Tag: brackets

How stats and history would pick the 2022 NCAA Tournament

by Will Warren, posted in Basketball Statistics

A year ago today, I published what became one of the most popular posts on this website, about how 20+ years of data accumulated from KenPom and Bart Torvik could tell you what might happen in March. Whether or not it was useful in any real way is frankly up to the reader, I guess. It got two of the four Final Four picks right, including the #1 value pick of the entire NCAA Tournament, Houston. However, it missed on five Elite Eight teams and seven Sweet Sixteen sides. However, it did go 27-5 in the first round, the best record I have ever posted in a bracket I’ve submitted to a bracket pool.

I’m going to be frank. The level of care I have for submitting brackets at this point is pretty low; I am doing this more because a lot of people really like it than because I personally desired to write this. But: there is some sort of enjoyment in sharing a relatively unique perspective of current statistics and previous history to attempt to inform your bracket.

For whenever this gets picked up by people who don’t normally read this website, many of these picks will be wrong. Even the very best brackets miss on an average of 13-15 picks out of 63 total a year. If I missed on 15 total picks, I would be beyond thrilled. I missed 22 last year; maybe that can get below 20 this year. Who knows. I look forward to seeing how it all unfolds.

As a reminder, here’s how all of this works: What this is is simply a game-by-game projection of the field of 68 based on a document I’ve put together since 2018. Bart Torvik has an amazing page on his site with detailed historical KenPom projections of each game over the last 20 years of postseason play. Using that, I’ve accumulated enough data to make informed, quality guesses on how the NCAA Tournament may go.

Along with that, this year’s projections will factor in heavily to provide a baseline of measurements. KenPom’s numbers expect about 5.04 Round of 64 upsets, the highest number I’ve seen in a projection in a few years; Torvik sits at 4.75, a little lower, but still in that 5 upset range. I’ll share the five upsets these numbers would most incline you to pick. Author’s note: many stats were also brought in from this amazing guide to the NCAAT I found on Reddit.

Onward? Onward.

Round of 64

West Region

(1) Gonzaga over (16) Georgia State. Well, obviously.

(9) Memphis over (8) Boise State. KenPom actually has Boise as a tiny favorite, but Torvik and EvanMiya have Memphis as a favorite. Either pick is fine because I don’t think either team is beating Gonzaga (more later), but the favorite in these 8/9 games over the last 21 years is 58-26.

(5) Connecticut over (12) New Mexico State. The metrics average here has UConn at about 71.5% to win; 5 seeds at 70% or better are 28-6.

(13) Vermont over (4) Arkansas. This is probably wild if you’re an SEC fan, but, again, riding the metrics. 4 seeds at 70% or worse to win since 2000 are 7-10. This is also a play against Arkansas going deep; teams at 40% or worse to make the Sweet Sixteen (Arkansas is at 38%) are 2 for 29 in doing so. Catamounts!

(11) Notre Dame over (6) Alabama…or (6) Alabama over (11) Rutgers. Precisely what you needed: a game that you have to wait until Wednesday at midnight to pick. Notre Dame would be at 42.1% to beat Alabama if they played; 11 seeds at 37% or better are 29-21 since 2000 in winning. Rutgers, however, would be at 34%, which is below that 37% threshold. 11 seeds below 37%: 7-27. I don’t know that I really like picking either, frankly, but again, a situation where either winner would be out in my Round of 32.

(3) Texas Tech over (14) Montana State. At 91.1%, Texas Tech is merely one of the eight largest 3-seed favorites since 2000. No 3 seed at 85% or better has lost their Round of 64 game, a perfect 36-0.

(7) Michigan State over (10) Davidson. Utterly disgusting. But: Michigan State, in the metrics average, is at 52.2% to win. The favorite in 7/10 games is 65-19.

(2) Duke over (15) Cal State Fullerton. The general threshold for “oh?” 2/15 games is about a 10% chance of winning for the 15 seed. Fullerton is at 7.6%. I’d love to see it, but.

South Region

(1) Arizona over (16) First Four Winner. No need to elaborate.

(9) TCU over (8) Seton Hall. Either pick is fine here. TCU is favored by Torvik; Seton Hall by KenPom. Tiebreaker goes to the better team as of late: TCU.

(5) Houston over (12) UAB. What a huge bummer UAB couldn’t be matched up elsewhere; I wanted to see them make a run. I guess they technically could. But: at 80.9% to win, Houston is one of the largest 5-seed favorites in modern Tournament history. No 5 seed at 76% or better has lost (17-0). This reminds me strongly of Villanova/Winthrop a year ago.

(4) Illinois over (13) Chattanooga. Technically, this meets our criteria for a 13 over 4. UTC has a 31.2% shot to win per KenPom; we’re focusing on 13 seeds at 30% or better this year. Unfortunately, Torvik has them at just under 23%, and the average takes them out of full consideration. I would not be surprised at all to see UTC win this, though; Illinois is not a strong 4 seed.

(11) Michigan over (6) Colorado State. About once every Tournament, a 6 seed will be at 54% or worse to win their first game. Those 6 seeds are 4-16 in the Round of 64 since 2000. Colorado State is at 50.4% on KenPom and a hilarious 41.7% on Torvik. Absolutely amazing draw for Michigan, at least for one game.

(3) Tennessee over (14) Longwood. Refer back to the Texas Tech stat: No 3 seed at 85% or better has lost their Round of 64 game, a perfect 36-0. Tennessee is at 92%. Feels like the Wright State game all over again.

(10) Loyola Chicago over (7) Ohio State. A tricky one: Loyola is at 54.3% to win per KenPom, but 47.3% on Torvik. Loyola wins both the averaging out and is the better team in their last 10 games.

(2) Villanova over (15) Delaware. So: remember the note about 2/15 games needing to be at that 10% threshold to be generally pretty interesting? Villanova sits at 89.4% to win by KenPom, 90.8% on Torvik. They’re the only 2 seed on either site to dip below 90% to win. Do I think Delaware wins this game? No. But 15 seeds at 10% or better to win, despite being 4-24, have an average margin of defeat a few points shorter than those worse than 10%. I think this one could be worth tracking.

Midwest Region

(1) Kansas over (16) First Four. Again, not expecting much. Similar to how the 2/15 games have a threshold, 1/16 games sit at 5% or above for interest and curiosity. This one is consistently at 3.6%. Skippable, especially since it’s the final game of Thursday. That being said: Texas Southern, if they win their First Four game, has the best defense of any 16 seed this year at 108th overall and did beat Florida in December.

(8) San Diego State over (9) Creighton. FINALLY! An 8/9 game that doesn’t require a coin-flip. San Diego State is at 62.2% to win; 8/9 seeds at 55% or better are 37-6. I would be very surprised to lose this one.

(5) Iowa over (12) Richmond. Iowa is at 82.5% to win; refer back to the “no 5 seed at 76% or higher has lost” stat.

(13) South Dakota State over (4) Providence. Apparently the Giant Killers system hates this pick, but whatever. Providence is at just 57.2% to win. Not only is this the third-lowest mark for a 4-seed ever, I had to institute a new part of the study for it: teams at 65% or worse are 3-7 in the Round of 64 all time, with none of them breaching the Sweet Sixteen.

(6) LSU over (11) Iowa State. Disgusting. Sickening. Makes me want to barf in a bag and pour it on my laptop. Unfortunately, someone must win this game. LSU sits at 62.2% to win, the highest of any 6 seed this tournament, unless Rutgers plays Alabama. As much as I’d like to see all four 6 seeds lose, one of them probably has to win. Even better: as you’ll see in the Round of 32, one of these two may be in the Sweet Sixteen.

