Men’s 2022 March Madness Odds and Upset Predictions
Click here to skip ahead to our 2022 March Madness odds long version and summary.
There is no better time to be a data junky than the week leading up to the March Madness tournament. If KenPom is the brightest star in the analytics sky, I’m simply one in the buckle of Orion’s belt. But, the best brackets predictions combine all the individual stars to form a constellation of information. Combining lots of different models tends to give you a better picture of the landscape than looking at any one in particular.
Before going in depth talking about upsets, the following graphic summarizes our estimates of the 2022 March Madness Odds.
Technique
We used our Bayes Ensemble method to come up with a rating for each team in the tournament in the same style as SRS (Simple Rating System). That is, the difference between two team’s ratings is a prediction of their point spread on a neutral court. Our model computes each team’s rankings by combining game scores and Vegas championship odds as a prior for ridge regression.
With these rankings and by using a normal distribution to convert from predicted point spread to win probability, we ran over 100,000 Monte Carlo simulations and counted how far each team advanced in each simulation in order to estimate the probability of each team to advance to each point.
Combining these raw probabilities with the recent article by The Data Jocks on how to pick Cinderellas in March Madness can give further insight into how to find the right team. In particular, we want to look at those teams that play at a slower-than-average pace and give them a boost above even what our model says.
2022 March Madness Odds
I’m going to first dump all the data into two tables for you to browse. The first table corresponds to the probabilities of each team advancing in the first two rounds. This table also contains each team’s raw rating. The second table contains data more closely related to securing a championship. Here is the table for the first weekend probabilities .
| Team | Rating | R32 | S16 |
|---|---|---|---|
| Gonzaga | 26.11 | 1.00 | 0.89 |
| Arizona | 25.82 | 1.00 | 0.89 |
| Kansas | 24.47 | 1.00 | 0.88 |
| Baylor | 24.23 | 1.00 | 0.82 |
| Duke | 23.95 | 1.00 | 0.76 |
| UCLA | 23.94 | 1.00 | 0.82 |
| Kentucky | 23.47 | 0.99 | 0.84 |
| Auburn | 22.92 | 0.99 | 0.80 |
| Villanova | 23.85 | 1.00 | 0.70 |
| Purdue | 23.82 | 1.00 | 0.68 |
| Illinois | 24.04 | 0.98 | 0.54 |
| Tennessee | 22.65 | 1.00 | 0.74 |
| Houston | 23.77 | 0.91 | 0.45 |
| Louisiana State | 21.93 | 0.81 | 0.62 |
| Texas Tech | 21.48 | 0.99 | 0.61 |
| Iowa | 20.31 | 0.90 | 0.60 |
| Connecticut | 20.63 | 0.93 | 0.55 |
| Alabama | 20.35 | 0.73 | 0.33 |
| Texas | 20.84 | 0.70 | 0.26 |
| Ohio State | 20.59 | 0.75 | 0.26 |
| Arkansas | 19.56 | 0.90 | 0.43 |
| Michigan State | 18.99 | 0.75 | 0.21 |
| Providence | 17.48 | 0.87 | 0.36 |
| Michigan | 18.07 | 0.62 | 0.18 |
| Wisconsin | 16.13 | 0.93 | 0.29 |
| North Carolina | 17.16 | 0.58 | 0.12 |
| Southern California | 16.09 | 0.56 | 0.12 |
| Memphis | 16.29 | 0.57 | 0.07 |
| Seton Hall | 16.10 | 0.63 | 0.08 |
| Virginia Tech | 16.55 | 0.30 | 0.06 |
| Indiana | 16.65 | 0.58 | 0.11 |
| San Francisco | 15.74 | 0.69 | 0.13 |
| San Diego State | 15.02 | 0.58 | 0.08 |
| Colorado State | 15.50 | 0.38 | 0.08 |
