Tuesday, March 18, 2014

Complete Midwest Region Analysis

Opening Thoughts:

This is obviously the Region Of Death. The top four seeds are ridiculous. But let me rant briefly about why these seeds are ridiculous. My regular readers and my twitter followers know that I've been talking about this for a long time, but I know that around Tournament time I get a lot of website visitors who are new to my site, and I want to explain this to you.

It's really not your fault. TV lies to you. If you watch any television analyst, or a major newspaper writer, or even people from the Selection Committee doing public interviews, they will tell you that they are picking "the 36 best teams", and that they want to seed teams by how good they are. Especially if you watch television you'll hear about "the eye test" all the time. How good is a team? How do they look? Surely we all saw how ridiculously good Louisville is playing right now! How can anybody think they're not one of the 12 best teams in the country!

Like I said, it's a lie. The "eye test"? Picking "the best teams"? It doesn't happen, and it's never happened. And in fact, that's for a very good reason, as you would hate what would happen to college basketball if it did (more on that in a moment). Teams are seeded by their resumes. Who did you beat? Who did you lose to? What's the difference between "best resumes" and "best teams"? Two things. First, schedule. If you schedule poorly in non-conference play, you won't pick up quality wins in non-conference no matter how well you play (lookin' at you, Louisville... and SMU). Second, luck in close games. The results of close games are almost entirely luck. The difference between a one point win and a one point loss is almost meaningless for any rating of how good a team is, but it matters a lot for your resume.

So witness this tweet from me about 2.5 hours before the bracket was to be revealed:

The Committee stuck to their guns and gave Louisville a 4 seed. Could you have argued for a 3 seed? Sure. A 3 or 4 seed. But nothing higher, even though Louisville is clearly one of the four best teams in the country right now.

But I know what Kentucky fans are saying (Calipari has been complaining to anybody that will listen about this): "We were 14th in RPI! If you're judging resumes, how are we an 8 seed!" Well, because RPI can be easily manipulated. The RPI is a crappy measure that can be goosed. Some coaches do this well (Calipari is one of the best) while others do this poorly. And if you want to goose your RPI, making 7 of your 13 non-conference games at home against mid-majors inside the RPI Top 200 that you're heavily favored against while only playing one outside the RPI Top 200 (which is what Kentucky did) is just about perfect. If you look at a better measure of resume, the Sagarin PURE_ELO, Kentucky's resume was only 33rd best. So an 8 seed was perfectly fair. In fact, anything higher than a 7 seed would have been the Selection Committee bowing to the name on the uniform. I'm glad they were able to judge Kentucky fairly.

By the way, why do we want teams seeded by their resumes rather than by how good they are? Two big reasons. First, scheduling suddenly wouldn't matter. Beating Murray State by 40 is better than beating Duke by 1, so why schedule Duke? Second, results in close games wouldn't matter. Whether a 30-foot buzzer beater goes in or not says really nothing about your team, and has no real impact on your computer ratings. But it matters a lot for your resume. Don't we want to care whether that buzzer beater goes in? Don't we want to reward teams for what they've actually achieved rather than on their future potential? So the current system makes sense, even if the Selection Committee lies in public about what they're doing.

Alright, rant over. Let's talk about the actual games:

First/Second Round:

My Picks:
16. Cal Poly over 16. Texas Southern
11. Tennessee over 11. Iowa
12. Xavier over 12. NC State

1. Wichita State over 16. Cal Poly
8. Kentucky over 9. Kansas State
4. Louisville over 13. Manhattan
12. Xavier over 5. Saint Louis
3. Duke over 14. Mercer
11. Tennessee over 6. UMass
7. Texas over 10. Arizona State
2. Michigan over 15. Wofford

I'm sure Mike Davis is getting a little satisfaction over making the NCAA Tournament in a year that the entire state of Indiana was shut out, but... come on... it's the SWAC. This Texas Southern team is one of the worst to ever make the NCAA Tournament. Of course, Texas Southern's defense against the Cal Poly offense is the movable force against the resistable object, so it's not like Cal Poly is an overwhelming favorite, but on paper they're the team to go with when making degenerate bets.

