March Madness 25 - ClevAnalytics

2025 MARCH MADNESS BRACKET PREVIEW

March Madness is upon us once again and for many this is the best sporting event of the year. Even for those who don’t care much about college basketball or even have not paid any attention during the season, there is a likelihood that you will be filling out a bracket. First and foremost, this is freaking hard and I hate to break it to everyone out there but there is a very high likelihood that you won’t win your pool. Even the people who have watched and studied the game all season, they struggle as much or even more to get things right. Personally, I love college basketball, but I have paid less attention to the sport this year than any other in my lifetime. Priorities get in the way and so I am less in tune with these team than ever before. Honestly outside of Cooper Flagg I would be hard pressed to name many other players. If you are looking for a deep dive into these teams, players, etc there are so many good resources out there who do a good job of that and will tweet a few this week.

With that being said, I have been filling out brackets since I was probably in 5^th grade. And I love it. I do believe that are some basic tenants to follow if you want to give yourself a decent chance at winning or placing in your pool (notice I said give yourself a chance, there are no guarantees with any of this). For maybe the 20-25% of the people out there these are things you already know and understand so it’s likely not going to be anything new or additive. But for the vast majority of everyday people who don’t follow the sport at all, and may not even know where to begin, the goal is to at least give a decent foundation to help guide you through bracket picks and maybe give you the tiniest of edges over the rest of your pool.

Below, I will take a look at some different categories of teams and the profiles of teams that historically have won the title, been knocked out early or have been the teams to upset multiple higher seeds and make deep runs. This is an inexact science to say the least and history doesn’t always repeat itself. This is a wide-open tournament so anything is possible! But I wanted to use this platform to potentially help you along the way. Feel free to follow some, all or completely ignore the historical profiles below. Your choice, I am just here to lay them out, in case you need the help.

(*KENPOM DATA GOES BACK TO 2001)

My Process

Truth be told, like 95% of people out there, I just don’t have the time to follow college basketball for a full season so I have really just dug deep over recent weeks. I don’t have the knowledge to handicap all 64+ teams. And with that, the best route for filling out a bracket is to use tournament history and some basic tenants, to give you the best chance to compete in your pools and has rewarded me with some success in the past. I like to use a more top down approach on the front end with some bottoms up team matchups analysis at the back end. I use somewhat of a decision tree approach in terms of bucketing these teams into different categories where historically they have proven to spit out a higher/lower likelihood of advancing. I essentially can work backwards to some degree and understand the group of teams that qualify for me in terms of Title winners, Final 4 participants and also try and exclude those higher seeded teams that have shown a propensity for early upsets.

Knowing Your Pool- Your bracket strategy should start with understanding the size of your pool that you are competing in and the level of knowledge of other participants, if possible. You should also understand the rules as some award more points based on the round and/or potentially based on the seed who wins. Based on the variation of the rules, if your pool awards risk taking then take some more risks at the appropriate times.

- The larger the pool (~100+ entrants) the more risks you should take although you should not be reckless. Obviously as you get larger and larger with the number of entrants, the more likely you will need to take some risks. Measured upsets are important but you really have to nail many of your final 4 teams and the winner to have a real chance. Its ok to make a couple of upset predictions among teams seeded 10-13 but only fade those favored teams that you don’t think will advance far anyway. If you do call for the upset, only advance them one round. A bracket full of 12 and 13 seeds in the sweet 16 and Elite 8 has virtually no shot of winning. You can utilize a resource like Yahoo or ESPNs Bracket Challenge Public Pick % ESPN Men’s Tournament Challenge – People’s Bracket that shows what the general public is picking in each round. Along the way of filling out your bracket, try and find good leverage spots where the public is over/undervaluing certain teams and matchups and go against the grain in those spots. A perfect example is right now Memphis as a 5 seed is being picked to win in round one in 64% of ESPN brackets over #12 seed Colorado St. The problem is that CSU is actually favored over Memphis by 2.5 points. This spread equates to ~58% win probability for CSU and a +12% win probability leverage spot to take CSU. That is one example and there are others where you can select specific upsets that helps you gain good leverage over your competition.