| Miami | 14.91 | 0.44 | 0.08 |
| Iowa State | 14.52 | 0.19 | 0.09 |
| Notre Dame | 14.96 | 0.27 | 0.06 |
| Davidson | 13.36 | 0.25 | 0.03 |
| Marquette | 15.37 | 0.42 | 0.06 |
| Saint Mary's | 15.04 | 0.42 | 0.06 |
| Creighton | 13.39 | 0.42 | 0.04 |
| Boise State | 14.83 | 0.43 | 0.04 |
| TCU | 13.27 | 0.37 | 0.03 |
| Loyola | 14.92 | 0.25 | 0.04 |
| Georgia State | -3.02 | 0.00 | 0.00 |
| New Mexico State | 7.97 | 0.07 | 0.01 |
| Vermont | 8.49 | 0.10 | 0.01 |
| Montana State | 1.68 | 0.01 | 0.00 |
| Cal State Fullerton | 0.83 | 0.00 | 0.00 |
| Norfolk State | -10.36 | 0.00 | 0.00 |
| Akron | -0.03 | 0.00 | 0.00 |
| Yale | 1.66 | 0.00 | 0.00 |
| Murray State | 11.42 | 0.31 | 0.03 |
| Saint Peter's | 2.60 | 0.01 | 0.00 |
| Wright State | -1.88 | 0.00 | 0.00 |
| UAB | 12.38 | 0.09 | 0.01 |
| Chattanooga | 5.91 | 0.02 | 0.00 |
| Longwood | -1.46 | 0.00 | 0.00 |
| Delaware | 1.35 | 0.00 | 0.00 |
| Texas Southern | -2.34 | 0.00 | 0.00 |
| Richmond | 9.32 | 0.10 | 0.02 |
| South Dakota State | 7.85 | 0.13 | 0.01 |
| Colgate | 3.55 | 0.07 | 0.00 |
| Jacksonville State | 3.01 | 0.01 | 0.00 |
The columns R32 and S16 indicate the probability of a team advancing to the Round of 32 and Sweet 16, respectively. The table containing the probabilities for the Elite 8, Final Four, and championship game is below.
| Team | E8 | F4 | Final | Title |
|---|---|---|---|---|
| Gonzaga | 0.68 | 0.45 | 0.28 | 0.17 |
| Arizona | 0.53 | 0.34 | 0.22 | 0.13 |
| Kansas | 0.65 | 0.41 | 0.21 | 0.11 |
| Baylor | 0.47 | 0.26 | 0.13 | 0.07 |
| Duke | 0.50 | 0.25 | 0.13 | 0.06 |
| UCLA | 0.45 | 0.24 | 0.12 | 0.06 |
| Kentucky | 0.45 | 0.23 | 0.11 | 0.05 |
| Auburn | 0.51 | 0.26 | 0.12 | 0.05 |
| Villanova | 0.43 | 0.19 | 0.11 | 0.05 |
| Purdue | 0.39 | 0.20 | 0.10 | 0.05 |
| Illinois | 0.25 | 0.14 | 0.08 | 0.04 |
| Tennessee | 0.37 | 0.15 | 0.07 | 0.03 |
| Houston | 0.20 | 0.11 | 0.06 | 0.03 |
| Louisiana State | 0.32 | 0.15 | 0.06 | 0.02 |
| Texas Tech | 0.27 | 0.10 | 0.04 | 0.02 |
| Iowa | 0.22 | 0.10 | 0.03 | 0.01 |
| Connecticut | 0.17 | 0.07 | 0.03 | 0.01 |
| Alabama | 0.13 | 0.04 | 0.02 | 0.01 |
| Texas | 0.11 | 0.04 | 0.02 | 0.01 |
| Ohio State | 0.12 | 0.04 | 0.02 | 0.01 |
| Arkansas | 0.12 | 0.04 | 0.01 | 0.00 |
| Michigan State | 0.09 | 0.03 | 0.01 | 0.00 |
| Providence | 0.09 | 0.03 | 0.01 | 0.00 |
| Michigan | 0.05 | 0.01 | 0.00 | 0.00 |
| Wisconsin | 0.08 | 0.02 | 0.00 | 0.00 |
| North Carolina | 0.03 | 0.01 | 0.00 | 0.00 |
| Southern California | 0.04 | 0.01 | 0.00 | 0.00 |
| Memphis | 0.02 | 0.01 | 0.00 | 0.00 |
| Seton Hall | 0.02 | 0.00 | 0.00 | 0.00 |
| Virginia Tech | 0.02 | 0.00 | 0.00 | 0.00 |
| Indiana | 0.03 | 0.01 | 0.00 | 0.00 |
| San Francisco | 0.03 | 0.01 | 0.00 | 0.00 |
| San Diego State | 0.03 | 0.01 | 0.00 | 0.00 |
| Colorado State | 0.02 | 0.00 | 0.00 | 0.00 |
| Miami | 0.02 | 0.00 | 0.00 | 0.00 |
| Iowa State | 0.02 | 0.00 | 0.00 | 0.00 |
| Notre Dame | 0.01 | 0.00 | 0.00 | 0.00 |
| Davidson | 0.01 | 0.00 | 0.00 | 0.00 |
| Marquette | 0.01 | 0.00 | 0.00 | 0.00 |
| Saint Mary's | 0.01 | 0.00 | 0.00 | 0.00 |
| Creighton | 0.01 | 0.00 | 0.00 | 0.00 |
| Boise State | 0.01 | 0.00 | 0.00 | 0.00 |
| TCU | 0.00 | 0.00 | 0.00 | 0.00 |
| Loyola | 0.01 | 0.00 | 0.00 | 0.00 |
| Georgia State | 0.00 | 0.00 | 0.00 | 0.00 |
| New Mexico State | 0.00 | 0.00 | 0.00 | 0.00 |
| Vermont | 0.00 | 0.00 | 0.00 | 0.00 |
