Using 20 Years of Data to Predict THIS Year’s March Madness Winner!

I analyzed the last 20 March Madness tournament winners to find the criteria that every championship team has met.

We will look at team records, winning streaks, strength of schedule, team ceilings, and more to help us discover every team that can win the big dance in 2025.

We’ll also look at what smaller schools can become the first mid major tournament winner in decades.

Since the ordering of these cuts was completely subjective, the order in which a team was eliminated in this article does not mean they are a better or worse championship contender than a team eliminated in different spot than them. Perhaps a better way to look at things is to see how many thresholds a team failed to cross. So, here are all the teams that missed out on the final list because of one single criteria failure.

Methodology

We are looking at every teams regular season and conference tournament as a whole. All statistics are the combination of these two periods.

We will mainly be looking for the minimum statistical values that each of the 20 previous tournament winners achieved. These minimums will help us to find the threshold that all championship teams met, allowing us to identify some sneaky teams that are true contenders to win it all.

Any obvious outliers will be addressed as we go.

If this sounds confusing right now, don’t worry. It will all make sense shortly.

Data Analysis

Below, you will find the questions I asked, my thought process of how I analyzed the historical data, and how it applied to this years teams.

If you just want to see the teams that met every requirement, go to the conclusions.

How many games does it take to win the NCAA Tournament?

It takes 6 wins in a row to win it all. So naturally, the first thing I wanted to know was if every championship team had a 6+ game winning streak before the NCAA tournament.

The answer was yes.

But the actual lowest max winning streak of each team was 7 games.

In other words, every single champion had proven before the tournament that they could win 7 games in a row. But for our purposes, let’s just focus on the 6 game requirement.

This will be our first criteria to narrow down the initial field of 68. If a team has not had a 6 game winning streak already, then history says it’s not happening in the tournament.

After applying this, we are already getting some interesting eliminations. #4 Maryland, #6 Illinois, #8 Gonzaga, and #9 Baylor are all single digit seeds who failed to win 6 games in a row this year. We have an additional 4 double digit seeds that also failed this requirement, leaving us with 8 teams gone, and 60 teams remaining.

How do we value different streaks?

Not every winning streak has the same value. A 10 game winning streak and a 6 game winning streak are both 1 streak of 6+ wins. But clearly, the 10 game streak is better.

I wanted to find a way to take this into account. Let’s take another look at the 10 game streak. If we think of each 6 game block as its own independent series, then the 10 game winning streak is really 5 unique 6 game winning streaks. Looking at it this way, we can give a little more value to a longer winning streak, while still focusing on the 6 game requirement.

Of the last 20 champions, they all had 3 or more unique 6 game winning streaks.

I like this static specifically because it proves a team is able to win the required 6 games in a row multiple times, even if it was just 1 actual streak.

Using this criteria on the remaining teams, we are able to eliminate 15 more teams. We have now knocked off 2 different 3 seeds in Kentucky and Texas Tech. Kentucky is an interesting case because they started their season 7-0, but then were unable to win more than 3 games in a row the rest of the season.

We also eliminated another 4 seed in Purdue, and 3 additional single digit seeds. There are now 45 teams remaining.

Now we know what teams are capable of winning 6 in a row, but thats just one aspect to consider. Winning 6 games in a row doesn’t actually mean you are good at winning consistently. Maybe that team just got hot once during an easy part of the schedule.

Let’s try to find ways to consider a teams actual strength. We’ll start with total wins.

What's the least amount of wins a championship team had?

Every tournament winner has had 25+ wins, except for the 2021 Baylor Bears. This was an anomaly. Baylor went 22-2 during a season in which they had 8 games cancelled. In a normal season, it’s pretty safe to assume they would have reached 25+ wins.

Most teams play 30-34 games during a year. For our previous champions, this meant that no team had more than 9 losses during a season. So if your favorite team is sitting at double digit losses, history says they are not going to win it all.

