams are still alive after the first two rounds.

Nate Silver, formerly with Baseball Prospectus and fresh off moving his 538 political analysis to the New York Times, jumped into the March Madness fray this year. Mr. Silver used a combination of computer power rankings and seeds, with adjustments for geography, injury, and other factors, to generate probabilities for each team reaching a particular round in the bracket.

Now that we're down to a manageable 16 teams, I thought I'd play around with some of his numbers to calculate the probabilities of some interesting results.

These all assume the outcome of each region is independent. I think my math is right, but feel free to correct me if you notice anything wrong.

72.5% probability of a #1 seed winning the championship

18 champs have been #1 seeds - most since seeding began in 1979

23.3% probability of three #1 seeds in the Final Four

Three or more #1 seeds have made it four times since seeding began

19.4% probability of an Ohio State vs. Kansas final

The most likely championship game pairing

12.7% probability of #2 seed winning the championship

Six champs have been #2 seeds - second most since seeding began

10.4% probability of Duke and UNC meeting in the Final Four

Also known as "The Game That Must Not Be Named"

5.7% probability of all-Big Ten final (Ohio State vs. Wisconsin)

Would be the first since Indiana-Michigan in 1976

3.1% probability of #10 seed or lower winning

Villanova (#8 seed) in 1985 is the lowest seed to win the championship

1.2% probability of a rematch of last year's Duke/Butler championship game

A championship game rematch has not happened since 1961 and 1962 (Cincinnati defeated Ohio State both years)

0.7% probability of all-SEC final (Kentucky vs. Florida)

There has never been an all-SEC final, despite ten championships between three teams

0.4% probability of both remaining Big East teams making the Final Four

More likely, there is an 87% chance neither will

While Joe Sheehan's scenario is no longer possible, there are still enough non-Big Six conference teams left to make an impact on the Final Four. But how much impact will they make?

Probability of a non-major team in he Final Four by region

Southeast (BYU or Butler) - 32.4%

West (SDSU) - 31.2%

Southwest (Richmond or VCU) - 14.5%

Probability of three non-majors in the Final Four - 1.5%

Probability of no non-major Final Four teams - 39.8%

Probability of a non-major national champion - 8.9%

Hey, it could happen.

Sweet 16 previews to come later this week.

All raw efficiency data is from the sublime kenpom.com. I used the ratings through last Sunday's games to make the comparisons. (For my second round reviews, I'll update the ratings through the end of the first round, and so on.)

Best games (by combined total NEM)

Top overall performances

Top offensive performances

Top defensive performances (lower scores are better)

One game by itself can be misleading, but net efficiency margin (NEM) at least gives us a way to compare each team's performance from the same baseline. So who had the best games over the first two (real) days of the tournament?

All raw efficiency data is from the sublime kenpom.com. I used the ratings through last Sunday's games to make the comparisons. (For my second round reviews, I'll update the ratings through the end of the first round, and so on.)

Top overall performances

Top offensive performances

Top defensive performances (lower scores are better)

Once everyone has played a game, I'll start posting net efficiency margin (NEM) results as a tempo-free and opponent-strength-independent team performance metric. If you still need some help with your bracket, here are a few teasers.

All raw efficiency data is from the sublime kenpom.com. I used the ratings through last Sunday's games to make the comparisons.

(6) St. John's vs. (11) Gonzaga - Southeast region, Denver pod

These are two teams heading in opposite directions, and it doesn't get any better for St. John's without D.J. Kennedy.

(6) Cincinnati vs. (11) Missouri - West region, Washington pod

Another case of teams heading in different directions, Cincinnati's Big East tourney loss notwithstanding. Missouri's downward trend has been going for a while.

(6) Georgetown vs. (11) Virginia Commonwealth - Southwest region, Chicago pod

Georgetown is expectd to regain the services of Chris Wright, who went out with a broken left hand on February 223. Since then, the Hoyas have not been themselves. Their NEM trend (blue is offense - higher is better - and red is defense - lower is better) shows the dropoff has come almost entirely on offense, so if Wright can step back in the Hoyas could make some noise.