(3) Wisconsin over (14) Colgate. Which is because the committee placed KenPom #34 at a 3 seed, the lowest-seeded 3 since 2011 New Mexico (#39), who got ransacked in the Round of 32 by an 11 seed and nearly lost to 14-seed Montana. Wisconsin sits at 75.5% to win on KenPom and 81.4% on Torvik. I would not blame you if you feel compelled to pick Colgate, because 3 seeds at 80% or worse are 21-6 at winning. Still, that’s 21-6. I know that I’m personally rooting for Colgate, because presumably, most ESPN users have Wisconsin at least in the Sweet Sixteen.

(7) USC over (10) Miami (FL). Neither one of these teams is very good, and both are the lowest-rated teams at their respective seedlines. Congratulations to Auburn on the Sweet Sixteen bid. USC is at a combined 51.9% average to win, and 10 seeds ranked 50th or worse on KenPom are 3-13 since 2006. Nasty, nasty game. Nasty!

(2) Auburn over (15) Jacksonville State. Ever since I saw this tweet:

I was rooting for Auburn to draw the hardest 15 seed imaginable. Of course, they drew one ranked ten spots below Missouri. Auburn is at 91.4% to win; the only upset path I can think of is one where Auburn foolishly attempts 30 threes and misses 24 of them. Wait a minute, that’s actually pretty realistic.

East Region

(1) Baylor over (16) Norfolk State. But with a warning: this is the only 1/16 game this year where the 1 is below 95% to win. Baylor is still at 94.7%, so something would have to go wildly wrong for an upset to happen. Still, maybe this one provides some interest at some point.

(8) North Carolina over (9) Marquette. Another easy one: UNC is at 56.1% to win; the KenPom favorite is 57-25 since 2000. Stylistically, Marquette generates almost no second-chances at all, which is a problem against a UNC team ranked #2 nationally in DREB%. This would require Marquette shooting 40% or better from deep to win.

(12) Indiana over (5) Saint Mary’s. I would actually prefer Wyoming win, because Wyoming is subjectively much more fun to watch for me. That being said, they would have about a 31-32% chance to win; while that’s still the best of any 12 seed this year, the hit rate for 12 seeds greatly increases beginning at 33-34%. Indiana would be at 41.2%, easily the best of any 12 seed. If you like upsets, root for Indiana; if you like fun basketball, root for Wyoming.

(4) UCLA over (13) Akron. In general, this is either the worst batch of 4 seeds or the best batch of 13 seeds in a decade. But this game sort of ruins the average on both sides. UCLA is a 2 seed in a 4 seed’s body; Akron is ranked below three teams seeded 14-15. UCLA is at nearly 90% to win; no 4 seed has ever lost above 83%.

(11) Virginia Tech over (6) Texas. I cannot stand picking super-popular upsets like this one. There’s actually more statistical value in taking Texas, because over half of ESPN users have selected VA Tech. The problem is that this upset simply makes a lot of sense. 11 seeds at 41% or better to win since 2000 are 23-12; all others are 13-36. VA Tech is at 44%, the second-best of any 6 seed this year. Considering the stats expectation is that 1.7 of the 6 seeds lose, I would pair this with the Colorado State pick and hope for the best.

(3) Purdue over (14) Yale. An upset would be fun, but Purdue is at 90% to win. Really wish Princeton had won the Ivy.

(10) San Francisco over (7) Murray State. San Francisco is at an impressive 57.6% to win and weirdly isn’t the Vegas favorite. The metrics favorite in 7/10 games is 65-19 over the last 21 tournaments.

(2) Kentucky over (15) Saint Peter’s. Somewhat similarly to Villanova/Delaware, I could see this one maybe being interesting, but it’s unlikely. Kentucky is at 90.7% to win, right near that 10% cutline, because Saint Peter’s is the best of the 15 seeds by some margin and has a legitimate top-40 defense. My guess is more that this is a lower-scoring affair – maybe something like Kentucky 69-56.

Round of 32

West Region

(1) Gonzaga over (9) Memphis. The wild thing about this game in particular is that, in theory, it could be Gonzaga’s single toughest game they play prior to the Final Four. Memphis has played like a top-10 team over the last several weeks and is a legitimate threat. Still, Gonzaga sits at 82.9% to make the Sweet Sixteen. No 1 seed at 82.5% or higher has missed it (18 for 18). I’ll believe it when I see it.

(13) Vermont over (5) Connecticut. The risk here is kind of obvious: what if Vermont loses in the first round? Well, then you lose a total of three points out of 192. Big deal. This is more of a three-pronged bet:

  • (4) Arkansas sits at 36% to make the Sweet Sixteen, per KenPom. 4 seeds at 40% or worse are 2-for-29 in making the second weekend since 2000. This includes Purdue and Oklahoma State from 2021.
  • (5) Connecticut sits consistently at 38-39% to make it on both KenPom and Torvik. 5 seeds at 41% or worse to make the second weekend: 10-for-56. Creighton broke the trend last year, but they did get to play a 13 seed on the way there.
  • (13) Vermont is at 13-15% depending where you look. Only ten 13 seeds since 2000 have been at 12% or better to make the second weekend; they’re 3-for-10 in doing it.

The problem is that only three 13 seeds period have made the Sweet Sixteen. If you want to take UConn here, that’s a fine pick, too. Somewhere, though, you have to try and create added value. Vermont has better versus-the-field value than nearly any other 13 seed to take the court before them. It’s worth a try.

(3) Texas Tech over any of (6) Alabama, (11) Notre Dame, or (11) Rutgers. This one is drama-free. Tech is at 66.2% to advance; 3 seeds at 57% or better are 20-for-26. If you really want to narrow it down, 3 seeds at 64% or better are 8-for-8.

(2) Duke over (7) Michigan State. Otherwise known as THE MOST INSUFFERABLE GAME IN THE HISTORY OF MANKIND. Unfortunately, Duke is at 63.6% to advance on KenPom and a bizarrely-high 69.8% on Torvik; 2 seeds at 63% or better are 32-for-42. If we’re lucky, this is Davidson vs. Cal State Fullerton instead.

South Region

(1) Arizona over (9) TCU. Same with Seton Hall. Arizona is at 76.3% on KenPom and 71.9% on Torvik; 1 seeds at 70% or higher are 51-for-57.

(5) Houston over (4) Illinois. With ya, Ken:

I just…don’t get it? Even if you like Illinois in this specific matchup, they’re at just 28.2% to touch the Sweet Sixteen. No 4 seed at 34% or worse has ever made the Sweet Sixteen (0-for-13). I guess Illinois could make it 1-for-14, but I’m not sure what about them is trustworthy.

(3) Tennessee over (11) Michigan. If you are a reader of this blog, you know that this will be a nightmare game for me. The good-ish news is that it doesn’t feel terribly dramatic. Tennessee is at 66.3% to touch the second weekend; refer back to the Tech stat for why those are absurdly good odds for a 3 seed.

