Final 4 Deep Dive
Game day is finally here. It was way too long of a week waiting for these games to finally come.The field of 64 teams has been cut down to only 4 remaining teams (in alphabetical order): Auburn, Michigan State, Texas Tech, and Virginia. Let’s break down each matchup and uncover the path to winning it all for each team.
(Like always, the code for our analysis is posted on our github giving you the ability to recreate any of the visuals)
Ken Pom Rankings
KenPom is an invaluable resource for college basketball fans and we highly recommend his work. He provides a solid path into a more advanced analytical mindset.
(darker = worse, lighter = better —– except for pace, darker = faster, lighter = slower)
While the Sweet 16 was mostly chalk, the weekend that followed was assuredly not. Three number 1 seeds went down, leaving only Virginia—coming off a historic loss to a 16 seed last year—as the sole remaining top seed. As we discussed last post, all four remaining teams excel defensively, highlighted by first, third, eighth, and forty-fourth ranked defensive efficiencies.
However, the defensive efficiency metric does not always do a great job showing how impactful these teams are on the defensive end. Take Auburn, for example. They are by far the lowest ranked defensive team remaining, but are first in total steals in the tournament (36), eighth in total turnovers caused (37), and fourth in blocks (18). Yes, they have let up 70+ points in each of their tournament games thus far. But their “active chaos” has led to a significant advantage in offensive opportunities created, perching Auburn atop the tournament in points per game (85.3). This frenetic “multiple” defensive setup—Auburn likes to mix in half-court and full-court presses, traps, and light pressure, as well as man-to-man and matchup zones—has helped them overcome the loss of their star Chuma Okeke, who missed the Kentucky game after tearing his ACL late in the game against UNC.
While Auburn is the most surprising team remaining in the tournament according to public perception, Virginia remains the biggest anomaly in college basketball. When creating a predictive model, you have to understand that the model’s goal is to predict results for every single team over every single season optimizing on the log loss function (I won’t get too technical here but most people may be familiar with the term “regression to the mean”).
What is the most likely outcome based on all of the possible scenarios that has ever been observed in the past. If team A played team B 10,000 times, what are the factors that could change from one game to the next to the next? And of those factors, which are sustainable, repeatable, and trackable? While Virginia ranks highly in many metrics across the board, their unique style, reduced possessions, and slow pace give our models pause.
Virginia’s style is extremely effective, and, to the eye, extremely successful year after year. But our model tends to “misprice” them in a tournament setting. Even with all of Virginia’s successes they are largely seen as an outlier team, winning with defensive, intentional offensive execution, and timely shoots. From a modeling perspective, this introduces significant variability game to game. In other words, in a tournament format, a team like Virginia that has to play mistake free to win (even if they consistently do) will be undervalued due to their small margin of error.
A perfect example of why the model has lower predicted probabilities on teams with small margins of errors is Virginia’s loss to a 16 seed last year. “On paper” they did not belong on the same court. Virginia played their style without any adjustments but UMBC continued hitting shots winning by 20 points making history as the first 16 seed to knock off a number 1 seed.
Now for the matchups:
Virginia – Auburn
Offense
Defense
A hot team from three against Virginia’s top ranked three-point defense. Haven’t we seen this before? In what is becoming a stunningly bizarre stretch of games for Virginia, they can’t seem to escape the hot team from down town. Consider this: in every game this tournament, they have faced a team in the top half of the nation in three point percentage, including four teams in the top-100 for total three pointers made this season (including Auburn-1 and Purdue-17 in the top 20).
Generally, a great offense struggles against a great defense, but a weird trend has followed Virginia throughout this tournament… they have been extremely susceptible from behind the arc. In every game so far this tournament, Virginia has allowed each team to not only eclipse their average three-point percentage and total threes per game average, but also eclipse Virginia’s own three-point against average and total threes per game (27.3% and 6 a game).