Follow along with my Selection Sunday observations here.

A BubbleGrid update and a few notes for your Sunday lunch. Less than five hours to go.(See this to learn what it's all about.)

• Chris Wright has reportedly been cleared to play for Georgetown in the NCAA tournament, so I'll probably leave their seed alone.
• Ohio State and Kansas are locks for #1 seeds and probably the top two overall in that order. Pitt should probably be the third, but Duke (pending today's outcome) and San Diego State are also in play for #1 seeds. I don't think North Carolina deserves a #1 seed if they win today, but I can see the committee doing that. They have a history of overseeding (in my opinion) teams that win their conference tournaments.
• SI.com's Andy Glockner gives Georgia the nod over Alabama, noting "there's no precedent for the committee to reward a team with an RPI like this that plays a terrible nonleague schedule this poorly." Kentucky in 2008 was similar, though their poor non-conference RPI (210) was caused more by their record (6-7) than the strength of schedule (99). Kentucky's conference performance (12-5, RPI 8) that year swayed the committee, and Alabama hopes for a similar result this year. Unfortunately, the Tide's numbers (236 non-conf RPI, 25 conf RPI) don't quite match what Kentucky did in 2008, so I agree with Glockner on this one.

BubbleGrid

• Hover over a value for more information
• Shaded teams are predicted to be in the field
• Teams with a * are the projected automatic qualifier (AQ) for their conference. Bold shaded teams have clinched the AQ berth
• Teams are listed in order of their selection confidence by the CTD model. In other words, the last shaded non-AQ team is the last team in, and the first non-shaded team is the first team out.

As others have noted, in previous years the first 30 or so (of 34) at-large teams are clear. After that, we're all just guessing.

By now, everyone knows there are now 37 at-large spots available. That means we're all just guessing at three more teams. Actually, it's worse because teams become (generally speaking) more bunched together as we move toward the median. Fortunately, we're not looking at the 172nd and 173rd best teams for the last few at-large spots, but as more at-large teams are added to the field, it gets harder to differentiate around the cut line.

This will make it more difficult for the committee to decide on those last few teams, and it will make it more difficult for those who try to predict how the committee will choose those teams. There's already been a fair amount of variation on the Bracket Matrix site, though that tends to firm up on Selection Sunday. I wouldn't be surprised to see the average number of at-large predictions missed to be worse than in years past.

If the CtD model misses any at-large teams, which I expect it will, I think it will come from the following pool: UAB, Clemson, St. Mary's, Alabama, Georgia, and VCU. Outside of that, I feel pretty good going into Selection Sunday.

It's homestretch time, with three more days to go. All conference tournaments have started, and 18 of the 31 automatic bids are still up for grabs.

Nobody

• Colorado

Completed the triple sweep of likely at-large Kansas State. With additional wins against Texas and Missouri, they may be hard for the committee to ignore.

• Georgia

Their win over a putrid Auburn team doesn't clinch anything, but it keeps them in the hunt.

• Notre Dame

If they weren't in the discussion for a #1 seed, they are now.

• UAB

The top seed in the CUSA tourney goes down, and they were probably overvalued by the selection model anyway.

• BYU

A close win against a bad TCU team can't help their post-Davies resume. Next up is their nemesis, New Mexico. I'm close to applying a seed drop based on how they've looked without Davies.

• Patriot League - Lafayette at Bucknell

Basically, everyone who is still on the bubble and still playing has a big game today. This includes Clemson, Virginia Tech, Michigan, Georgia, Penn State, Michigan State, Memphis, and Alabama, to name a few. The bubble is so weak and fluid this year, it would be a bad idea to give the committee a reason to doubt you.

Three days to go. (See this to learn what it's all about.)

• Hover over a value for more information
• Shaded teams are predicted to be in the field
• Teams with a * are the projected automatic qualifier (AQ) for their conference. Bold shaded teams have clinched the AQ berth
• Teams are listed in order of their selection confidence by the CTD model. In other words, the last shaded non-AQ team is the last team in, and the first non-shaded team is the first team out.