Let’s eliminate all teams with less than 25 wins and 10 or more losses.

Iowa state is the 3rd 3 seed to be eliminated. They finished just short of the threshold with a record of 24-9, as did Oregon and BYU.

With Arizona and Texas A&M failing to reach 25 wins, the final two 4 seeds are out. Additionally, there are zero remaining teams in the field ranked 6, 9, or 16.

This was our biggest hurdle for teams to clear so far, with 18 teams removed, and just 27 remaining.

We have now cut the field down significantly, but we have more to consider. Teams play wildly different schedules which means that not all 25 win seasons are the same quality. Generally, a team from a power conference is going to be play a tougher schedule than a mid major team. But we don’t want to eliminate mid major teams from consideration, so let’s find an alternative way to look at things.

Strength of Schedule

To do this, we will look at each remaining teams strength of victory and strength of defeat. Strength of victory is the combined winning percentage of all the teams they beat, while strength of defeat is the combined winning percentage of all the teams they lost to. Among championship teams, the lowest strength of victory winning percentage was 48.4. This is roughly equivalent to a 15 and 16 team.

A percentage lower than this might mean that a team was beating up on lower quality opponents.

Additionally, the worst strength of defeat among championship teams was 59.1 percent. This means that championship teams really were only losing to good teams on average.

Strength of Victory

Let’s eliminate all the teams that fail this strength of victory test first.

Unsurprisingly, 4 mid major teams failed this test. I would like to give a shoutout to the Mcneese Cowboys though. They had a great season, going 27-6, with one of those losses being by 8 points at 2 seed Alabama and another being a 3 point loss vs 8 seed Mississippi State. I would not be surprised at all to see them play 5th seed Clemson tough in round 1.

Strength of Defeat

The strength of defeats criteria results in 4 more mid major teams getting cut from our list. These 4 teams all had great seasons, but have shown themselves to be vulnerable to losing games they shouldn’t.

For an example, let’s take a look at Liberty. They went 28-6 on their way to a Conference USA title, and never lost to a team with a losing record. But they did lose to three teams with winning percentages below 59.1 percent. Why exactly does this matter? Well, nearly every team in the NCAA tournament has a winning percentage above 60%. Losing these games during the season shows you are just not quite in that elite tier of teams in terms of consistency.

These two tests resulted in the loss of 8 mid major teams, leaving us with just 19 teams left.

So championship quality teams play quality schedules, and on average they only lose to good teams, but surely some of them have lost to a quote unquote “bad” team, right? Well, no. Not really.

What’s the worst loss by a championship team during the season?

The worst loss, by record, of the last 20 championship teams was against a 15-16 team. The next worst loss was versus a 16-17 team. In other words, no championship team ever lost to a team who finished more than 1 game below .500.

In my opinion, this is a good stat to see if a team is susceptible to a let down game. Championship teams consistently beat the teams they should and do not lose to teams that are having bad seasons.

Applying this criteria results in the loss of 6 more teams, including 2 of the 5 seeds in Clemson and Michigan. So what went wrong for these 2 teams?

Clemson lost an overtime game at in-state rival South Carolina, a 12-20 team that finished last in the SEC.

For Michigan, the story is a little different. They barely missed the threshold by losing on the road to a 15-17 Minnesota team in overtime. Michigan came into that game on a 5 game winning streak.

So in one instance, an intense rivalry got the better of really good Clemson team, and in the other, Michigan just seemed to have a let down game. But it just goes to show how tough it is to stack 6 wins in a row and how its tough to win every time the pressure is at its highest.

13 teams remain, including every 1 and 2 seed, but we do have 2 double digit seeds and an 8 seed who have cleared every test so far.

What is the best team you have proven you can beat?

Now we will look at a team’s ceiling.

Among championship teams, the lowest quality best win by winning percentage was the 2006 Florida Gators victory over a 23-11 team, which is a 67.6% winning percentage.