Virginia allowed nine threes against Gardner-Webb at a 39.1% rate (GW averaged 8 threes a game at a 39.1% rate for the season—spooky); eight threes against Oklahoma at a 36.4% rate (Oklahoma averaged 7 a game at a 34.6% rate); 9 threes against Oregon at a 36% rate (Oregon averaged 8 a game at a 35.1% rate); and 14 threes against Purdue at a 43.8% rate (Purdue averaged 10 a game at 36.8%).
Unfortunately for the Hoos, Auburn comes in as the hottest three-point shooting team in the country—and trust me, Auburn loves shooting threes, ranking second in the country with 1083 attempts. After a stretch of eight straight games with twelve or more made threes, Auburn only had seven against Kentucky, but almost all of them were timely and sparked a big second half run to close out the game. Even when Auburn misses, they keep shooting, and as a plus offensive rebounding team, they may be able to take advantage of second-chance opportunities against Virginia.
A common misconception of Virginia’s “pack line” defense is that it gives the other team the open three in exchange for a clogged lane. Instead of focusing only on stopping dribble penetration, Virginia’s incredible “switchability” and closeout ability (specifically help defenders from the weak side to the strong side) allows them to both clog the paint and get a hand up to stop the easy shot from outside.
6’7” wing De’Andre Hunter is a huge reason Virginia is able to get to the open shooter so quickly. He can guard 1-4 with ease and his ability to switch and stifle opposing team’s top options allows the active hands of Mamadi Diakite and Ty Jerome to create a nightmare for teams trying to find the open shot. It’s commonly known that Virginia wants you to attack their half-court defense. This is where they excel, but Auburn’s style could help neutralize this strength.
Auburn’s own defensive pressure and ability to go coast-to-coast with ease could force Virginia into an uncomfortable up-and-down game, especially if they get behind early. Jared Harper may be the fastest point guard in the nation and Auburn loves to take a rebound and sprint to the basket for an easy layup. If they are able to force Virginia into turnovers—something very rare, but has happened against other quick teams (see Gardner-Webb)—Auburn should be able to take advantage of three-pointers in transition and layups.
Auburn has seen quite a few different styles in this tournament, including teams on either ends of the pace extremes (UNC-seventh fastest team in the nation and Virginia-slowest team in the nation). This game will be a pace tug-of-war and one that looks to be decided at the three-point line. I’m especially excited to see Bryce Brown’s matchup with 42%+ three-point shooter Kyle Guy. Guy was mired in a miserable slump prior to finding his shot against Purdue.
Bryce Brown is now Auburn’s go-to scorer without Okeke and put in 24 points against Kentucky. I wonder if Virginia will deploy Hunter on Brown all game or if they have enough trust in their scheme to switch on Brown at the top of the key. Our model doesn’t like Virginia—they should have lost to Oregon according to the model, but my eyes have trouble seeing anything other than a Virginia victory. This should be fun.
Model: Auburn 63%
Monte Carlo: Auburn 63%
Michigan State – Texas Tech
Offense
Defense
Stout defense. Efficient offense. Sound familiar? While Michigan State is closer in style and form to Michigan than Gonzaga, they are certainly no offensive slouch, ranking number three overall in offensive efficiency. This presents many similar offensive problems to Texas Tech.
While both teams are stout defensively, Texas Tech is on another level, considered a top two defensive team in the nation all year. Texas Tech plays an extremely aggressive and active man-to-man defense that is concerned with cutting passing lanes and bodying cutters. This aggressive style of defense generally results in many committed fouls—Tech was 270th in the nation in fewest fouls committed. But strangely enough in this tournament they have done a great job avoiding foul trouble, especially in the second halves of games. Playing defense without fouling is an area to watch in this matchup as Michigan State excels at the line (in the regular season they were just outside of the top 50 in free throws attempted and ranked 22nd in free throw percentage).