The two bubble team play-in games are much more interesting. This Iowa/Tennessee game is far and away the highest level play-in game we've ever had. According to the Pomeroy ratings, Tennessee is the 13th best team in the nation and Iowa is 27th. Sagarin actually has them reversed, with Iowa the 13th best team and Tennessee 20th. The Iowa rating is confusing for people who have seen them lose 6 of 7 down the stretch and finish only 20-12 overall, and have been told by every pundit on tv that they're "overrated". Of course, if everybody on tv calls you "overrated", you're basically by definition underrated, and Iowa is underrated due to horrific luck in close games (against Big Ten opponents they went 0-6 in games decided by six points or less). In Big Ten play their +0.07 PPP differential was better than Ohio State and Nebraska and not far behind conference champ Michigan (+0.10 PPP). They are a really good team. And same goes for Tennessee, by the way, with their 1-7 record in games decided by seven points or less and a really nice +0.13 PPP differential in SEC play (for comparison, Kentucky was only +0.09 PPP, and even Florida was only +0.19 PPP en route to going 18-0).

How do we separate Tennessee and Iowa? Well Tennessee has one of the goofier stats in the country for a team that doesn't take many threes: they are 4-10 when shooting below 31% behind the arc and 17-2 when shooting better than 31% behind the arc. Basically, they tend to be streaky, and they can win a game with crazy shooting (like they did against Virginia) or they can shoot themselves out of a game with terrible shooting (like they did when they shot 1-for-19 behind the arc in a big loss to Florida). Iowa's defense has struggled, though, particularly on the perimeter. They were below average in the Big Ten in 3PA/FGA defense and 3P% defense, and their eFG% defense was dead last in Big Ten play. Iowa attacks the offensive glass hard and likes to draw fouls, but Tennessee led the SEC in DR% and defensive FTRate. Either Iowa or Tennessee will be a solid favorite over UMass, but I like Tennessee to be the team to get that chance.

I think the NC State/Xavier game comes down to what you think about Matt Stainbrook's health. NC State's defense is bad (12th in the ACC) and their offense can't shoot (14th in 3P%, 10th in eFG%), so they need to dominate the glass to win. But with Stainbrook fully healthy, I don't see that happening. This NC State team really shouldn't be in this NCAA Tournament, no matter how you want to pick your bracket, and I think they go down to Xavier here.

Onto the Round of 64 games, obviously you don't need to worry about Wichita State or Michigan. Wofford is easily the weakest 15 seed in the NCAA Tournament and will not pose a threat to Michigan. Mercer over Duke is getting a little bit of hype, but I think we all see how silly that is. Mercer was 7th in the Atlantic Sun in defensive 3PA/FGA ratio, and Duke loves to chuck up threes. Meanwhile, Mercer's offense is heavily driven by threes, while Duke led the ACC in 3PA/FGA defense. And, by the way, who on Mercer is guarding Jabari Parker? I wouldn't worry about Manhattan vs Louisville either. George Beamon is a really good player, but Russ Smith might be the scariest defender in the nation to go against. Louisville is vulnerable against big front line that can get after the glass, but Manhattan is a small team that was barely above average at rebounding the ball in the MAAC.

The first interesting game is the 5/12 match-up, and it's probably the most interesting 5/12 match-up in the whole Tournament. Whether you have NC State or Xavier here presumably doesn't matter in your bracket competition, and Saint Louis is definitely a soft 5 seed. It's not just that Saint Louis has lost four of five (which is due in large part to luck) but that they've been overrated all season long. They simply haven't been as good as their resume, and are actually only 42nd in the Sagarin PREDICTOR. On paper, Xavier and Saint Louis are almost exactly even. Saint Louis has a really good defense and they led the A-10 in 2P% defense, but they are vulnerable to big men who can get the ball inside their tough perimeter defense. And while Xavier's defense isn't very good, Saint Louis's offense is bad, finishing 11th in the A-10 and failing to crack 1.00 PPP in four of their last five games. Xavier has lost one game in the calendar year 2014 by double digits, so this game will be close. In the end, I think this is where we finally have that 12/5 upset.

NC State is a bit harder of a pick over Saint Louis, just because they're not really very good. But if you believe in the "last team in the Field of 68 always does well because they're highly motivated by the hate" logic, NC State can potentially score just enough points off of offensive rebounds against an occasionally shaky Saint Louis defensive rebounding squad to pull off the upset.