(10) Loyola Chicago over (2) Villanova. Here is the thing: in a one-off affair, Villanova would be favored by four points or so. That still means Villanova is the most likely Sweet 16 squad. They’re at 58.8% to make it, per KenPom. And I know Tennessee fans probably want no part of another Sister Jean run. But whatever. I’m showing my work:

  • 2 seeds at 63% or worse to make the Sweet Sixteen are 19-for-42 in doing so. Villanova is the only 2 seed below 63% this season on KenPom. (Duke is at 63.6%, for the record.)
  • 2 seeds at 60% or worse, per Torvik’s database, sit at 14-for-35 in making the second weekend, just a 40% success rate.
  • 2 seeds with a spot differential of 20 or lower in KenPom are 16-12 since 2006. Loyola Chicago is 13 spots behind Villanova, while Ohio State is also pretty close at 21.
  • 2 seeds favored by 5 or fewer points, since 2006: 9-11. Villanova, per KenPom, would be favored by 4.08 points.
  • Loyola sits at 22.2% to make the Sweet Sixteen, per KenPom. 10 seeds at 17% or higher to make the Sweet Sixteen are 11 for 28 in doing so; all others are 2 for 56. (For what it’s worth, San Francisco and Davidson also meet this. Go 10s!) 20% or higher: 9 for 20.
  • Even if this is Ohio State instead, it’s still worth rooting for, but Ohio State is below the 20% threshold for 7 seeds to make the second weekend. (Below: 5-for-46. Above: 14-for-38.)

To win a bracket pool, you have to swing for the fences on a couple of picks. Just like last year, I’m hoping Loyola repays the faith. All of this over a two-point pick. Why do I do this?

Midwest Region

(8) San Diego State over (1) Kansas. This pick was originally Kansas over SDSU. This one is absurdly tough, and it’s one where I have to go against my first instinct, which is to pass on the upset. Kansas is the only 1 seed below 70% on KenPom to make the Sweet Sixteen, and there is real value to be had in betting against them. The problem is two-fold: 1. San Diego State’s offense, which ranks 157th on KenPom; 2. Since 2006, 1 seeds playing 8/9 seeds with offenses ranked worse than 50th are 23-2.

The odds of all four 1 seeds making the Sweet Sixteen are just 31%. This is fully an odds play; I am trusting that San Diego State will overcome their offense to go deep. They are the best 8 seed on the board by some distance, and their odds are significantly better than those of 8 seed counterpart UNC. Gotta swing for it.

(5) Iowa over (13) South Dakota State. Even if Providence wins, this is very easy: Iowa is at 60.4% to make the Sweet Sixteen, and 5 seeds at 45% or better are 13-for-16 in advancing. Quick sidebar: Iowa’s 60.4% Sweet Sixteen odds are the highest ever for a 5 seed.

(6) LSU over (3) Wisconsin. LOL. But I can defend this. LSU has the best odds of anyone in this quadrant to advance at 40.2%; 6 seeds at 34% or better are 11-for-23 in advancing. This is more a bet against Wisconsin, who sits at 34.5% to advance. 3 seeds at 42% or worse are 2-for-11 in making it. What an awful quadrant. This could just as easily be Iowa State/Colgate, frankly.

(2) Auburn over (7) USC. As funny as an Auburn loss would be, it seems unlikely. Auburn sits at 69.5% to advance; 2 seeds at 63% or better are 32-for-42 in making it.

East Region

(1) Baylor over (8) North Carolina. This is the toughest one in the entire bracket for me.

Do I like this pick all that much? Frankly, no. I don’t like either Baylor or Kansas to go far at all, really. But this came down to two things:

  1. Significantly more people are picking UNC than San Diego State despite SDSU having the better upset odds;
  2. In a larger bracket pool, that makes San Diego State the superior value pick over UNC.

I believe that one of the 1 seeds will be going home before the Sweet Sixteen, based on the numbers that show me 1.02 are expected to go home. I think that it will either be Baylor or Kansas. I just simply think Kansas may have the worse matchup.

(4) UCLA over (5)/(12) Doesn’t Matter. UCLA is at 58.2% to advance per KenPom and 58.1% per Torvik; both are well above the standard 53% rate that I normally look at for obvious 4-seed advancements. 53% or better: 15-for-17 making S16. I suppose I’m rooting for the 12 seed to win so UCLA’s path is better.

(11) Virginia Tech over (3) Purdue. Slightly over half of all 3 seeds make the Sweet Sixteen: 48-for-84 since 2000, or 57%. We all expect 3 seeds to make it every year, particularly when their paired 6 seeds lose over 50% of the time in the first round since 2011. That’s just not the case. 3 seeds that actually make the Sweet Sixteen have insane win rates – 50% in the next round! – but getting there is a struggle.

Anyway, Purdue is at 50.5% to advance, per KenPom. 3 seeds at 52% or lower to make the next round are just 19-for-48; all others are 29-for-36. Even if you bump that up to 57% or lower, it’s still just 27-for-58, or a 46.5% shot at getting through. The other numbers are this: Ken’s numbers expect about 2.36 10+ seeds in the Sweet Sixteen. Torvik’s: 2.13. EvanMiya: about 1.96. The message is that at least two 10+ seeds should make the Sweet Sixteen. Three may be aggressive, but in 13 straight NCAA Tournaments, at least one 10+ seed has made the second weekend. The most common number of 10+ seeds in the Sweet Sixteen: three, which has happened 13 times in 36 tries. Try, try again.

(2) Kentucky over (10) San Francisco. This one is sad to say no to. Kentucky comes close to meeting the metrics for a two-seed loss, but at 64.9%, I can’t pull the trigger. Also, this is the nation’s #1 rim FG% offense going up against a team that ranked 90th-best in a significantly worse conference. For the record, I think Murray State would be an even less optimal matchup. The best possible single-game matchup here among the 7/10s may have been USC or Loyola.

Sweet Sixteen

West Region

(1) Gonzaga over (13) Vermont. Nice and calm. Gonzaga is at 70.8% to make the Elite Eight. Not only are those insanely high odds, but no 1 seed at 65% or better has missed the Elite Eight (13-for-13).

(3) Texas Tech over (2) Duke. You may have to wait until the second weekend to blissfully rid yourselves of the Retirement Tour™, but when it happens, I think it comes at the hands of this scary Texas Tech team. Not only would Tech currently be favored on a neutral court, they have superior Elite Eight odds to Duke (38.2% vs. 35.5%). 3 seeds at 31% or better: 16-for-24. All others: 8-for-60. 2 seeds at 40% or lower: 12-for-47. All others: 24-for-37.

South Region

(1) Arizona over (5) Houston. This one was a tough one to say no to. Both of these teams are utterly terrific, and the fact that KenPom’s #2 and #4 teams are forced to play each other in the Sweet Sixteen is an insanity that only European football’s seeding system can match. Again, this is a numbers game we’re playing. The numbers say this: Arizona has the second-best Elite Eight odds of the 1 seeds. 1.97 of the four 1 seeds are expected to get there, per KenPom. We need to drop two of them.

The issue is this: I just think Houston > Arizona is a taller mental gymnastics task than Iowa > Kansas. (Spoiler.) Teams with 44% or worse odds to make the Elite Eight are just 7-for-19; Baylor is at 43.3% on KenPom, while Kansas sits at 44.3%, right on the cutline. Arizona being at 47.2% was enough to push them just over the edge. Houston would have the on-paper shot volume edge, but Arizona would win in the foul department with relative ease and should out-shoot Houston. Tough, but fair.

