For as long as I can remember, I have loved the NCAA Basketball Tournament. I love the bracket. I love the underdogs. I love One Shining Moment. I even love the CBS theme music.
As a kid I filled out hand-drawn brackets and scoured the morning newspaper for results of late night games. As I got older, I started tracking scores using a increasing complex set of spreadsheets. Over time, as my analysis became more sophisticated, I began to notice certain patterns to the Madness
I have found that I can use modern analytics and computational tools to gain a better understanding of the tournament itself and perhaps even extract some hints as to how the tournament might play out. Last year, I used this analysis to correctly predict that No. 4 seed UConn win the National Title in addition to other notable upsets.
There is no foolproof way to dominate your office pool, but it is possible to spot upsets that are more likely than others and teams that are likely to go on a run or flame out early. While we wait for the main set of games to start on Thursday, "Dr. Green and White" is here to help you fill out your 2024 bracket.
Before we did into the current bracket in detail, let's start with an overview of my methods and general trends to expect in 2024.
Methodology Overview
The foundation of my methodology is an observation that I made several years ago that boils down to this:
When it comes to NCAA Tournament upsets, the behavior is exactly the same as in regular season games. The odds are largely predictable based on Vegas points spreads and by tools that can predict point spreads, such as Kenpom efficiency margin data.
All of my analysis of college basketball odds is based on this same premise. Kenpom efficiency data can be used to assign probabilities to any arbitrary basketball match-up. Knowing this, the full season and any tournament can be mathematically modeled and its odds can be calculated.
My favorite plot to highlight this fact is shown below.
 |
Figure 1: Correlation between NCAA Tournament upsets and the odds predicted using Kenpom efficiency data.
|
|
This figure compares the winning percentage for the higher seeds in the NCAA Tournament to the odds expected based on the average point spread of games with that seed combination. The figure shows that data for all seed combinations that have occurred at least 40 times.
Figure 1 tells us why No. 16 seeds have won two times over the past 38 years (1.3% of the time). It is because on average No. 16 seeds are 22.5-point underdogs and 22.5-point underdogs win straight up 1.4% of the time whether the game in played in March or in November.
There are a few notable deviations from this correlation. For example, No. 10 seeds have surprisingly good luck against No. 2 seeds and No. 9 and No. 5 seeds do not upset No. 1 seeds in the second round or in the Sweet 16 as often as expected. (This could be a discussion topic if No. 9 Michigan State survives in the first round against No. 8 Mississippi State on Thursday.) In general, the correlation is very strong.
The Vegas points spreads and the point differentials predicted by Kenpom efficiency margins, also correlate very strongly. Figure 2 below shows how strong this correlation is for the first-round games in the 2024 NCAA Tournament.
 |
Figure 2: Correlation between the Vegas lines and the point differentials predicted using Kenpom efficiency margins for the 2024 NCAA Men's Basketball Tournament.
|
|
Figure 2 gives us confidence that Kenpom efficiencies can be used to model the results of the NCAA Tournament.
2024 Bracket Overview
The lead up to this year's Tournament featured an unprecedented number of upsets in the conference tournaments. Of the 32 total conference tournaments, only 11 were won by the No. 1 seed, nine were won by the No. 2 or No. 3 seed, and the remaining 12 were won by a lower seed.
Will all of this conference madness spill into the national tournament?
I attempted to explore this question by simulating the results of the 2024 tournament 5,000 times and counting the number of upsets that occurred in each round. I then compared these values to simulations of previous tournaments and to the results of the past 21 actual tournaments. The results are shown below in Figure 3.
 |
Figure 3: Number of projected upsets per round of the 2024 NCAA Tournament based on a Monte Carlo Simulation and compared to the historical value and the average of the series of historical simulations.
|
|
Based on the results of this simulation, the answer to the question above is 'no.' In fact, the simulation suggests that the 2024 NCAA Tournament is more likely to see slightly fewer upsets than usual, especially in the first two rounds.