The Iowa/Tennessee game versus UMass is fascinating because whichever 11 seed wins that play-in game will be a clear Vegas favorite over UMass. While Louisville is a great example of a team clearly better than their seed, UMass is a great example of a team clearly worse than their seed. UMass was 11-4 in games decided by six points or less, and by both Pomeroy and Sagarin are worse than several teams that were relegated to the NIT. What does UMass have going for them? Chaz Williams, of course, one of the most explosive playmakers in the nation. But other than Cady Lalanne's shot blocking, they aren't particularly good at anything else. UMass was also 8th in the Atlantic Ten in defensive rebounding, facing Iowa and Tennessee teams that both really get after the offensive glass. The reality is that whoever you have winning the Iowa/Tennessee game (and in most bracket competitions, they won't even ask you to pick one) should be favored over UMass.

Arizona State and Texas are two somewhat overrated teams playing in a 7/10 game. On paper these are two relatively even teams. The difference, in my opinion, is in the paint. Arizona State has a fearsome defensive presence in 7'2" Jordan Bachynski, but he's their only true big man, and Arizona State was near the bottom of the Pac-12 in both offensive and defensive rebounding. And that's a problem against a Texas team with multiple quality bids (Cameron Ridley, Jonathan Holmes and Connor Lammert) who all go aggressively after the offensive glass. Texas was the best offensive rebounding team in the Big 12, and it was what drove their success this season. Unless Jahii Carson dominates Isaiah Taylor, Texas is the favorite.

It's weird to think of Kentucky as underrated, but they are underrated. They certainly deserved the 8 seed (stop whining, Calipari), but they're a lot better than their resume. A 2-8 record in games decided by six points or less will do that. And Kansas State, while a solid defensive squad, is pretty mediocre and limited offensively (only 8th best in the Big 12), and their only road victories all season long came over Texas Tech and TCU. Kansas State's offense has had particularly trouble away from home. They failed to hit 1.00 PPP five times in conference play, and all five came on the road. I just don't see an argument for taking them over Kentucky.

Third Round:

My Picks:
1. Wichita State over 8. Kentucky
4. Louisville over 12. Xavier
3. Duke over 11. Tennessee
2. Michigan over 7. Texas

Kentucky over Wichita State is going to be a popular upset pick, and certainly I understand it. Kentucky is a ridiculously good 8 seed and Wichita State is a very soft 1 seed (though I'll repeat myself again: both teams deserved the seeds they got based off of resume). I'd expect Wichita State to be favored in Vegas over Kentucky, but only by a couple of points. This is a close game. But I'm going with Wichita State for two reasons. First, this Kentucky defense is good at stopping three-pointers, but they are not good at help defense and have trouble keeping opponents out of the paint, while Wichita State derives a large portion of their offense in the paint (they're actually a pretty mediocre three-point shooting team). Also, Kentucky depends on crashing the offensive glass for points, and were second in the nation in OR%, but Wichita State was 5th in the nation in DR%, so Kentucky's going to need to find other ways to score (which they're not good at). I understand why a lot of people are taking Kentucky, but in my opinion Wichita State is the favorite.

Louisville is going to be a fairly large favorite against either Xavier or Saint Louis. Of the two, Xavier is probably the tougher opponent because Semaj Christon and Isaiah Philmore can both get to the free throw line and the team can get after the glass, the two weaknesses of Louisville, Saint Louis is not a team that gets any easy baskets. They'll struggle to crack 60 points against the Cardinals. Either way, Louisville is a pretty easy pick to take to the Sweet 16.

One of the things I often talk about is that the difference between being a 1 seed and 2 seed and 3 seed really isn't that big, on average. You can be a 1 seed and get screwed in your (draw like Wichita State) or be a 3 seed like Duke and get a really generous draw. I don't think there's any question that Duke has the easier path to the Final Four than Wichita State, and it's not even really close. And this Round of 32 games against UMass or a play-in team is a big reason why. Tennessee or Iowa would obviously be tougher tests for Duke than UMass, but the Blue Devils will be a pretty solid favorite against anybody.