(3) Tennessee over (10) Loyola Chicago. I would also take Tennessee over Villanova, for the record, as they’re the higher-rated team on both KP and Torvik. I promise this makes some pretty good sense, though, and not just out of homerism. Tennessee’s 39.2% Elite Eight odds are the eighth-highest ever for a 3 seed, per Torvik. The seven teams all ahead of them: Elite Eight entrants. Alternately, just refer back to the Texas Tech stat. There are two terrific 3 seeds this year and two meh ones; ride the two terrific ones.

Midwest Region

(5) Iowa over (8) San Diego State. Even if this is Kansas instead, Iowa is actually ranked ahead of Kansas on Haslametrics right now, and it genuinely may be defensible. Since January 15th – two months ago today – Iowa has played at the level of the 4th-best team in America, per Torvik. Kansas is third, but it’s a virtual tie. Frankly, it just comes back to the numbers: Iowa has the HIGHEST ELITE EIGHT ODDS EVER for a 5 seed at 31.4%. If not now, when? I shudder to think of the Fran McCaffery takes I’ll have to delete.

(2) Auburn over (6) LSU. Unfortunately, this one is straight forward: Auburn is at 48% to make the Elite Eight; 2 seeds at 40% or better are 24-for-37 (12-for-47 all others) and they very nearly crack the 50% super-safe barrier. Here’s hoping for an upset somewhere along the line. Jacksonville State?

East Region

(4) UCLA over (1) Baylor. 4 seeds at 25% or better to make the Elite Eight are 6-for-15, and UCLA sits at about 31-32%. Baylor is at just 43.3% to reach the Elite Eight, the lowest of any 1 seed. With the knowledge that 1 seeds at 44% or worse to make the Elite Eight are 7-for-19 in doing so, I am simply playing the odds.

(2) Kentucky over (11) Virginia Tech. Again, hoping for an upset, but I imagine the miracle VA Tech run ends here if it gets that far. Kentucky is *just* above that 40% barrier to crack the Elite Eight at 41.3%. Statistically, we can expect 0.73 10+ seeds to make the Elite Eight. I feel fairly confident that Virginia Tech is the most likely team to make it happen. Go Hokies?

Elite Eight

West Region

(1) Gonzaga over (3) Texas Tech. Gonzaga is at an astounding 53.7% to make the Final Four. Only 16 1 seeds have ever cracked 46%, and they’re at a collective 15-for-16. Gonzaga did it last year; I bet they do it again this time out, too.

South Region

This should go over well.

(3) Tennessee over (1) Arizona. Again: I’m merely playing the numbers. This is not a homer pick. If I could pick against Tennessee, I would, because doing this makes me nauseous. But I want to show you a couple of things.

These are the current odds on KenPom and Torvik. Arizona sits at 29.1% to make the Final Four on one site and 21% on the other. I think Torvik’s number is a little wild, but bear with me. 1 seeds with 33% or lower odds to make the Final Four are 4 of 40 in doing so (all others 28 of 44). Only one 1 seed this year is better than 33%: Gonzaga. The expected number of Final Four teams that are 1 seeds: 1.36 per KenPom, 1.29 per Torvik. That’s not two. That’s one.

Tennessee’s odds, for a 3 seed, are consistently at 20% to make the Final Four. Those are the seventh-highest odds for any 3 seed since 2000. They’re 5% higher than Texas Tech’s in 2019. 7% higher than Michigan’s in 2018. They are 1.7% higher than Florida in 2006. What I am telling you is this: it is okay to believe that the chance is real. If you don’t know, it was probably even harder to believe Tennessee would win the SEC Tournament, a tournament they had a 20.4% chance of winning.

The chance is there. Tennessee has a 53% chance of playing someone other than Arizona if they can make the Elite Eight. We’ll see if it happens.

Midwest Region

(5) Iowa over (2) Auburn. I don’t trust anyone in this region at all for a variety of reasons. Kansas’s metrics are very weak for a 1 seed hoping to make the Final Four; they’re out. Wisconsin is one of the weakest 3-seeds ever. Ditto Providence at the 4. 6-seed LSU has good metrics, but questionable motivation after firing their head coach. USC is the weakest 7-seed in the field. Miami is the weakest 10. San Diego State is the best 8, but their offense ranks in the 150s. Creighton lost a starter and is a big underdog in their first game. Iowa State looked great two months ago but appears to have firmly ran out of gas, scoring 41 points in their only Big 12 Tournament game.

That leaves you with Iowa, a team that hasn’t touched the Sweet Sixteen since 1999 or the Elite Eight since 1987. It has never made a Final Four in the 64-team era. It also leaves you with Auburn, a team that was #1 at one point this season. Auburn was great for a while, but they have some extremely obvious shortcomings: 258th in 3PT%, a bad offensive steal rate, and a 5-4 finish to the regular season that included an awful SEC quarterfinals loss. High seeds that lose in the conference tournament quarterfinals rarely make positive history.

However, high seeds that lose in the conference tournament quarterfinals rarely get such an advantageous draw. Think about Baylor, another team that copied Auburn’s result. Baylor’s path to the Elite Eight is likely #28 and #8 in KenPom. Auburn: #42 and #19, if LSU holds it together long enough. The problem with that is that Auburn just got done losing to #43 on a neutral court. Iowa, meanwhile, has lost twice since January 31 and has risen to #13 in KenPom, just three spots behind Auburn.

Frankly, this is not a pick I love. If any region seems destined to have a truly absurd champion, it is the Midwest, a region where teams seeded 5 or worse have a 39% chance of coming out on top. That is insane, especially when you consider the other three regions are at 21.5% (East), 14.2% (West), and 29.7% (South). The South Region reasonably could blow up, sure, but most of that value is generated by Houston with 17.6%. The non-Iowa 5+ seeds are at 20.7%, and Iowa has better Final Four odds than any other 5+ seed out there. Ride the insanity.

East Region

(4) UCLA over (2) Kentucky. Enough words. Straight to it: 2 seeds at 25% or worse to make the Final Four are 4-for-62 in getting there. Auburn is the only 2 seed that qualifies to go far enough this year, so maybe I lose it there, but whatever. At 17% to get to New Orleans, UCLA has the sixth-highest odds of any 4 seed since 2000 to do it. I recommend attempting to take advantage of one of the worst selection committee jobs in recent memory.

Final Four

(1) Gonzaga over (4) UCLA. When you say it like that it does sound pretty crazy: a straight-up Final Four rematch. It’s happened before, and UCLA was even involved in it with Florida, but it’s rare. Still, I just like the value of UCLA enough to take a swing at it. I think it ends against Gonzaga. Everything at this point of the tournament is a pure coin-flip, but Gonzaga sitting at 38.5% to make the title game is pretty good. 1 seeds at 33% or better, per Torvik, are 10-for-12 in getting there.

(3) Tennessee over (5) Iowa. Look, if you’re still reading, I think you know how stupid and insane this game even looks on paper. Tennessee ranks higher on every metrics site but Haslametrics and is one of the best 3 seeds ever. Give it a whirl, see how it feels.

THEN:

(1) Gonzaga over (3) Tennessee. Whether Tennessee makes it or not is sort of besides the point for this exercise. Gonzaga is at 27.5% to win it all; no other team is above 9%. Even Arizona is the only other team above 6.6%. Baylor entered at 8.2% last year, but those numbers were COVID-dampened. 2019: Virginia, 21.4%. 2018: Villanova, 18.1%. 2017: UNC, 10%. Weird champions happen, but the majority were at least in the double-digits. Only three champions have been below 7% to win it all since 2006: both UConn titles and 2015 Duke. That’s 11 of 14 tournaments where an unsurprising champion came out on top. Barring some sort of serious surprise, your champion this year is either Gonzaga or Arizona. I have Gonzaga as the only one of these in the title game, so there you go.