The overall average number of first round upsets is 8.6 (of 32 games) but the simulation result calls for almost a full game less at 7.8. The data also suggest a relatively quiet second round (4.7 upsets compared to 5.1 upsets).
It may be counterintuitive, but it makes sense that the chaos in the conference tournaments will translate into a relatively calm first two round of the NCAA Tournament. This is a manifestation of what I like to call the law of conservation of upsets. Usually it appears to the first and second round of the tournament, but it can also apply here.
A large number of conference tournament or first round upsets results in some of the higher quality teams being weeded out prematurely. As a result, the teams that do survive face weaker opponents in the subsequent first or second round games.
Starting with the Sweet 16 round, the number of projected upsets in 2024 tracks with the simulation averages over the past 20 years. Expect to see between two and three upsets in the second round, between one and two in the Regional Finals and maybe one upset in the three games that make up the Final Four.
Speaking of the Final Four, I am also able to use the results of my simulation to project the distribution of seeds that will advance to the final weekend. Most "experts" will frequently select three or even all four No. 1 seeds to make the Final Four. The reality is quite different.
Figure 4 shows distribution that I obtained in my simulation of the 2024 tournament. This distribution is very consistent with past simulations and the actual distribution of Final Four seeds over the years. Recall that the 2023 Final Four was comprised of a No. 4, two No. 5, and one No. 9 seed.
 |
| Figure 4: Projected distribution of seeds in the 2024 Final Four based on the results of a 5,000 full tournament simulations. |
Simulation and history shows that a "typical" Final Four actually is made up of a No. 1 seed, a No. 1 or a No. 2 seed, a No. 2 or No. 3 seed, and one lower seed. More than two No. 1 seeds have made it to the Final Four only five times since seeding began in 1979.
Compared to past simulations and history, the 2024 Final Four is likely to look very similar to past tournaments. It is quite likely (84% odds) that at least one of the No. 1 seeds will advance to the final weekend. The odds are only 46% that two of the No. 1 seeds make it and they are only 13% that three No. 1 seeds make the Final Four.
The mostly likely scenario is that the second and third highest seeds in the Final Four will be No. 2 seeds. The lowest seed to appear in the 2024 Final Four is most likely a No. 4 seed.
With this background knowledge now in place, the next step is to dig into the brackets, look for upsets, and make some predictions. That will be the subject of part two of this analysis. Stay tuned.
Part 2: Breaking Down the Brackets
Yesterday I introduced my methodology of how I use Kenpom efficiency data to understand why NCAA Tournament upsets happen at the frequency that they do. I used the data for the 2024 Tournament bracket to predict that we are likely to see fewer upsets in the first two round especially. I also made some predictions about the make-up of the Final Four.
In the second and final part of this series, it is time to dig into the four regions in detail. Which teams will advance to Phoenix? Which big upsets will create buzz in the first two rounds, and how far will the Michigan State Spartans advance? Let's jump in.
East Region
Table 1 below summarizes all of the relevant data for the East Region.
 |
Table 1: 2024 NCAA Tournament East Region odds
|
This table gives a lot of information that we will use to make our picks. The left side of the table shows the pre-tournament Kenpom adjusted efficiency margin for each team. The shaded cells on the left side of the table provide a comparison of each team's efficiency relative to the historical average of teams of that seed. This provides a simple way to look at the relative strength or weakness of each team and the bracket as a whole.
The middle of the table shows the odds for each team to advance through each round of the tournament. The teams are sorted not by seed, but by the odds for each team to advance to the Final Four. The red or green shaded cells on the far right are the relative odds for each team to advance compared to historical averages for that seed.
Finally, there is a column labeled "SoD" which stands for "strength of draw." This calculation starts with the odds for a historically average No. 1 seed to advance to the Final Four from any of the 16 positions on this year's bracket. I then compare those odds to the odds that the same historically average No. 1 seed would have to reach the Final Four in a historically average NCAA Tournament bracket.