I actually think Iowa would be the tougher opponent for Duke than Tennessee, because of the aggressive way they get after the glass and draw fouls. If they can get Jabari Parker in foul trouble then they can really just play volleyball off of their missed jumpers (and boy does Iowa like to shoot jumpers). Tennessee has the better post scoring options, so they would be difficult for Duke in their own way. How does Duke match up against Jarnell Stokes? How do they keep Jeronne Maymon off the glass? Either game will be tough. What are the worries these two teams would have? Well, Iowa's poor three-point defense is a problem if Duke gets going. This could be one of those games where Andre Dawkins comes off the bench and hits six three-pointers. With Tennessee, I think the question is if they have the offensive skill to take advantage of Duke's defensive liabilities. Duke's perimeter defense is still feisty, even if they lack a real paint presence, and opponents often struggle to get the ball into the paint. In the end, Duke is the favorite against either team, but neither will be an easy opponent.

If you have UMass pulling the upset over the 11 seed, I wouldn't really consider taking them over Duke. Duke has a couple of quality perimeter defenders that can take a shot at Chaz Williams, and the UMass perimeter defense is pretty poor. The skill difference between these two teams is pretty vast, and Duke would be something like a 7 or 8 point favorite in Vegas. It could be a romp.

Michigan is a relatively soft 2 seed. They won the Big Ten regular season title, but primarily by being lucky in close games (8-0 in Big Ten games decided by seven points or less). Their efficiency margin in conference play was tied with Wisconsin for second best, and just narrowly ahead of fourth place Iowa. That said, they got a lucky draw to the Sweet 16 (leaving aside Duke waiting for them in the Sweet 16, of course). Neither Texas or Arizona State is particularly good, Also, Michigan's achilles heel is their defense, but both Texas and Arizona State were below average offenses in their conferences. Texas is probably the bigger concern of the two because of their big bodies. Without Mitch McGary, Michigan has struggled at times against big front lines, and Texas certainly has that. And so they'd have a shot to pull the upset for that reason. But before you pick Texas, just think about what will happen if Texas doesn't totally annihilate Michigan on the glass. Michigan is just so much more skilled offensively. And I don't really see an argument for Arizona State anywhere.

Sweet 16:

My Picks:
4. Louisville over 1. Wichita State
3. Duke over 2. Michigan

If Wichita State and Louisville play, Louisville will likely be something like a 5 point favorite in Vegas. They've been the better team this season, and the better team down the stretch, and I don't think there's really much debate about that (at least among those that understand advanced metrics). But how do these two teams actually match up against each other? Well, Wichita State is good across the board... but they're not really great at anything. That seems weird because Pomeroy rates both their offense and defense in the top ten in the country, but they're rated highly because they're very good at everything. They rebound well, they get open shots pretty well, they take care of the ball and force turnovers pretty well, etc. But what are they best at? It's not quite clear, and part of that problem comes from the fact that their schedule was pretty soft. We haven't gotten to see them play teams that are great at something to see how Wichita State's roster responds.

We know what Louisville can do, which is to play pressure defense and force turnovers. Fred Van Vleet is very smooth with the ball, and it's hard to see him melting down against anybody, but Wichita State as a whole wasn't the greatest ball handling team (just 4th in the Valley in offensive turnover rate, though they did lead the lead in steals allowed per game). What about Louisville's weakness, which is big men that can crash the glass and draw fouls? Cleanthony Early doesn't really get a lot of offensive rebounds... I guess Darius Carter is the best shot there.

I think you have to ask yourself what you think of this Wichita State team's success. Obviously going undefeated for 34 (and presumably at least 35... or by this round it would be 36) games is extraordinarily impressive, even against that schedule. But while they've been really dominant over the last few weeks, that wasn't totally the case earlier. They were good-but-not-great in non-conference play, squeezing by teams like Davidson, and they had a bunch of close calls early in conference play. They're on a real hot streak, but history says that hot/cold streaks don't carry over into the NCAA Tournament. Besides, Louisville is on their own hot streak, and if they were in a different conference and had scheduled better they easily could be sitting here as a 1 seed. Throw in the fact that Louisville as the 4 seed has an easier opponent in the Round of 32 than 1 seeded Wichita State (thanks, Selection Committee!) and I think the only logical choice here is Louisville over Wichita State.