One final plea: please do not use these picks to bet on games or futures or whatever. These are for fun and are the product of a lot of weird research I feel bizarrely compelled to work on. If you want to win a bracket pool, please do not expect much, because I did not win mine last year and have not since 2010. However, I hope this helps.

2021-22 Bracketology, Vol. 1: where Tennessee stands, how they could rise/fall, and best/worst NCAA Tournament draws

by Will Warren, posted in Tennessee Basketball

Normally, I don’t do any NCAA Tournament or bracket-related posts until the Monday after the Super Bowl. Most readers generally can’t get into it until that date, and neither can I. However, 2022 presents a new challenge: the NFL has turned their season into 18 weeks, and therefore, the Super Bowl is now the second Sunday of February, not the first. This is fine, because I respect and worship The Shield™, but it’s not great for content planning or breaking a long-standing routine.

So: a compromise. This is a post that serves as an introduction to Tennessee’s 2022 bracket concerns. There will be two more posts – one February 22, one March 8 – that go into further detail, answer certain questions, and explore some ideas of what works and what doesn’t in March. For now, think of this as a post that answers three concerns:

  1. Where does Tennessee stand right now from a seeding perspective?
  2. With eight games left, how will various finishes and final records leave Tennessee looking, both in SEC standings and in Tournament seeding?
  3. What are Tennessee’s best and worst possible draws?

This is a post that comes just shy of 3,000 words, so we’ll dive right in.

1. Where does Tennessee stand as of February 8?

Pretty well! Thanks for asking.

…anyway, Tennessee sets up pretty nicely for a run to the finish. Bracket Matrix, the bracket consensus site, has not updated since February 4 at the time this piece was typed. As of last Friday, they had Tennessee as the highest 5 seed, which is confusing because they have the same number of good wins and two fewer bad losses than 4 seed Michigan State, but if it’s surprising that Michigan State gets more love than Tennessee I have a beach house in Idaho for you.

More reliable, at least for the purposes of what we’re discussing, is Bart Torvik’s TourneyCast tool. It’s the only tool out there I trust that accurately represents the instability being a month-plus away from the NCAA Tournament possesses. When you’re in Tennessee’s position – good enough to be widely considered no worse than a top-20 team, not good enough to be considered a top-10 one – you have quite the amount of uncertainty. Factor in Tennessee’s 3-6 record against Quadrant 1 teams and you can see where a bracketologist might be a little lower on Tennessee than a metrics guy.

Still, things look pretty good. Here’s what the TourneyCast has to say about Tennessee’s seeding odds the remainder of the way. I only used those with at least a 5% chance of happening, because otherwise, I will get a reply to this article asking me about the possibility of being an 8 seed.

  • 1 seed: 5.2% chance
  • 2 seed: 17.6%
  • 3 seed: 21%
  • 4 seed: 21.3%
  • 5 seed: 19.6%
  • 6 seed: 11.2%

Tennessee’s average seed on TourneyCast is a 3.8, which would translate to them being roughly the 14th-ranked team in the field on Bracket Matrix. Again, seems like a fair guess for a team ranked 13th on KenPom and 18th in Wins Above Bubble.

Prefer something you think sounds more reliable? Say, a little tool from the Worldwide Leader? Well, wouldn’t you know it, ESPN now has a seed projection tool that has Tennessee at…

…oh dear. Well, forget you saw that. As a reminder, BPI believed Tennessee was the best team in the SEC last year entering the SEC Tournament, which was mighty hard to defend if you look at any metrics site in existence. I can’t figure out how they weigh certain games; if I find out I’ll let you know.

Here’s two more that I feel less confident in, but exist and are worth taking a look at. First, TeamRankings.com has a bracketology tool that places Tennessee as the 16th-ranked team in the field. They don’t provide an average seed, but their graph is less friendly to Tennessee, giving them no real shot at a 1 seed (fine) and a >5% shot at an 8 seed (wait a minute).

Lastly, there’s this mysterious website called INCC Stats. Its first function appears to be as some sort of cross-country running site for Indiana high school athletes, but its secondary function is that of a college basketball ratings site that offers a remarkably strong resemblance to a blend of Sports-Reference and KenPom. Tennessee ranks 14th in their field as well, with an average seed of 4.3. (If Tennessee beats Mississippi State, they’re expected to improve to 3.9.) This is a little harder to read because this goes along with a chart of various SEC records, but just follow the basic outline at the bottom.

So, what have we learned? What can you learn?

  • Tennessee, on the majority of sites with an NCAA Tournament forecasting tool, has roughly a 60% chance of finishing somewhere on the 3-5 seed lines. That doesn’t mean Tennessee will do this; it just means that as of February 8, it’s what they’re most likely gonna do.
  • Tennessee’s most likely outcome is a 4 seed. Again, this isn’t a guarantee; it’s just what’s most likely. On the three sites that list their odds in full, Tennessee has between a 21-27% shot of being a 4 seed.
  • There are scenarios, some more realistic than others, where Tennessee could end up a 7 seed, 6 seed, 2 seed, or even a 1. I don’t think any of them sound super realistic, but ESPN sure seems like a believer.
  • To ensure Tennessee’s status as a 4 seed or better, they’ve gotta win some big games. Which leads us to the next question.

2. What records get you what seeds heading into the SEC Tournament?

With the help of both Torvik and INCC Stats, I’m just going with remaining records that at least have a puncher’s chance of happening. Even KenPom, which is a little rosier than Torvik on Tennessee, only gives the Volunteers a 4.8% chance at going 8-0 to finish the season. Could that happen? Sure. Will it? Probably not. Likewise, Tennessee going 0-8, 1-7, or even 2-6 is extremely unlikely – none of those scenarios have more than a 2% chance of coming to fruition. So we’ll focus in on five specific runs to the finish, all of which have at least a 7% chance of happening: 3-5, 4-4, 5-3, 6-2, and 7-1.

The average seeds by record are via INCC Stats; considering Torvik is a little more positive on their seeding outcomes, you can bump up the seed listed by 0.1-0.2 if you want his numbers instead.

19-11, 10-8 SEC (3-5 finish)

  • Losses to no fewer than one Quadrant 2/3/4 opponent
  • Assuming 0-1 wins over Quadrant 1
  • Expected SEC finish: 4th or 5th
  • Average seed: 6.9
  • Chances of happening: 6.9% INCC, 6.4% Torvik, 4.4% KenPom

This would be a disastrous run to the finish, of course. Tennessee currently sits in a position where they’re >75% likely to finish in the top 3 in the SEC and >90% likely to finish top 4; this is the only realistic outcome that would put a top-four finish in very real doubt, and if you’re going 3-5 to the finish against this schedule, anything is possible in the SEC Tournament in a bad way.

Tennessee would have done enough to likely be a 7 seed, but if Tennessee gets here, it means they’ve lost all four of their coin-flip games (Mississippi State, Kentucky, Arkansas, Auburn) and suffered a real upset at the hands of, say, Missouri. It’s the type of finish that makes you an obvious fade bet in the first weekend. Let’s pretend we didn’t see this.