One clear observation from Table 1 is that the despite UConn being the No. 1 overall seed, the Huskies were placed in the most difficult region in the tournament. The primary indicator of this fact is the that all 16 teams have a notably negative strength of draw. The teams in the East Region have between a four percent and eight percent more difficult draw than teams in a historically average region.
The primary reason for this difficulty is evident from the data in the relative Kenpom efficiency column. The top four seeds and five of the top six seeds are significantly above average compared to past teams of their seed.
All of the top four seeds have over a 10% chance to advance to the Final Four, but the clear outlier is No. 4 Auburn. The Tigers have the second best odds to advance to the Final Four at 21%, which is 11 percentage points higher than a historically average No. 4 seed. One other potential team to watch in the East is above average No. 6 BYU.
South Region
Table 2 below summarizes all of the relevant data for the South Region.
 |
Table 2: 2024 NCAA Tournament South Region odds
|
|
The South Region is the second most difficult region in the 2024 Tournament based on average strength of draw, but it is significantly easier that the East Region. Based on the data in Table 2, No. 1 seed Houston is the clear favorite to advance to the Final Four. The Cougars have a 46% chance to advance to the final weekend, which is by far the best odds of any team in the Tournament.
Unlike the East, No. 2 Marquette, and No. 3 Kentucky are relatively weak. In fact, the team with the second best chance to win the South Region is No. 4 Duke (15%). If Houston can advance past the Sweet 16, their path to the Final Four will likely be clear.
The reason for this speculation is that the teams seeded No. 7 through No. 10 in the South Region are relatively strong. Therefore, there is significant potential for chaos on the bottom half of this bracket.
Specifically, I will be on the lookout for No. 10 Colorado. The Buffalos are above average for their seed and could wreak havoc in the bottom of the bracket, if they can escape from the First Four on Wednesday evening.
Midwest Region
Table 3 below summarizes all of the relevant data for the Midwest Region.
 |
Table 3: 2024 NCAA Tournament Midwest Region odds
|
In the Midwest Region, the colors in the strength of draw column have shifted from mostly red to mostly white and blue, indicating that the Midwest is of average difficultly.
Unlike the East and South Regions, there is no obvious lower-seeded dark horse contender. No. 1 Purdue, No. 2 Tennessee, and No. 3 Creighton are all slightly above average for their seed. In fact, the only teams that are more than a point per 100 possessions above average are No. 5 Gonzaga (+1.43) and No. 6 TCU (+1.61).
The Midwest Region does contain a lot of below average double-digit seeds, however, which reduces the odds of most first round upsets. There are two exceptions, however. No. 4 Kansas and No. 6 South Carolina. Don't expect either of these teams to hang around very long.
Based on the raw numbers, No. 1 Purdue (39%) and No. 2 Tennessee (24%) have the best odds to advance to the Final Four. But betting on either of these teams seems fool hearty for reasons that Kenpom efficiency margins do not capture.
Despite Purdue's dominance of the Big Ten in back-to-back years, the Boilermakers have not hung a Final Four banner since 1980 and have lost to double-digit seeds in each of the last three Tournaments. Meanwhile, Tennessee head coach Rich Barnes owns the worst score in the Performance Against Seed Expectation (PASE) NCAA Tournament metric in history.
Betting on either team to win the Midwest would be betting against history. I will have to see it to believe it.
West Region
Table 4 below summarizes all of the relevant data for the Midwest Region.
 |
Table 4: 2024 NCAA Tournament Midwest Region odds
|
The prevalence of a darker blue color in the strength of draw column indicates that the West Region is the easiest of the four regions in the 2024 NCAA Tournament. Unlike the other three regions, the very weak No. 1 seed, North Carolina, does not have the best odds to make the Final Four. That honor falls to slightly above average No. 2 Arizona.