By the way, what if you have Kentucky over Wichita State? Well, I'm sure it will be pointed out that Kentucky beat Louisville earlier this season. But I watched that entire game, and it's really not a good game to use as a judge. Louisville was struggling to find their rotation early in the season, and weren't playing at anywhere near their current levels, and they also couldn't shoot to save their lives (6-for-26 behind the arc). Kentucky's backcourt is not playing particularly well, and Russ Smith and Chris Jones will run all over them. That said, the one answer Kentucky has is Julius Randle, and he's where Louisville is really missing a guy like Gorgui Dieng. I'm not sure Montrezl Harrell can limit Randle by himself.

If Duke plays Michigan in the Sweet 16 (and certainly that seems very likely), expect to see Duke a narrow favorite in Vegas. Now maybe Duke will get an extra point because they're a "public" team, but the computers agree that Duke was a little bit better than Michigan this year. That said, these teams are ironically very similar in strengths and weaknesses. Both of these teams are incredibly explosive offenses but extremely mediocre defenses, and history says that teams as poor defensively as Duke and Michigan almost never make a Final Four. By the advanced metrics, the last team with a defense even close to as bad as this year's Duke and Michigan teams that made a Final Four was Dwyane Wade's 2003 Marquette squad (that team was still ranked narrowly outside the Pomeroy Top 100 defensively, which is where Duke and Michigan are now, after their Final Four run). Both Duke and Michigan are thin on the front lines and are vulnerable against teams that can crash the glass. Both are really good jump shooting teams, though very dependent on that jump shooting. Basically, these two teams are awfully similar to each other.

I think the one big difference between these two teams is Jabari Parker. He's the best player on either team, and he has the ability to force constant double teams from Michigan's defense. You can make a really good case that after Doug McDermott the best player in college basketball is Jabari Parker (I'd take Russ Smith, but you can make the case for Jabari). Parker was limited to 12 points when these two teams played in December, but Mitch McGary was playing that day. Michigan is always going to be fairly dependent on hitting outside shots, as even Nik Stauskas sometimes gets tentative and a little bit too comfortable jacking up threes. So Parker is the reason I give Duke the edge.

Elite 8:

My Pick:
4. Louisville over 3. Duke

By some computer ratings, Wichita State is better than Duke, but by other computers ratings it's the other way around. And certainly every computer rating I'm aware of agrees that Louisville is the best team in this region (and I think pretty much every analyst agrees as well). So Duke vs Louisville, while a 3 seed vs a 4 seed, is probably the match-up between the two best teams in the region. And so it's really not much of an upset. A whole lot of your friends will have this match-up also.

To me, though, I really can't take Duke (or Michigan, for that matter) to the Final Four. And it's for the reasons I stated above. History says that you want to be Top 20 in Pomeroy offense and defense to make the Final Four - it doesn't fit all teams, but it fits most teams. You can get away with being outside the Top 20 on one side of the ball if you're really elite on the other side of the ball, but there's only so far you can push that. Like I said above, the last team to have a defense as bad as Duke or Michigan and make a Final Four was Dwyane Wade's Marquette team 11 seasons ago. That's a pretty daunting precedent, particularly with a really strong Louisville team standing in their way.

And if you look at the match-ups, Duke and Michigan don't even match up particularly well here. Louisville is vulnerable against teams with multiple bigs who can get offensive rebounds, score second chance points, and draw fouls. Duke has a guy who can do that in Jabari Parker, but that's it, and overall they're not a great offensive rebounding team. Michigan doesn't have anybody who fits those characteristics. Michigan was 7th in the nation with only 9.4 turnover per game, so they can generally handle aggressive defenses like Louisville, but Louisville can score just fine without turnovers. They will be aggressive off of defensive rebounds and even off of made shots, and Michigan's transition defense was been pretty terrible this season (opponents have a 62% eFG% in transition compared to 46% otherwise). To me, Louisville is a pretty solid favorite over either of these teams.

If you have Wichita State getting past Louisville, I think they'd be favored over Duke and Michigan for the same reasons. You just need a really good reason to take a team that bad defensively to the Final Four. Where do Duke or Michigan have a significant match-up advantage? Even Jabari Parker doesn't even fit here, as Wichita State has several solid defensive big men that Gregg Marshall can throw at him.