20-10, 11-7 SEC (4-4 finish)

  • Assuming 0-4 or 1-3 Quadrant 1 record
  • Expected SEC finish: likely 4th, but a decent shot at 3rd
  • Average seed: 5.5
  • Chances of happening: 16.8% INCC, 18.3% Torvik, 13.1% KenPom

It says a lot about the quality of the SEC beyond the top three that even a 4-4, 11-7 SEC finish would still probably keep Tennessee at an SEC Tournament double-bye and maybe even the third overall seed. Still, this would be disappointing, even with Olivier Nkamhoua’s injury.

As of now, the 5.5 average seed would actually make Tennessee the highest-rated 6 seed, but the 6 seed line must be avoided at all costs. Since the field was expanded to 68 in 2011, 6 seeds are 19-21 in the first round and have just a 15% success rate in making the Sweet Sixteen

21-9, 12-6 SEC (5-3 finish)

  • Assuming no worse than one additional Quadrant 1 win, or two Quadrant 1 wins but one Quadrant 2-4 loss
  • Expected SEC finish: 3rd
  • Average seed: 4.6
  • Chances of happening: 20.5% INCC, 29.6% Torvik, 25.5% KenPom

This is where I make a special note than KenPom’s numbers give Tennessee a better shot at finishing 13-5 than 12-6, but the difference on both Torvik and KenPom is small. You’re looking at a coin-flip to either finish with 12 or 13 SEC wins, essentially. Given the Nkamhoua news, maybe that coin flip goes against you now. Who knows?

Anyway, this means you have defeated at least one and possibly two of State/Kentucky/Arkansas away/Auburn while either losing the rest of losing to one of Missouri/Arkansas home. It’s a decent outcome, but you’d essentially go into the SEC Tournament needing a win over likely 2-seed Kentucky to move from the 5 line to the 4. Once again, this is important; 5 seeds are 24-16 in the first round since the field expanded, while 4 seeds are 31-9 and have made the Sweet Sixteen twice as often (23 to 11).

22-8, 13-5 SEC (6-2 finish)

  • Assuming a 2-2 record against Quadrant 1 opponents or a 3-1 record with one Q2-Q4 loss
  • Expected SEC finish: likely 3rd, outside shot at 2nd
  • Average seed: 3.8
  • Chances of happening: 27.1% INCC, 27.7% Torvik, 29.9% KenPom

If you combine the odds of these two sites + KenPom, this is Tennessee’s most likely outcome. What an outcome it would be: multiple wins over the State/Kentucky/Arkansas away/Auburn grouping, which would push you to five Quadrant 1 wins. As long as you draw any of Arkansas/Florida/Mississippi State in the SEC Tournament quarterfinals, you have an opportunity to add a sixth. That’s important.

In this scenario, assuming an SEC Tournament semifinal loss, Tennessee would close the season with an 11-9 record against Quadrant 1 + 2 opponents combined. Only 23 teams in 2018-19 played at least 20 games against Q1+2 competition and finished with a winning record. That, combined with Tennessee’s flawless record against Quadrant 3 & 4 opponents, would likely lock up no worse than a 4 seed.

There is one moderately realistic outcome that’s better, though.

23-7, 14-4 SEC (7-1 finish)

  • No worse than a 3-1 record against Quadrant 1 opponents
  • No losses to Q2-Q4 teams
  • Expected SEC finish: no worse than 3rd, with a tiebreaker with Kentucky determining 2nd
  • Average seed: 2.9
  • Chances of happening: 16.6% INCC, 13.7% Torvik, 20.4% (!) KenPom

Well, it’s not out of the question, but it’s less likely than 13-5, 12-6, and 11-7. Still, it’s worth exploring. Here’s the path I would deem most likely:

  1. Tennessee holds serve in three should-be-easy games: home Vanderbilt (88% to win, per KenPom), road Missouri (83%), road Georgia (91%). The odds of these three events happening are 66.5%.
  2. Tennessee beats Arkansas (78%) on Senior Day. Arkansas is good and getting better, but it’s a home game at a place Arkansas rarely wins. This plus the three gimmes: 51.9%. Still looking good.
  3. Tennessee beats Mississippi State (60%) tomorrow. 31.1%, which I swear to you is pretty good for a series of five independent events.
  4. Tennessee wins two of three against the following: Arkansas on the road (57%), #1 Auburn (55%), or #5 Kentucky (49%). If you treat these three as independent events, which they are, Tennessee has a 90% shot at getting one win. Against this trio specifically, Tennessee’s most likely outcome is 2-1. That’s if they proceed through the 31% likelihood of going against the other five teams unscathed.

You can see why this is unlikely, even though Tennessee is technically favored in seven of their final eight. (Because I am sure I’ll be asked, their average seed if they win out is 2.0, but again, that’s a 6% event. Can happen, very likely won’t.) Still: the path is there, and you don’t have to squint much at all to see it. If Tennessee can get a few lucky bounces here or there, 7-1 is 7-1.

3. As of today, what are Tennessee’s best and worst NCAA Tournament draws?

All seedings are based on two things:

  1. Where they are on the Bracket Matrix consensus as of right now;
  2. For seeds 12-16, if they have a >15% chance of making the NCAA Tournament, per Bart Torvik’s TourneyCast. Upsets happen in conference tournaments, and this is our way of getting them in a projected field. (I needed a cutoff point and 15% sounded like an accurate midpoint between 10-20%.)

Also, for the purposes of our…”study,” I guess, we’re assuming that Tennessee makes the NCAA Tournament as a 4 seed. It’s their most likely outcome on pretty much every site I read, and we can adjust that projection in a couple of weeks if needed. All numbers in the projections are via KenPom.

The GOAT Draw

  1. Auburn (SEC)
  2. Duke (ACC)
  3. Wisconsin (B1G)
  4. Tennessee (SEC)
  5. Ohio State (B1G)
  6. Texas (B12)
  7. Saint Mary’s (WCC)
  8. Boise State (MWC)
  9. Loyola Chicago (MVC)
  10. Seton Hall (BE)
  11. North Carolina (ACC)
  12. North Texas (CUSA)
  13. Hofstra (CAA)
  14. Liberty (A-Sun)
  15. Colgate (Patriot)
  16. Penn (Ivy)

“But Will,” you holler, “you put the #1 team in Tennessee’s bracket. Surely, this must be a mistake.” Shirley, I can be serious. Of the current #1 seeds – Auburn, Gonzaga, Purdue, and Kansas – it is Auburn who ranks lowest in a metrics average of the four, settling in around the eighth-best team in America. If you want to replace Auburn with Kansas, who is a spot or two ahead of them in some places, go for it. But I’m standing by this.

Auburn has very much earned their status as #1 in the AP Poll through a variety of great wins and timely cash-ins of luck. But the very best teams in college basketball rarely have to escape the bottom two teams in their conference by a combined three points. Sure, Auburn won both games…but did it actually make you feel better about their odds of somehow going undefeated in SEC play? I’m banking on two things here: a moderate reversal of fortune (i.e., Auburn loses 2-3 SEC games, loses in the SEC Tournament) and a genuine belief that even a really, really good team is capable of a bad night at the wrong time in March. This is the same team that’s nearly lost to South Florida, Saint Louis, Missouri, and Georgia. They’re not perfect.

This draw is complicated because, on one hand, it would be perfect if Tennessee drew literally the worst possible 1-16 seeds at every line. However, I have an alternate plan: what if Tennessee’s potential Round of 64, Round of 32, and Sweet Sixteen higher seeds are all below-average, but their bottom-bracket adversaries (seeds 6, 7, 10, 11) are all really good? If Tennessee made the Elite Eight and all they had to do was beat a 6, 7, 10, or 11 seed, you would take that gamble every single time, even if it didn’t work out. Therefore, you want really, really good lower seeds on the bottom half to knock out the 2 and 3 seeds.