As I've noted before, no computer model is perfect, especially one that attempts to predict the behavior of unpredictable people. Some of its whiffs come out of the blue, but some are obvious if you look hard enough. Here are a few teams that I think the selection model is currently overvaluing.

I don't know where they'll show up tomorrow, but as of today the model had them comfortably in. Oddly enough, the loss to ECU bumped the Pirates into the top 100, which (marginally, albeit) increases the value of UAB's regular season sweep.

One problem is that they technically have a top-50 win, but it was against the #50 team (VCU) and it was at home. They are 9-7 against top 100, which is better than many of the bubble teams, but only 1-4 against top 50, and that one is just barely top 50.

Each has a bad overall record, which the committee will certainly notice. For each, a run in the Big Ten tourney could mitigate that factor. Penn State is slightly more overvalued than the Spartans. They have a similar profile on the surface to Michigan and Michigan State, but it is qualitatively worse. Their best road win was at #75 Minnesota and the overall quality of their wins (and losses) don't stack up. (Comparison table described here.)

Wins

Losses

Wins

Losses

The Tigers have four losses outside the RPI top 100. Their best road win is at #71 College of Charleston. Their 8-6 record against the top 100 is good, but somewhat deceptive. Five of those wins were over teams ranked between 71 and 100.

Harvard's non-conference RPI (30th) is very good, but misleading. They played four non-conference road games against the top 57, which boosts their RPI, but they only won one. Their selection score is 7.8 (out of 100), meaning if they lose to Princeton, they'll probably fall out anyway, so I'm not too worried about them.

Several more bids were doled out, and the big conferences got started Tuesday.

• Horizon - Butler 59, Wisconsin Milwaukee 44
• Sun Belt - Arkansas Little Rock 64, North Texas 63
• Summit - Oakland 90, Oral Roberts 76

• Princeton

The Tigers forced a one-game playoff with Harvard for the Ivy Leagu's automatic bid by beating Penn at the Palestra. The playoff game will take place Saturday at Yale.

• Villanova

Villanova is in no danger of missing the tournament, but their seeding will suffer from the ugly loss to South Florida. The Wildcats have now lost their last five and 10 of 15 after starting 16-1. As expected in the Big East, many of the losses were to good teams, but the margin in some of these should be troubling. After starting with a NEM of 26.7 through their January 22 win at Syracuse, Villanova has dropped off to 12.9 over their last 12.

• Big Sky - Montana at Northern Colorado
• Northeast - Robert Morris at Long Island

Another four bids are up for grabs today, but otherwise it's a slow Monday. Six days to Selection Sunday.

• Missouri Valley - Indiana State 60, Missouri State 56

• St. Mary's and Gonzaga

CtD has each in as of today, but neither could have afforded a bad loss in the WCC tournament. The loser of today's championship game still isn't safe.

• Conference top seeds

It was a bad day for several conference tourney #1 seeds. Fairfield (MAAC), George Mason (CAA), Vermont (America East), Missouri State (MVC) lost in their conference tournaments. Among those, George Mason is the only team with a realistic chance at an at-large bid. As of today, the Patriots are in the field

• CAA - VCU vs Old Dominion
• Metro Atlantic - St. Peter's vs. Iona
• Southern - Wofford vs. College of Charleston
• WCC - St. Mary's vs. Gonzaga

Sunday is a slow day on the calendar. Only one conference championship game today, and few games with bubble implications. Seven days to Selection Sunday.

• Atlantic Sun - Belmont 87, North Florida 46
• Big South - UNC Asheville 60, Coastal Carolina 47
• Ohio Valley - Morehead State 80, Tennessee Tech 73

• Harvard

The Crimson clinched at least a share of first place in the Ivy League. Princeton must now win Wednesday at Penn to force a one-game playoff for the automatic bid.

• Notre Dame

Notre Dame withstood a Connecticut run after Ben Hansbrough fouled out with 8:24 left. Barring a complete collapse in the Big East tourney, the Irish are in good shape for a #2 seed at worse.