- In smaller pools (<50-100 entrants), you really don’t have to make too many big risks with upsets. Fading popular favorites is really the best route but in general you don’t need to do too much more than backing the teams with the highest win probability throughout or using kenpom ratings as a proxy for win probability. In both pool size instances, pay attention to the spreads of the round one games and if a double digit seed is favored over the higher seeded team then that could be the type of seed “upset” to pick. This really isn’t an upset based on win probability but the average contestant might not have in their bracket. Also compare the true implied betting odds for each round and to win the tournament for teams versus the public bracket challenge data posted on ESPN, Yahoo, etc. If you see a wide gap between actual win probability odds and what the general public is selecting in pools, that gives you a good proxy for your pool and potentially selecting the undervalued team and lead to differentiation. The easiest way to start is just advancing every team seeded 1-4 thru the the 2nd round and every 1 seed to the Elite 8. Try and include at least a couple of the most likely Championship teams (using my table below) in the Final 4. If you want to take a team to reach the Final 4 that is not one of the top teams, try and find a team that is a top 4 seed but there is a gap between their title odds and their seeding. Two years ago, UCONN was a 4 seed but they were #4 in the Kenpom rankings and their implied betting odds to win it all (5.9%) showed the 2nd widest gap vs what the public was picking on ESPN (1.8%). That led to me recommending them as one of the top teams to select to win the title (23 Madness Bracket Preview – ClevAnalytics). The goal with any of these brackets is not to pick the perfect bracket (that will lead to a likely poor bracket) but the goal is to just beat the competition.

- If you are in a really sharp pool that is not very large, you likely will have to be creative. Everyone else has the same data and will envoke a similar strategy to the basics I am laying out here. So there has to be some differentiator. At that point you may have to double back and actually go with the more popular title teams or just fade some analytics darlings to separate a bit from the expert masses.

- If your pool awards more points based on the seed of the team that wins then evaluate the point spreads and obviously try and pick a couple of the double digit seeds where the spread is tight. For example, VCU is an 11 seed but is only a 3/3.5 point underdog vs BYU. Unless you feel strongly about BYU advancing far, taking VCU there makes a lot of sense. Same with UC SD as a 12 seed and only a 3.5 point dog to Michigan.

Past Data– Using historical data, you can try and narrow down the field to the handful of teams with the most realistic shot to win. Same with fading those higher ranked teams (seeds 1-7) who have historically profiled as teams that don’t typically advance far in the tournament. Again, nothing is perfect and even 20+ years of results is a fairly small sample size. There will be outliers but most likely very few if anyone else in your pool will have those outliers winning it all so you don’t have to be perfect. (see below for much more detail)

Seed Analysis/Notes- Understanding what tournament history says about how often and how far different seeded teams advance and using that baseline to help you project your bracket. It’s just a nice guide and absolutely doesn’t mean this tournament will fall exactly as it has throughout history. But just some notes to keep in mind:

The non 16 seed play in game has existed for the last 13 seasons. In 12 of those tournaments, at least one has advanced to the round of 32. UCLA in 2021 and VCU in 2011 made it all the way to the Final 4 and three others reached the Sweet 16. Last year, we saw Colorado upset Florida in round one. Don’t be afraid to advance one of these teams.

A top 4 seed has lost in the opening round in 15 of the last 16 tournaments. Last year two teams (#3 Kentucky and # 4 Auburn) lost in round one, both as double digit favorites. Five teams seeded 4th or better have lost in round one to a double digit seed in the last two years. If you are in a very small pool only pick an upset here in round one if you don’t think that team is advancing far anyways.

Last year was a pretty chalky tournament in general but with #11 seed NC State Wolfpack advancing to the Final 4, we have now seen 16 straight tournaments where a double digit seeed team has made it to the Sweet 16.

Matchup Model– After using the historical data and current odds as a backdrop, I use season long data to evaluate matchups and try and best model the outcome of game to game selections. If you are in a smaller pool you likely don’t even need this, you can just advance the higher ranked kenpom team (proxy for favored teams) throughout with a couple of decisions in close games.

2025 Tournament Team Data (co/Kenpom.com)

Title Contenders

Lets take a historical look at the profile of an NCAA tournament champion. Below are the pre tournament Kenpom rankings across raw points per possession, adjusted offense, raw points allowed per possession, adjusted defensive efficiency and total efficiency margin since 2001.

No team has won the NCAA tournament in the 23 years of available Kenpom data, without entering the tournament in the top 25 of Kenpom’s overall ranking.