Along with that, we’re making the 14, 15, and 16 seeds super-powered, or at least as much as I realistically can. (I debated doing the same with the 8/9 seeds, but it’s probably in Tennessee’s advantage to just draw an average 8 or 9 seed, not 2020-21 Loyola Chicago or something.) (Also, I would prefer a 12 seed that is good enough to be threatening but not scary enough to knock Tennessee out themselves. Learned my lesson last year.) The goal is this: Tennessee makes the Elite Eight while facing a 13, a 12, and a 9 seed. Worst case scenario: you draw the worst 13, worst 5, and worst 1 seed in the field.

What this produces is the following:

  • Baseline Tennessee NCAA Tournament odds, per Torvik: 80.7% to make Round of 32; 47.8% to make Sweet 16; 22.9% to make Elite Eight; 10.6% to make Final Four
  • GOAT Draw: 90.4% Round of 32; 64.7% Sweet 16; 33.6% Elite Eight; 18.2% Final Four

This draw almost doubles Tennessee’s odds of the first-ever Final Four run in school history. Not only is that nice, but check out the rest of this bracket. The 3 seed is only 27% likely to make the Sweet Sixteen. The 2 seed has just a 44% shot to see the Elite Eight. Your second and third most-likely draws from the bottom half of the bracket are the 6 and 7 seeds. If I, today, told you Tennessee could have a one-in-three chance to make the Elite Eight…you’d take that, right? That’s why this is the GOAT Draw.

The Poop Draw

  1. Gonzaga (WCC)
  2. Arizona (Pac-12)
  3. Houston (American)
  4. Tennessee (SEC)
  5. Texas (Big 12)
  6. Iowa State (Big 12)
  7. Indiana (B1G)
  8. Boise State (MWC)
  9. Davidson (Atlantic 10)
  10. Miami (FL) (ACC)
  11. Wake Forest (ACC)
  12. UAB (Conference USA)
  13. Furman (Southern)
  14. Navy (Patriot)
  15. Cleveland State (Horizon)
  16. Nicholls State (Southland)

A truly funny and terrible thing is that this draw actually gives you a better shot to exit the first round and an equal shot to make the Sweet Sixteen as the baseline. This is because the baseline has a lot of uncertainty involved and considers pretty much anything from a 2 to 6 seed for Tennessee to at least be mildly possible.

For this, we’re giving Tennessee a true murderer’s row to get through. If you can escape Furman – the highest rated potential 13 seed in KenPom at #67 – you’re either rewarded with a Texas team you just lost to a week ago or a UAB team that’s 38th in KenPom. Escape that grueling weekend? Congrats. Your reward, in 87% of simulations, is #1 Gonzaga. If you somehow pull off the upset of a lifetime, your next reward is either #2 Arizona or #4 Houston.

My hope with this draw is to illustrate what a true Selection Hell looks like: Tennessee gets the worst possible 1-3 and 5 seeds, gets zero help from the 6-11 seeds, then proceeds to draw the toughest 12 and 13. It’s an absolute nightmare…and yet: that absolute nightmare still results in a Sweet Sixteen almost half the time. I guess it could be worse!

A summary of what has been learned so far:

  • Tennessee will most likely be a 3, 4, or 5 seed.
  • Their most likely final SEC record (per KenPom) is 13-5, closely followed by 12-6, 14-4, then 11-7.
  • Hope you don’t draw Furman or UAB.
  • Hope your enemy seed draws Furman or UAB.
  • Hope you get a bad 1 seed or your 1 seed draws Dream Killer Loyola Chicago.

More to come soon.

The best (and worst) value picks for the 2021 NCAA Tournament

by Will Warren, posted in Basketball Statistics

No fancy intro here – just big, dumb stats stuff.

The values below are based off of the traditional bracket scoring of 1 point per Round of 64 win, 2 per Round of 32, 4 per Sweet Sixteen, 8 per Elite Eight, 16 per Final Four, and 32 for getting the champion right.

Round of 64

As a reminder, picks in this round are just one team against another; if one team is a great value pick, their opponent is a bad one. Pretty simple.

Best Values

(9) St. Bonaventure over (8) LSU (50.1% to advance/31.9% ESPN picks/+.182 expected points added). Only 31.9% of the public have picked the Bonnies, but St. Bonaventure is the KenPom favorite and, in my combined metrics, a 50.1% likely winner. Is this a safe pick? No, but it could be a huge first-round swing pick.

(11) Utah State over (6) Texas Tech (39.7% to advance/22% ESPN/+.177 EPA).

(13) North Texas over (4) Purdue (28.6% to advance/11% ESPN/+.176 EPA).

(10) Rutgers over (7) Clemson (53.6% to advance/39.3% ESPN/.143 EPA). I assume that this is some sort of weird college football residual affection for Clemson?

Worthy Moon Shots

(14) Eastern Washington over (3) Kansas (23.5% to advance/7.1% ESPN/+.164 EPA). No team in this bracket has less value than Kansas, to be honest, though I see this being more of an issue in the Round of 32 than here.

(14) Abilene Christian over (3) Texas (23.2% to advance/7.3% ESPN/+.159 EPA).

(13) Liberty over (4) Oklahoma State (26.1% to advance/14% ESPN/+.121 EPA).

(14) Colgate over (3) Arkansas (25.3% to advance/13.9% ESPN/+.114 EPA).

Round of 32

Best Values

(6) USC over either (3) Kansas or (14) Eastern Washington (44.6% to advance/24.9% ESPN/+.394 EPA). USC is well-rated in the metrics, would be a neutral-court favorite over both teams, and both were true before Kansas announced a starter would be out for at least the first weekend. As mentioned above, Kansas may be the single-worst value pick in this Tournament.

(6) BYU to the Sweet Sixteen (34.8% to advance/18.9% ESPN/+.318 EPA). I really do not like this pick, for the record. BYU is going to be very vulnerable in the first round. This is more of an anti-Texas pick, which will be explained later.

(8) Loyola Chicago over (1) Illinois (21.3% to advance/6.5% ESPN/+.296 EPA). Explained yesterday.

Teams to Avoid

(3) Kansas (36.8% to advance/68.2% ESPN/-.628 EPA). If you add Jalen Wilson back to these projections, Kansas is still the lowest-value pick of the first two rounds. They haven’t gotten above 19th in KenPom since Tennessee blew them out in late January, and the last time they actually played like a 3 seed (AKA, being ranked 9th-12th in KenPom) was January 9.

(3) Texas (41.9% to advance/70% ESPN/-.562 EPA). Texas has a really tough Round of 64 matchup and would either face BYU or a very good 11 seed in the next round. The Big 12 wasn’t quite as good this year as it has been the last several, and I think in general, the league’s being overvalued this March.

(4) Oklahoma State (38.2% to advance/59.4% ESPN/-.424 EPA). I promise this wasn’t all that intentional, but again, another Big 12 team to avoid. Oklahoma State never got higher than 30th in KenPom, which is the range for 8 seeds traditionally.

Sweet Sixteen

Best Values

This round isn’t quite as ripe as the others, to be honest. This is more of a round to avoid certain teams than to go all-in on some.

(6) USC (18.6% to advance/7.6% ESPN/+.44 EPA). Not a great value, to be honest, but it’s #1.