The teams seeded between No. 3 Baylor and No. 7 Dayton are all historically below average which the exception of slightly above average No. 4 Alabama. But there are two potentially interesting dark horses lower in the bracket. No. 9 Michigan State (4.4%) and No. 11 New Mexico (3.8%) own the sixth and seventh best odds to win the region.
The majority of the double-digit seeds are also below average.
From a Michigan State point of view, this is clearly good news. In the weeks leading up to the Tournament, many experts expected that the Spartans would land somewhere between a No. 7 and No. 10 seed. Many speculated that avoiding a No. 1 seed by landing on a No. 7 or No. 10 seed would be best.
However, the Spartans' placement as the No. 9 seed in weakest tournament region is actually the best possible draw that Michigan State could have received. The Spartans have the easiest strength of draw (+1.3%) of any team seeded between No. 7 and No. 10 in the 2024 Tournament.
The Spartans are currently a one-point favorite over No. 8 Mississippi State in the first round game this Thursday. If Michigan State advances to face No. 1 North Carolina in the second round, the Tar Heels are only predicted to be favored by between three and three-and-a-half points. This spread is closer to a No. 5/12 seed match up than a typical No. 1/9 seed match up where the spread is closer to 10 points
If Michigan State is able to advance to the second weekend, potential match-ups with No. 4 Alabama (-2) and No. 2 Arizona (-4.5) would be challenging, but not impossible. At least according to the math, a classic Tom Izzo March run is very much a possibility.
First Round Upset Analysis
Tables 1-4 provide a great snapshot of each region, but in any tournament it is the individual match-ups that ultimately matter. Which upsets are the most likely? Figure 1 below helps to answer that question.
 |
Figure 1: Odds for the higher seeded teams to win for each seed pairing, relative to the historical odds (shown in blue) for all first round games.
|
In my opinion, the data in Figure 1 is the most useful when filling out my bracket. The blue line in both panels shows the historical odds for the higher-seeded team to win each match-up. The labeled data points show the actual odds, based on Kempon efficiency data (which accurately mirrors the Vegas spread).
If a data point falls below the line, an upset is more likely than average. If a data point is above the line, an upset is less likely than average. The farther a data point is from the line, the more likely the upset.
One of the key conclusions from part one of this series was that between seven and eight upsets are most likely in the first round this year. Figure 1 shows that the most like candidates are in the contests where the better seed is actually the underdog in Las Vegas. This year there are four first round games that fall into this category:
-No. 11 New Mexico (-2) over No. 6 Clemson
-No. 9 TCU (-4.5) over No. 8 Utah State
-No. 9 Michigan State (-1) over No. 8 Mississippi State
-No. 10 Nevada (-1.5) over No. 7 Dayton
Earlier in the week, No. 11 Oregon was favored over No. 6 South Carolina, but the line has since flipped to the Gamecocks side. My analysis still likes the Ducks in the upset.
My analysis also suggests that the winner of Wednesday night's play-in game between No. 10 Colorado (-2.5) and No. 10 Boise State is also a good candidate to upset No. 7 Florida.
As for other, bigger upsets, the left panel suggests that any upset of a No. 1, No. 2 or No. 3 seed is less likely than usual this year. Most of the No. 4 seeds are also safe as well. The only "big" upset that my analysis suggests is:
-No. 13 Samford over No. 4 Kansas.
The No. 12 seed upset over the No. 5 seed is always a popular office pool pick, but this year there is no obvious candidates. If one feels inclined to pick a No. 12 seed to advance, Grand Canyon over No. 5 St. Mary's is the best option.