While they haven’t been quite as aggressive in this tournament, Michigan State’s ability to get “cheap” points from the line could completely derail the defensive game plan for Texas Tech. Interestingly, though, if Sparty can’t get to the line, their proficient offensive attack that relies on assists—they were tops in the nation in that category—can actually play to Tech’s favor. Gonzaga, owners of the third most assists in the nation, committed 16 turnovers (next to 15 assists) against Tech in the elite 8, including a number of drops caused by active hands and the unmatched length on the wings. Tech’s ability to fake open passing lines could be a big issue for a team that loves to share the ball as much as Michigan State does. The Spartans will have to find ways to keep plays alive and avoid easy run-out opportunities that Tech uses to power its sometimes plodding offense.
Luckily, Michigan State is an extremely aggressive and proficient offensive rebounding team—in fact, they are top five in the tournament in all rebounding metrics—and will look to use offensive rebounds to increase the number of possessions in this game. Something interesting to note: Tech is an excellent offensive and defensive rebounding team but has been out rebounded in two straight games, something MSU should look to take advantage of.
While pace is variable for both teams, they have both shown a penchant to use their active defensive positions to lead to easy offensive opportunities. Look for Texas Tech to do this with steals and for Michigan State to take advantage of their extremely high block rate to spark easy transitioning opportunities. Michigan State ranks number two nationally in two-point field goal defensive, mainly due to their “pack the paint” philosophy. So if Tech wants to score, they need to run or hit on their threes at a rate they did against Gonzaga.
I’m intrigued by Cassius Winston’s matchup with Tech’s Davide Moretti. Winston, considered by most to be the best point guard in the nation (tournament-form Jared Harper on line 2) is the main (read: only) ball-handler for Michigan State, and while Tech doesn’t try to stop penetration as much as they try to clog the passing lanes, Moretti has the ability to stay in front of Winston and force the ball either out of his hand or into a quick shot. Jarrett Culver will have to be a star for Tech and Matt Mooney’s creativity will be greatly needed against a front-line that includes Xaiver Tillman and Kenny Goins.
This game, as most deep in tournaments, seems to favor the team that can make tough shots, especially tough threes. Tech has shown an ability to do that over the last few rounds, but the sample size is extremely small compared to their decent averages prior to the tournament. Michigan State has shown the ability to consistently make threes and if they are able to keep calm and avoid unnecessary turnovers, they should be able to make enough to hold off Tech.
Model: Michigan State 64%
Monte Carlo: Michigan State 64%
Path to the title:
Net Ratings
Net Rating is a metric encapsulating a team’s offensive and defensive ratings. When looking at the distribution curves for each team, a wider curve, like Texas Tech’s curve above, means they have much higher variation from game to game. They could play amazingly one night and terribly the next. A tighter curve, like Michigan State, means they are more consistent.
Offense
What each team needs to do to win a National Championship
Auburn:
It would be easy to call Auburn a case study in situational success. Losing a star player this deep into a tournament is obviously a tough blow, but injuries can sometimes galvanize a group and turn mediocre players into primetime performers. However, to pin success on Auburn solely on heart or togetherness would be completely missing the larger picture of Auburn’s season… they are simply really good.
Their three-point prowess is unmatched across the country, and their speed is almost unheard of within the college game. With a multiple defensive set that can guard you full-court, they have ways to change the timing of an offensive by only moving around one or two defensive pieces. Plus, when they are moving on defense, their offense seems to flow effortlessly, players cutting from all directions, offensive rebounds there for the taking. While losing Okeke hurts from a pure scoring standpoint, their defensive activity is more than enough to create easy offensive runouts.
Auburn will win if… They make threes and win the fast break point battle. Auburn’s goals are simple: They want to force turnovers and turn those turnovers into transition points, usually in the form of layups or three-pointers. If they can get back to twelve-plus made threes a game for the final two games, they will be awfully hard to beat, especially because teams struggle coming back against this group when they get hot. If their shots aren’t falling they need to rely on getting out in open space as their half-court offense tends to sputter when their threes aren’t falling. It’s not a bad thing for Auburn to fall behind early, but they have gotten blown out on numerous occasions this year by hoping the three can get them back in it. Auburn has a big hill in front of them, but they have the game’s biggest weapon in their back pocket.