In all but three cases (2003 Syracuse, 2011 Uconn and 2014 Uconn), every champion has been a top 6 overall team entering the tournament. In all three cases where the Champion did not rank in the top 6 overall by Kenpom, those teams did have at least have one victory during the regular season over a Kenpom top 5 team. So at least they showed the ability to beat an elite team during the season.

Only one time (2014 Uconn) has a team with an adjusted offensive efficiency of worse than 21st won the title. 18/23 champions have entered as a top 10 offense by adjusted offensive efficiency and the median rank entering the tournament of the last 23 title winnners has been 3rd overall. Only once (again 2014 Uconn) has a title winner entered with a raw points per possesion ranking of worse than 45th.

Defense looks to be a bit looser when it comes to the high level filters. The 2021 Baylor Bears (44th) and 2015 Duke Blue Devils (37th) are the only teams to win the title with an adjusted defense that ranked outside the top 30. However, both of those teams possessed elite offensive ratings (top 3 in adjusted off efficiency). So if you are going to look at a team with defensive metrics outside of the top 30, they better be elite on offense to make up for that deficiency. The median title winner defensive efficiency ranking is 11th overall.

In the last decade, as basketball in general has moved much more into the pace and space, we have seen some of these rankings evolve into the best offenses overpowering the elite defenses. Since 2015, no title winner has entered with an adjusted offensive efficiency worst than 15th and all but one (2016 Villanova) has entered top 7 in offense. At the same time, we have only had two title winners (2016 Villanova and 2019 Virginia) enter as a top 20 adjusted defensive team. In fact, four of the last nine winners have entered 25th or worse on defense. The two biggest outliers on defense in the last 23 years of Kenpom data (21 Baylor, 15 Duke) have come in the last 10 seasons. Eery single title winner has entered the tournament inside the top 6 in overall Kenpom ratings. An elite offense has evolved into the better predictor of success from a title perspecitve as long as they have some ability to get stops. If you are going to lean into one specific category to pick your winner among the top teams, leaning more toward offense likely is the smartest bet.

According to Kenpom, since 1985, there have been 39 teams as a #1 or #2 seed but were not ranked in the preseason AP poll. Not only have none won the Championship but none of those 39 teams even made the Final 4. Priors and wisdom of the crowd entering the season seems to hold some water as a predictor for a deep tournament run. I chalk this mainly up to school pedigree and talent levels reverting back to the mean if a team overperformed throughout a regular season. They can be really good regular season teams but when it comes to winning four+ straight in a row versus the top competition, usually coaching and talent supercedes everything else. None have exceeded their seeding expectations. A #1 seed is expected to make the Final 4 whereas a #2 seed is expected to make the Elite 8. This season Michigan State and St. Johns qualify as #2 seeds.

When you include the runners-up as well, no team has made the NCAA tournament championship game with an adjusted offensive rating worse than 62^nd, a defensive rating of worse than 69th or with a total net efficiency margin worse than 44^th. The 2011 Butler Bulldogs were a big outlier, ranking 69^th in adjusted defense and 44^th in overall net rating. If you exclude that team, no runner up or title winner entered the tournament worse than 29^th in overall net rating.

Who are the likeliest title contenders based on historical criteria? To keep it fairly simple, I am using a filter for the teams that rank in the Kenpom top 25 overall, are top 20 in adjusted offense (also top 50 in raw pts/poss), top 50 in adjusted defense and if not in the top 6 overall in Kenpom ratings has at least one win over a top five Kenpom team this season. In addition, I eliminated two teams that were not ranked in the AP preseason top 25 (MSU and St John’s).

My historical filters has led me to a list of 7 true title contenders. Duke, Florida, Houston, Auburn, Tennessee, Alabama and Texas Tech. These will be the only teams that I will be choosing from in my brackets to win it all.

Even if St. John’s was not excluded for the preseason top 25 filter, they still would be eliminated for me due to the fact that their adjusted offensive ranking of 65th falls well outside of historical champions. It would be eight spots worse than by far the biggest outlier we have seen in modern times from a champion. Michigan St. fits most every category besides the AP poll filter. However, they are outside the top six in total Kenpom rankings and have not beaten a single top five team all season. Again these are just guides and not hard and fast rules so if you feel strongly about including Sparty I would not blame you for advancing them far if you like their matchups.

Other teams that are in the “just missed” category of teams that would be an outlier but not completely out of nowhere if they won would be Kentucky, Missouri, Clemson, Texas A&M and Kansas. At the same time, a couple of these teams (Missouri and Kentucky) have defensive metrics that also put them in the bucket of teams that could also lose very early in the tournament.