(8) Loyola Chicago (13% to advance/3% ESPN/+.389 EPA). 

(7) Connecticut (15.8% to advance/6.2% ESPN/+.384 EPA). I actually like this one, for what it’s worth.

(6) BYU (15.6% to advance/6.4% ESPN/+.367 EPA).

Teams to Avoid

(1) Baylor (50% to advance/70.1% ESPN/-.806 EPA). Baylor has to go through a really good 8/9 seed combo and either Purdue or Villanova in the Sweet Sixteen. Let me put it in perspective: Gonzaga actually does have 70% odds to advance. Baylor does not.

(1) Michigan (44% to advance/63.7% ESPN/-.79 EPA). This hurts, but no Livers is a real killer. If you haven’t watched them, Livers would pretty easily be the best player on Tennessee, for example.

(1) Illinois (51.1% to advance/69.1% ESPN/-.719 EPA). 

(3) Kansas (12.7% to advance/28.3% ESPN/-.626 EPA).

This team is actually lower than some others, but they required a shoutout: (2) Ohio State (43.7% to advance/57.6% ESPN/-.554 EPA). 

As well as this one: (2) Alabama (35.5% to advance/47.1% ESPN/-.462 EPA).

Elite Eight

Best Values

(2) Houston (31.3% to advance/15.3% ESPN/+1.28 EPA). This is our first >1 point value add of the Tournament and one of just four the entire way through. Two of the other three? Also Houston. I cannot believe the Cougars can be had at this “price,” if you will; they were in KenPom’s top 7 for basically the entire season and never wavered. There isn’t a realistic reason for Illinois to be receiving 53.9% of Final Four picks from this region when Houston would only be a one-point underdog on a neutral court.

(2) Iowa (20.9% to advance/14% ESPN/+.552 EPA). Iowa is up here simply because everyone is taking Gonzaga to the Final Four and further. Gonzaga has received an astounding 67.2% of ESPN user picks in this region, a full 13% higher than any other team in the field. I’m taking Gonzaga there myself, but there is real value in giving Iowa a shot if you’re in a bracket group of, say, 50+ people.

Teams to Avoid

(1) Illinois (31.7% to advance/53.9% ESPN/-1.773 EPA). Because Illinois has gotten hot at the right time, they’ve received the second-highest percentage of Final Four picks in the field. This is a lot for a team that has a negative turnover margin, ranks 235th in FT%, and has been the serious beneficiary of an opponent cold streak from three over their seven-game win streak (expected 3PT% 35.7%; actual, 29.7%). Illinois is a wonderful team, but they aren’t a lock to even get to the Sweet Sixteen given that they may be playing the ninth-ranked team in KenPom in the Round of 32.

(1) Baylor (33.6% to advance/50.1% ESPN/-1.32 EPA). Baylor’s struggles as of recent are probably overblown, but they kind of quietly received a real bummer of a draw. In the Round of 32, the Bears either have to face KenPom’s #10 team (Wisconsin) or the #28 team (North Carolina) that’s risen 14 spots in the last three weeks. The Sweet Sixteen opponent is likely either Purdue (#13 KenPom) or Villanova (#12). Lastly, they’d likely have to get past #7 Ohio State to make the Final Four. That’s as many as three Top 12 opponents in four games, which is a huge ask for pretty much any team. It’s like having to beat a 3 seed three straight times despite playing an 8, a 5, and a 2.

(1) Michigan (28.3% to advance/42% ESPN/-1.093 EPA). This is entirely based on the Livers injury. If he were available, Michigan would actually be the best value pick of any 1 seed.

(3) Texas (7.8% to advance/15.1% ESPN/-.583 EPA). Lowest value pick of any non-1 seed.

Final Four

Best Values

(2) Houston (19% to advance/7% ESPN/+1.919 EPA). The second-highest value by any team in any round.

(2) Iowa (13.1% to advance/7.6% ESPN+.879 EPA).

Teams to Avoid

(1) Illinois (19.8% to advance/30.6% ESPN/-1.722 EPA). 

(1) Baylor (19.4% to advance/29.6% ESPN/-1.629 EPA).

Champions

Best Values

(2) Houston (9.5% to win/2.5% ESPN/+2.229 EPA).

(2) Iowa (7.9% to win/3.4% ESPN/+1.426 EPA).

Way down the board, but still worthy of consideration in, say, 150+ person bracket groups: (2) Ohio State (4.7% to win/3.4% ESPN/+0.41 EPA).

Teams to Avoid

(1) Illinois (10.2% to win/14.9% ESPN/-1.516 EPA). Perhaps call this the anti-hot hand.

(1) Gonzaga (32.6% to win/37% ESPN/-1.415 EPA). Gonzaga has the best chance to win the NCAA Tournament of any team since 2000-01 Duke, and truth be told, they are a perfectly fine pick. If you’re in a 100-person bracket group, on average, 37 people will pick Gonzaga. If you’re one of those 37, you have to nail the earlier rounds to scratch out an advantage. That’s why Gonzaga is both the smartest pick in the field and one of the worst.

Thanks for tagging along! One final ranking is below.

Teams with the most total added points value in the field:

  1. Houston +5.514 expected points added
  2. Iowa +2.588
  3. Loyola Chicago +1.702
  4. USC +1.5
  5. Wisconsin +1.482

Teams with the lowest added points values:

  1. Illinois -6.136
  2. Baylor -4.741
  3. Gonzaga -3.339
  4. Michigan -3.112
  5. Kansas -2.656

How stats and history would pick the 2021 NCAA Tournament

by Will Warren, posted in Basketball Statistics

BIG LARGE UPDATE: I made a Google Doc showing how this method has performed over the last five NCAA Tournaments. It would have correctly selected three of five champions and 14 of 20 Final Four teams. You can check it out here. 

Last year, to avoid the depression of knowing the NCAA Tournament was cancelled, I posted a super-long statistical breakdown of how the stats I’ve collected over the years would’ve picked every game of the NCAA Tournament. It was really fun to put together, and I made myself a promise that when the 2021 Tournament came around, I’d do it again for the real thing.

Well, here we are. After the longest year in human history, the 2021 Tournament is right around the corner. Matchups were revealed yesterday evening, and I’ve spent the night and morning putting together what I think is 99% likely to be my own bracket. Does this mean you should copy it exactly? Of course not. What’s the fun in that?

What this is is simply a game-by-game projection of the field of 68 based on a document I’ve put together since 2018. Bart Torvik has an amazing page on his site with detailed historical KenPom projections of each game over the last 20 years of postseason play. Using that, I’ve accumulated enough data to make informed, quality guesses on how the NCAA Tournament may go.

The only NCAA Tournament I’ve had the chance to apply this to was 2019’s. It correctly identified three of the Final Four teams, missing only on Gonzaga, and said Virginia would be the national champion. It also correctly picked three Round of 64 upsets and went 26-6 in the first round. I’m hoping to beat both numbers this year, and we’ll simply have to see how it plays out.

First disclaimer: I feel confident that at least 90%, if not more, of the picks on the following pages will be the same ones I make in an actual ESPN bracket selection. That said, I reserve the right to change selections up to Friday morning, and if they’re meaningful, I’ll send out the article again with alterations.

Final disclaimer: this method is going to end up missing on several games and picks throughout March and early April. This is simply how it works, because it is impossible to have a perfect bracket.

To click ahead to a preferred section, use the below menu:

NEXT PAGE: Round of 64 picks

Pages: 1 2 3 4 5