Michigan State:
Tom Izzo in March is as good as they come. But recently, Tom Izzo at the end of March and in April has been less than ideal. Michigan State’s last trip to the finals was in 2009 and their last championship was in 2000. While this Michigan State team may not be the most talented team he’s had in the last few years due to injuries and graduation, it could be his most complete. With an All-American Point Guard leading the show, Izzo has made sure to make Michigan State the model of offensive efficiency. They share the basketball with ease and have so many options on the wings that they can lineup switch you to victory. For a team that has had such an up-and-down year as this, it must be extremely validating for Izzo to know that they are playing their best ball at the most important time.
Michigan State will win if… They win the individual scoring battles. Teams know that Cassius Winston is MSU’s main (only) ball handler. They know the offense flows through assists. So teams try to force the ball out of Winston’s hands. If Michigan State is able to exploit individual one-on-one matchups on the wings to stretch the defense out, the lanes should be wide open for their brilliant offensive movement. Add this to their very underrated transition game and you have a team that is hard to stop once it gets rolling. Because Michigan State is so proficient defending the two point shot, they should be able to take advantage of long rebounds from three-point range to start their offense and get into the flow prior to the defense being set. While individually they aren’t as strong as some of MSU’s previous players, collectively they are incredibly efficient and a number of different players can score when called upon.
Texas Tech:
Villanova. UNC. Villanova. Duke. UConn. The last five national champions were incredibly skilled on the offensive end. They were not defensive slouches by any means, but their focus and skillsets led to offensive output few other teams could match. Texas Tech is looking to flip the script and win on their broad defensive shoulders. To do so would be a remarkable feat for AP National Coach of the year Chris Beard. A team made up of two transfers in the starting lineup, one international player, one defensive-only center, and a rising NBA talent has exceeded all expectations and one-upped their improbable elite eight run from last year. And they are doing it with team defense. No single player really stands out defensively—Culver’s motor has shown his defensive upside at the next level, but their system, active hands, and chaotic switches cause nightmares for opposing teams.
Texas Tech will win if… they take more shots than their opponents. The beauty of Tech’s defense is they can get away with being okay offensively. Their defense forces opponents into long stretches of not scoring and their length causes easy runouts in transition. They have shown a penchant for offensive rebounding in this tournament and have actually been a plus three-point shooting team over their last few games. But it all comes down to opportunities. They must play solid defense and have more chances to score, because out of the teams left, they are the weakest on that side of the ball.
Virginia:
While models may disagree about Virginia’s National Championship validity, one thing’s for certain: Virginia is Virginia, and their gameplan is as obvious as a five-year-old’s hide-and-seek hiding spot (the closet, it’s always the closet). However, obvious doesn’t mean bad. In fact, obvious can help a team focus on maximizing their potential within an obvious scheme, something Virginia does better than anyone in the country.
Virginia will win if… they go back to their elite 3-point defense and limit their opponent to five or less a game. Without the three-point barrages, Virginia wins their previous matchups fairly comfortably due to their molasses pace and need for opposing teams to be perfect offensively with reduced possessions. But if the threes continue to fall for their opponent, Virginia has players to bail them out offensively. This team will go as far as the shooting arms of Kyle Guy and Ty Jerome and the offensive prowess of De’Andre Hunter takes them. Limit threes. Make threes. Slow the game down. Virginia cuts down the nets.
Championship Probabilities:
Michigan State: 40%
Auburn: 29%
Texas Tech: 17%
Virginia: 14%
*A quick disclaimer on KenPom’s “luck” index. Teams with 1 or 2 very good to elite free throw shooters tend to be considered lucky, particuarly solid but not great teams. That is partially because those teams fare well in close games. However, anyone who watches enough basketball knows that more often then not, when you hit your free throws, you win the game. There are certainly some valid aspects to the computation but we would be remiss not to mention its shortcomings.