Final 4 Contenders:

In the last 23 tournaments, among the 92 Final 4 participants, only one (VCU in 201 at #82), came into the tournament with an overall Kenpom rating worse than 44^th overall.

Among the 92 Final 4 teams, only two (Louisville in 2012 and S Carolina in 2017) made it with an adjusted offensive efficiency ranking worse than 75^th. They ranked 127^th and 149^th, respectively. However, they both were elite on defense, ranking 2^nd and 3^rd in those seasons. If a major outlier on offense is going to make it, they better be among the best defensive teams in the country.

Only five of the 92 Final 4 teams since 2001 ranked worse than 69^th on defense (Marquette in 2003, VCU in 2011, UCLA in 2021, Miami FL in 2023 and Alabama in 2024). VCU is an outlier like no other team, in virtually every way. Marquette, UCLA, Miami FL & Alabama ranked top 30 in adjusted offense and all four fit the top 44 total Kenpom ranks. So again, if you are going to be an outlier on one side of the ball, you better be an elite team on the other side.

88% of Final 4 teams since 2001 came into the tournament ranked inside the Kenpom top 25 overall. 77% ranked inside the top 25 in adjusted offense and 72% defensively. 58% of the 92 Final 4 teams ranked inside the top 25 in both. There will always be outliers but sticking to these rates will be a good benchmark.

Just 9 (9.8%) of the 92 Final 4 teams since 2001 entered the tournament outside of both Kenpom’s top 25 in adjusted offense and defense. All but one of those nine ranked at least top 35 in either offensive or defensive efficiency.

There has not been a Final 4 since 2001 that did not include at least one top 10 overall Kenpom rated team. A top six overall Kenpom rated team has made the Final 4 in 22 of the last 23 tournaments. Interestingly, since 2010, all but one Final 4 has included at least one team ranked 20^th or worse in Kenpom overall ratings so there are almost always at least one team very few expect.

According to Kenpom, since 1985, there have been 39 teams as a #1 or #2 seed but were not ranked in the preseason AP poll. None have made a Final 4. Priors and wisdom of the crowd entering the season seems to hold some water as a predictor for a deep tournament run. I chalk this mainly up to school pedigree and talent levels reverting back to the mean if a team overperformed throughout a regular season. They can be really good regular season teams but when it comes to winning four+ straight in a row versus the top competition, usually coaching and talent supercedes everything else. None have exceeded their seeding expectations. A #1 seed is expected to make the Final 4 whereas a #2 seed is expected to make the Elite 8. This season Michigan State and St. Johns qualify as #2 seeds.

Potential Vulnerable Top Seeds

Who are the top seeds that could be primed for a big upset early? Historically, we have seen a pattern among higher seeded teams that are prolific on offense but mediocre to poor on defense, be ripe for an early upset. Nothing is hard and fast as a rule but more of blinking red warning lights. Teams that don’t play consistently good “enough” defense and get key stops find it very difficult to string together multiple tournament wins in a row for a deep run. However, because we are looking for higher seeded teams (7 seed or better) that may fit the upset bill, they must have likely also been really good on offense in order to receive such a high seed. Those are the types of teams we could fade from making a deep tournament run. Again, in smaller pools you may not need to pick many upsets but potentially fading 1-2 of these teams from making a deep run could give you some options.

Going back to 2001, there have been 45 teams that have been a top 7 seed ranked top 25 in raw offensive points per possession but with a raw defensive efficiency ranking of 135th or worse. Among these 45 teams, only two have made the Final 4 and ironically it has occurred in each of the last two seasons (Miami FL in 2023 and Alabama in 2024). Five others made the Elite 8 and 13 teams were upset by a double-digit seed in round one. That includes two last year with #3 seeded Kentucky losing to #14 Oakland and #7 Florida losing to #10 Colorado, both in round one. #3 Baylor survived the first round but got upset in round two by #6 Clemson.

Among the 45 teams that have qualified historically in my database, 71% have won their round one game vs an expectation of 75% based on their seed. These teams have won 0.17 games in the tournament less than expected based on their seed. In aggregate these teams have gone 19-25-1 ATS in round one games with a -2.0 ATS margin. If you isolate just those teams as big favorites (-10 or more), they have gone just 4-13-1 ATS in round one with a -4.8 ATS margin.