Second Round Upsets
Moving onto the second round and beyond, Figure 2 shows a similar analysis which for consistency assumes that the favorites all win in the first round.
 |
Figure 2: Odds for the higher seeded teams to win for each seed pairing, relative to the historical odds (shown in blue) for all second round games (left panel) and the Sweet 16 and Regional Final Rounds (right panel)
|
In part one of this analysis, my simulation suggested that a total of four to five upsets are most likely in the second round. According to Figure 2 and my analysis, the following upsets are the most likely:
-No. 6 Texas Tech over No. 3 Kentucky
-No. 10 Colorado over No. 2 Marquette
-No. 9 Michigan State over No. 1 North Carolina
-No. 11 New Mexico over No. 3 Baylor
Note that my analysis also favors No. 5 Gonzaga to upset No. 4 Kansas if the Jayhawks survive the first round. Similarly, my analysis would favor No. 7 Florida upsetting No. 2 Marquette.
There is no obvious option for additional second round upsets. My computer has its electronic eye on No. 6 BYU over No. 3 Illinois, but I am not so sure. Personally, for the historical reasons mentioned above, I am tempted instead to take he following upset to blow up the Midwest Region.
-No. 9 TCU over No. 1 Purdue
(Sorry, Boiler Makers.)
Sweet 16 and Beyond
If I make assumptions above, I project that the tournament would play out as follows from the Sweet 16, using a combination of math and intuition.
In the East Region, I have the first two rounds going as chalk leaving No. 1 UConn to face No. 4 Auburn and No. 2 Iowa State facing No. 3 Illinois. The right panel of Figure 2 suggests that Auburn is a good candidate to upset the defending National Champions. I have the Tigers meeting and defeating No. 2 Iowa State in the Regional Final.
In the South Region, No. 1 Houston faces a tough challenge against No. 4 Duke in the Sweet 16, but I see the Cougars surviving and advancing in that match-up as the last remaining No. 1 seed. The other half of the bracket features No. 6 Texas Tech versus No. 10 Colorado. I project that No. 1 Houston would beat No. 6 Texas Tech in an all-Big 12 Regional Final to advance to the Final Four.
In the Midwest Region, my analysis results in Sweet 16 match-ups featuring No. 5 Gonzaga versus No. 9 TCU and No. 3 Creighton versus No. 2 Tennessee. Even if Purdue were to survive into the second weekend, I would pick No. 3 Creighton and No. 5 Gonzaga to advance to the Regional Final, where I have the Blue Jays advancing.
In the West Region, my Sweet 16 match-ups feature No. 4 Alabama versus No. 9 Michigan State and No. 2 Arizona versus No. 11 New Mexico. In this case, my computer likes the Spartans to continue their March run, but I do have the season coming to an end in the Regional Final against the Arizona Wildcats.
In summary, I have three Sweet 16 upsets and one upset in the Regional Finals:
-No. 4 Auburn over No. 1 UConn
-No. 3 Creighton over No. 2 Tennessee
-No. 9 Michigan State over No. 4 Alabama
-No. 4 Auburn over No. 2 Iowa State
This results in a Final Four of No. 4 Auburn versus No. 2 Arizona and No. 1 Houston versus No. 3 Creighton. As a general rule, once the teams reach the Final Four, the projected favorites usually win.
Despite the fact that Auburn is just a No. 4 seed, the Tigers are ranked No. 4 in Kenpom. For this reason I have No. 4 Auburn facing and losing to No. 1 Houston in the National Championship Game.
As a final note to this analysis, history shows that 18 of the past 21 champions have entered the NCAA Tournament ranked in the top six of Kenpom efficiency margin. If this trend holds in 2024, one of the following teams will win the National Title:
-UConn (No. 1 seed in the East)
-Houston (No. 1 seed in the South)
-Purdue (No. 1 seed in the Midwest
-Auburn (No. 4 seed in the East)
-Iowa State (No. 2 seed in the East)
-Arizona (No. seed in the West)
It is easy to see more clearly why the East is, by far, the most difficult region. My analysis has three of those six teams advancing to the Final Four, two of those teams (UConn and Iowa State) losing to one of the other teams in the group (Auburn), and the chronic choker of the group (Purdue) doing what they do best.
That all feels right.
That is all the advice that I have to share today. Enjoy the Madness.
Comments
Post a Comment