| Montana State | 0.00 | 0.00 | 0.00 | 0.00 |
| Cal State Fullerton | 0.00 | 0.00 | 0.00 | 0.00 |
| Norfolk State | 0.00 | 0.00 | 0.00 | 0.00 |
| Akron | 0.00 | 0.00 | 0.00 | 0.00 |
| Yale | 0.00 | 0.00 | 0.00 | 0.00 |
| Murray State | 0.00 | 0.00 | 0.00 | 0.00 |
| Saint Peter's | 0.00 | 0.00 | 0.00 | 0.00 |
| Wright State | 0.00 | 0.00 | 0.00 | 0.00 |
| UAB | 0.00 | 0.00 | 0.00 | 0.00 |
| Chattanooga | 0.00 | 0.00 | 0.00 | 0.00 |
| Longwood | 0.00 | 0.00 | 0.00 | 0.00 |
| Delaware | 0.00 | 0.00 | 0.00 | 0.00 |
| Texas Southern | 0.00 | 0.00 | 0.00 | 0.00 |
| Richmond | 0.00 | 0.00 | 0.00 | 0.00 |
| South Dakota State | 0.00 | 0.00 | 0.00 | 0.00 |
| Colgate | 0.00 | 0.00 | 0.00 | 0.00 |
| Jacksonville State | 0.00 | 0.00 | 0.00 | 0.00 |
Let’s start with the title favorites. Just like last year, my model thinks Gonzaga is the favorite. We give them roughly a 45% chance to make the Final Four and about a 17% chance to win the whole thing. Even crazier: if you take the Vegas data out and leave just the raw game scores, the model moves even further in Gonzaga’s direction. That is, the numbers say the Zags are much better than the betters think they are. Re-running the model without the Vegas data gives the Zags a 34% chance to win the Championship. Those would be the highest March Madness Odds I ever remember seeing. I think the 17% chance is closer to reality.
The other three one seeds round out the top 3 title favorites. This year the one seeds only total up to about 46% of the titles; it is more likely a 2+ seed wins this year than a 1 seed does. Perhaps the most surprising of the remaining teams is fourth-seeded UCLA. They have better title odds than anyone other than a 1 seed or Duke.
The only other notable thing I see is that Wisconsin is WAY down the ratings for a 3 seed. I’m not the first person to question Wisconsin’s #3 seed. In fact, if you sort all the teams in the first table by their raw rating, you’ll see that there are 3 teams with seeds 11 or greater who are “better” teams than Wisconsin. That’s right: Michigan, Indiana, and Virginia Tech are all seen as better NCAAM teams than #3 seed Wisconsin. I can’t explain it, but the numbers are the numbers.
I’m going to group the teams a little bit now to talk about them more in depth.
March Madness Upsets: Seeds 13-16
There isn’t a ton to say here. Seeds 13 or greater averaged 0.35 wins per simulation. To put that another way, roughly every one in three simulations a seed in the 13-16 range wins. Not great.
South Dakota State leads the way, advancing past the Friars about one in eight times. Much like Wisconsin, Providence has been largely hated by the computers this year. Next is the Vermont Catamounts who beat Musselman and the Hogs about one in ten times. Even better – Vermont has the 6th slowest pace of all tournament teams. If you must pick an upset in the 13+ range, I think they’re the ones to go with.. Finally, Colgate beats Wisconsin – who I already discussed – about 7% of the time.
March Madness Upsets: 12 Seeds
Now things get really fun. Common wisdom says at least one 12 seed wins more often than not. The same is true this year: a 12 seed advanced to at least the second round 67% of the time. Indiana is actually favored to win over St. Mary’s. The other three 12 seeds are in the 5-10% chance of winning range.