The overall straight up win percentage vs expected is not overwhelmingly poor but if you dig into #2 seeds, their results are a bit stunning. There have been seven #2 seeds that have fit this criteria since 2001 and five have won their 1st round games vs the 15 seed. Those two Ls may not sound like much but considering we have only seen seven #2 seeds since 2002 lose to a 15 seed out of 88 games and two of those seven have come with teams that fit this category, it is worth paying some attention. These #2 seeds had a pre game win expectation of 90% and they only won 5/7 (71%). They have also gone 0-6-1 ATS in those games with a -8.3 ATS margin. Further, out of the seven #2 seeds, two lost in round one and the five remaining teams that made it out of round one ALL HAVE LOST IN THE 2ND ROUND. Amazingly, all five of those #2 seeds lost by at least six points to their opponent including two losses by double digits. The teams that fit this criteria are below including the sole #2 seed this year- Alabama. The Crimson Tide are currently a 23.5 point favorite over Robert Morris.

Double Digit Cinderella

Every year the biggest question is who can be this year’s Cinderella? They come in all shapes and sizes but I will do my best to try and profile what historically has proven to be the characteristics that can match some teams this season. For point of reference, there have been 17 teams seeded 11 or higher to make the Sweet 16 since 2014 and at least one has made it in every year. Since 2014, 11 of the 17 teams seeded 11 or higher to make the Sweet 16 have come from one of the big conference schools. Last season, NC State from the ACC was the only double digit seed to make the Sweet 16 as an 11 seed. Now let’s remember that this is a small sample of teams and there is lots of variance needed to win in an upset. Sometimes it’s as simple as a team getting hot from 3 or the opponent going ice cold, etc.

But a few things do stand out. For one, the median adjusted tempo that these teams played with ranked 232nd in the country. Seven of the 17 teams that have pulled multiple upsets, had a tempo that ranked over 300th in the country and only one (2021 Oral Roberts) was a fast paced team inisde the top 100 nationally. The two teams that played the fastest tempo in this group (2021 Oral Roberts and 2015 UCLA) also were excellent 3-point shooting teams. Oral ranked 20th and UCLA 71st in 3pt percentage so they could get away with a lot of possessions since they made a lot of 3s. But in general, a big chunk of these 11+ seeded teams that reached the Sweet 16 in recent memory, played at a generally slower pace. And that makes sense since in order to increase your odds of beating a better, more talented team, you want to maximize the variance by decreasing the number of possessions as much as possible. Three of the recent 11+ seeded upset teams that came from non Power 5- Loyola in 2018, Saint Peter’s in 2022 and Princeton last season, were extremely slow and all ranked top 90 in lowest raw points per possession allowed.

The other thing that really helps is taking advantage of what your opponent is weak at defending. If the higher seeded team lets you have extra chances to score (opponent offensive rebounding) or doesn’t take away chances for you to score (bad turnover rate on defense) then that will help the underdog. So finding a good combo of a double digit seed who keeps things slow, can make some threes but also plays an opponent who doesn’t minimize your chances to score, is ideal for an upset or two. Here are some candidates:

When evaluating those potential 11+ seed Cinderella teams, we would love to marry those that can slow down the pace, hit some 3s and face an opponent who doesn’t turn over teams at a high rate and will allow some offensive rebound opportunities. Top seeded teams like Illinois, Alabama, Wisconsin, Kentucky, Michigan and Kansas all pop as teams that allow a good amount of 2nd chances and possessions for opponents. These teams all rank 260th or worse nationally in opponent effective possession ratio which essentially measures the number of shot opportunities allowed per possession.

Using Actual Odds vs Public Picks

Comparing betting odds to win the title versus what the general public is picking can also help find good value. Using the table below to find teams that are undervalued from a championship standpoint in brackets is useful for large pools. (only teams with 1% or more win probability listed).

As you can see among the top seeds, Auburn and Houston are clearly showing the best “value” between true odds to win the Championship and public perception. Duke and Florida are the most overvalued teams but quite honestly as compared to prior years, 1% (Florida) and 2.1% (Duke) are not too egregious. By comparison, last year we had two teams (UCONN and UNC) overvalued by more than 4%. I would not necessarily fade either Florida or Duke based on this but in very large pools you should also strongly consider Auburn and Houston if entering multiple brackets.