• Alabama

Despite a 12-4 conference record and the SEC West crown, Alabama has one of the more hotly-debated at-large profiles. Their win over likely at-large Georgia won't hurt, but they still have work to do in Atlanta.

• Michigan

• Georgetown 17th on the CtD S-Curve (#5 seed)

The free fall continues for the Hoyas after point guard Chris Wright broke his hand on February 23. Georgetown has dropped four of their last five. All have been to tournament teams, but two were at home and yesterday's 22-point loss at Cincinnati won't impress the committee regardless of Wright's status. He expects to return for the NCAA tournament, and Georgetown will certainly let the committe know that.

• Purdue 6th on the CtD S-Curve (#2 seed)

A loss at Big Ten cellar-dwellar Iowa cost Purdue any hope they had at sneaking up to the #1 seed line.

• Missouri Valley - Indiana State vs. Missouri State

Despite inexplicably picking up a few votes in this week's coaches poll, Missouri State likely won't make the field without a win over Indiana State.

• Wisconsin at Ohio State

The Buckeyes are currently the #1 overall seed in CtD's eyes, and will be hard to knock off the #1 seed line if they pick up the win on senior day.

• Kentucky at Tennessee

Tennessee's travails have been well-documented this season. I'm not sure a home loss to Kentucky make them fall back on the bubble, but they don't have much room for error.

• Florida State at North Carolina State and Wake Forest at Boston College

The ACC's mediocrity (by its own standards, anyway) has also been well-documented. If Florida State and Boston College have any hope at an at-large bid, neither can afford to slip up.

Each day (more or less) over the next eight days before Selection Sunday we'll take a look back at the previous day and a look forward at the next.

• Atlantic Sun - North Florida vs. Belmont (Macon, Georgia)

Belmont's profile has some at-large characteristics, but they're not quite in as of todoay. A loss would certainly keep them out.

• Big South - UNC Asheville at Coastal Carolina

The Chanticleers have a chance for revenge at home after UNC Asheville's Matt Dickey hit a three pointer at the buzzer to give the Bulldogs a win ten days ago.

• Ivy League

This one has an asterisk. If Princeton wins, they clinch the regular season championsihp and the Ivy's automatic bid. If Harvard wins, they clinch at worst a tie for first place and put pressure on Princeton to win at Penn next week to force a one-game playoff.

• Duke at North Carolina

The outright regular season championship and first seed in the ACC Tournament are at stake in Chapel Hill. Duke seems likely to be placed in the Charlotte pod, but the Heels could make a claim for the other spot with a win today and strong showing next week.

• Michigan State at Michigan

Despite their troubles this year, the Spartans are closer to an at-large bid than are their in-state rivals. A win at home would bring the Wolverines even with Michigan State at .500 in the Big Ten. A loss probably means the NIT.

• Wyoming at BYU

BYU has precious few chances to show that Wednesday's 18 point home loss to New Mexico doesn't represent the new normal without Brandon Davies.

• Horizon semifinals (Valparaiso at UW Milwaukee, Cleveland State vs. Butler)

UW Milwaukee surprisingly claimed the top seed and home court in the Horizon tourney, but today's semi Cleveland State and Butler has the only two teams with realistic at-large chance. The loser is probably NIT-bound, and the winner may still have to claim the automatic bid on the road.

Each year I like to take a look at some of the teams that the selection and seeding models may be mishandling. I'll do a little more of that over the next week, but I'll start with a quickie side-by-side comparison of St. Mary's and Gonzaga.

(Background here)

The CTD selection model currently has St. Mary's as one of the last few that would be in if they we'rent already the automatic qualifier. Gonzaga, on the other hand, is the seventh team out.

It's not as though St. Mary's, like many of the teams at the tail end of the field, has a profile that historically matches at-large selections. The CTD model compares each team against profiles from the past and assigns each a score between 0 (not an at-large profile) and 100 (perfect at-large match). St Mary's score (through March 2 games) was 2.4 - not exactly a ringing endorsement.

St. Mary's best win is better than Gonzaga's, but after that Gonzaga has the better wins. St. Mary's inexplicable loss to San Diego will probably be its biggest problem with the committee.

Wins

Losses