Now if you somehow have Kentucky getting through here... I think you might want to take Duke or Michigan simply on the "strength of schedule" argument. Kentucky has to get through Kansas State, Wichita State and Louisville just to reach an Elite 8, while Duke and Michigan both have a really soft path until they play each other in the Sweet 16. Unless you're a huge Kentucky fan, do you really want to rely on this disappointing Kentucky team to pull it together for four games in a row like this? Particularly with how spotty their backcourt play and defense have been? I wouldn't want to take that chance in my bracket, for sure.


Anonymous said...

Nice "analysis", dude. I read the entire post twice and still have no clue which team is guttiest. Fail.

Anonymous said...

Is the stat about "no defense this bad since Marquette" calculated based on D ratings going into the tourney or after the tourney is completed. In other words are there teams since Marquette that had D ratings close to Michigan or Duke going into the tourney but by the time they made a final four run the ratings had improved.

Jeff said...

I'm basing it on post-tourney stats, because I don't have access to pre-tourney Pomeroy D stats, but there's a pretty big gap back to those three teams. I think the next worst defense since '03 to make a Final Four finished with a Pomeroy defense around 50th.... somewhere around there.

Anonymous said...

Pre-tournament. Post-tournament. Doesn't matter. I think Jeff's point is that both Duke and Michigan are business class because their defenses are flying coach. Therefore Louisville should be favored versus either as they would have priority boarding.

Scott said...

Question re: log5

Does kenpom recalculate the probabilities after the first round (aka play-in) games? Is there a shortcut for doing this?

Jeff said...

Log5 math is very easy to do yourself. You can google the formula. As for how much the Pomeroy ratings will change after each game... that isn't so easy. You probably can't do that yourself. But Ken usually updates his numbers within an hour or so of each game ending.

Unknown said...
This comment has been removed by the author.
Unknown said...

I think the assertion that Wichita State is soft is interesting. You point to advanced stats, but they are very comparable (and arguably more balanced) than the other #1 seeds (ken pom, colley matrix). If you're only considering Sagarin, then yes, they are 'soft'. They didn't play crap for a schedule, but what we do know is they've returned 7 of 9 from last years final 4 team, and one of the lost players (PG Armstead 2012/13 stats: fg% = 39%, A/T < 1.7/1) was arguably one of their weakest links. He wasn't nearly as efficient as Van Fleet (2013/14 stats: fg% = 49%, A/T = 3.8/1).

I think it's pretty clear that this Wichita State team is better than last years team, and I also think that this Louisville team is about the same or slightly worse than last years team.

Louisville won by 4 last year (Wichita State led by 12 in the 2nd half), more Van Fleet and no Armstead makes this potential matchup very intriguing. I don't know if Louisville can force enough turnovers with Van Fleet being the major decision maker for the Shockers. I think there's a good chance that the winner of that game will end up in the Championship game.

The potential Louisville/Wichita State match-up and the Michigan State/Iowa State match-up are the most intriguing in the tournament IMO. If Izzo decides to play uptempo with the Cyclones, they might get themselves into trouble.

Anonymous said...

Where do you get the B10 conference efficiency numbers? You said Michigan was tied with Wisconsin and just above Iowa. According to Kenpom (http://kenpom.com/conf.php?c=B10), Michigan was first in margin with 10.7, Wisconsin's margin was 7.9, and Iowa's was 6.8.

Jeff said...

I generally take conference efficiency numbers from bbstate.com. Those two websites calculate possessions slightly differently, which is why they have slightly different numbers.

Unknown said...

Van Fleet would be such a better name than VanVleet, not sure why I thought it was the former. This difference alone my force me to take Louisville over Wichita State.

Anonymous said...

What are your thoughts if I have the winner of tenn/Iowa beating duke and facing michigan? I needed a high seed in my sweet 16 and am picking them based on them being a seemingly stronger high seed than others..

Jeff said...

That's going to come down to just how much your bracket competition rewards upsets. In general I advise people who pick big upsets to have that team lose in the next round, because you don't want to totally screw up your bracket if the upset you picked doesn't come to pass.

But if your bracket competition gives vastly more credit for upsets than for favorites, obviously the actual gap in quality between Michigan and Tennessee/Iowa is pretty small. Michigan would be a very small favorite in Vegas.