March Madness Odds: 11 seeds
An 11-seed winning is not much of an upset, but they are still noteworthy. Our models think that on average 1.4 11-seeds will advance in the 2022 March Madness tournament. The best of the 11 seeds is Michigan. We’re giving Michigan a 62% chance to beat Colorado State, nearly 2:1 odds
.
Close behind is Virginia Tech: the numbers say the Hoakies beat Texas 31% of the time. Even better, we’ve determined in the past that teams which play at a slow pace are more likely to upset their higher-seeded opponents. Virginia Tech has the slowest pace of all tournament teams, the 10th slowest in all of division one. These two things together make me think Virginia Tech has close to ‘coin-flip’ odds against Texas. Funny enough, this is the second year in a row I’ve put Texas on upset alert in the first round.
The other 11 seeds – Notre Dame/Rutgers and Iowa State have about a one in four or five chance to advance. Not great March Madness Odds, but I wouldn’t be surprised if one or the other advances.
March Madness Odds: 10 Seeds
The 10 seed matchups are always difficult to pick. Not only do they all have a shot, they also all have a chance in the second round. Thus, picking the right 10 seed upsets is important for a good bracket. Our model finds an average of 1.7 10 seeds advance to the second round. The best 10 seed to pick is the San Francisco Dons. Not only are they the highest rated of the 10 seeds, Murray State is the worst of the 7 seeds. This is a dangerous combination leading to the Dons having nearly a 70% chance of advancing. That is fairly astronomical for a so-called underdog.
The Miami Hurricanes have the second best odds of advancing, about 45%. Davidson and Loyola round out the rest of the 10 seeds with about a one in four shot each.
March Madness Odds: 8 & 9 Seeds
The 8 and 9 Seeds are notoriously difficult to predict. This year this is especially the case. Only one team (Seton Hall) in this group has more than a 59% chance of winning their first round matchup. Honestly, truly, these picks could not be more inconsequential in how your bracket performs. The 8 and 9 seeds are usually destined to get handled easily in the second round, so correctly picking these games is typically a one point difference. I’ll probably flip a coin for each of these matchups except for Seton Hall.
Only one team in this group has above an 8% chance at upsetting the one seed in the following round. The Tar Heels end up beating Baylor in about one in eight of our simulations. About half the time a one seed loses in the second round, but because it is so difficult to pick whether it is the 8 or the 9 that does the upsetting, I don’t recommend picking any upsets of that form in your bracket.
Cinderella Watch
Instead of just looking at the first round matchups, I want to look forward to the Sweet 16 and the Elite 8 to see which double digit seed has the best chance at making a deep run. While none of these teams have particularly good odds, the probability of at least one of these teams advancing to the Sweet 16 is only about 60%. This feels much lower than a typical tournament. I might not pick any double digit deep runs in my bracket.
We know that in the 1-4 seed range, the 1 seeds and UCLA are going to have the best tournament. What about teams in the 5-9 range? Overall in this range, the three best teams are Houston, LSU, and Texas. However, the best teams won’t always have the best tournaments because they might have a more difficult route. I will also point out that LSU is a very curious case because they no longer have the coach who got them to this point in the season.
In the 5-9 seed range, there are three teams that have a greater than 50% chance to make it to the Sweet 16 (requiring at least one upset in the process). These are 6th seeded LSU, and the pair of 5-seeds Iowa and UConn.
One other curiosity is the second round matchup between Houston and Illinois. The computers have been huge fans of the Cougars all year.. The raw numbers have Houston as the 10th best team from a ratings perspective. This would make their 5 seed an egregious affront to their quality. Unfortunately for them, their likely 2nd round matchup, Illinois, is also criminally under-seeded. My model has Illinois as the fifth best team overall. Unfortunately these two juggernauts will go head-to-head in the second round.
The winner of the Houston-Illinois game makes the Elite 8 about 45% of the time and the Final 4 about 25% of the time. This is ridiculously high for the fourth and fifth seed in a region.
Final Thoughts
To summarize, the upsets I am pulling for this year are:
-
Relatively High Confidence:
- San Francisco over Murray State
- Indiana over St. Mary’s
- Michigan over Colorado State
-
Toss-Ups
- Virginia Tech over Texas
- Miami (FL) over USC
In addition to this, we think the following teams have the best March Madness odds to win a Championship
- 1 Seed: Gonzaga (16.7%)
- 2 Seed: Duke (6.4%)
- 3 Seed: Purdue (5%)
- 4 Seed: UCLA (6.1%) [Very high for a 4 Seed!]
- 5 Seed: Houston (3.1%)
One Reply to “Men’s 2022 March Madness Odds and Upset Predictions”
Comments are closed.