All 13 teams remaining have met the 67.6% best win percentage requirement, so they all pass this test at first glance.

But let’s take another look at the best wins. Florida’s win vs 23-11 team was a quality win, but in terms of best wins, this number truly was significantly lower than the other 19 champions.

The next lowest best win by percentage was 79.3% by Baylor. As mentioned before, the Baylor year had quite a few regular season cancellations, but they still beat a team that was 23-6. 4 other teams had their best wins vs 27-7 teams, which is a 79.4% winning percentage.

Logically, this does look like the 2006 Florida year was an outlier, but let’s check the math.

If we plug all of the best win percentages into a standard deviation calculator with a 99.9% confidence interval, we can see that the 67.6% is well below the 3 standard deviations minimum and could be considered an outlier. Like everything in this article, we are only using a 20 team sample size, so nothing is completely foolproof.

For our sake, let’s throw out the outlier here and use 79.3% as our threshold.

This criteria resulted in a major shakeup in our remaining teams. Not only did we lose 5 of the 13 teams, but we lost our first #1 seed in Houston, and our first two #2 seeds in Michigan State and St John’s.

Of these 5 teams, neither St John’s nor Drake had the opportunity to play a team this consistent at winning. While it may seem a bit unfair to cut them with zero chances, the prior champions all proved themselves capable of winning these games before the tournament.

However, I would like to note that St John’s was particularly close to the threshold with a win vs a 78.8 winning percentage team.

So we’ve shown that these teams can all win consistently, and that they can beat really good teams. But we have yet to show they can beat tournament caliber teams consistently. To do this, we’ll look at every teams record vs 20 win teams.

Can you beat 20+ win teams consistently?

Nearly all tournament teams will have 20+ wins, so your record vs them is a good indicator of how you might fare in the tournament.

All of the previous 20 championship teams won 50% or more of their games vs 20+ win teams, and each had a minimum of 4 wins.

Of the remaining teams, only Wisconsin failed to reach the 50% win mark vs 20+ win teams. Their record of 5-8 is winning percentage of just 38.5%.

While they did reach the 4 win total, Wisconsin has not shown themselves to be consistent at winning games vs tournament caliber teams.

We are down to 7 teams from our initial field of 68.

Now lets consider one final aspect.

Whats the minimum number of players that average 10+ppg on championship teams?

Every championship team, other than Louisville, has had anywhere from 3-6 different players average 10+ ppg. But even Louisville was extremely close to reaching this threshold, with their third and fourth top scorers each averaging 9.8 ppg.

So, how many of the remaining teams have 3 or more players at 10+ ppg? All of them. In fact, everyone except Duke has 4 or more players in double digits. But Duke does have 3 players averaging 12+ ppg.

In my opinion, having multiple scorers matters a lot. It’s pretty likely that over a 6 game run, a teams top 1 or 2 players will have at least 1 mediocre or bad game. Having 3 players that can score gives the ability to overcome this, and win the tough games.

Conclusions

There’s no magic formula for picking your bracket, and there are many different ways to analyze a team’s strength. With all that being said, here are the 7 teams that fit the mold of what we think a championship team looks like based on the last 20 tournament winners.

I gotta say, I really don’t understand why Louisville is such a low seed, considering everything we’ve covered. For even more context, they ended the year ranked 10th in the AP poll following their 2nd place finish in both the ACC regular season and conference tournament. If you just want a sleeper team based on seeding, Louisville could be a good option.

6 of these teams are from power conferences, with 4 of them being from the SEC, and the other 2 from the ACC. Perhaps more interestingly, we do have 1 mid major team in Memphis. Unfortunately, they did lose starting guard Tyrese Hunter to an injury during their conference tournament. While it sounds like he may miss some time, let’s hope for the best and just treat them based on their resume.

Theres not a lot of love for Memphis in some of the best well known team strength models. For example, Kenpom, which also doesn’t consider injuries in their formula, has Memphis ranked as the 51st best team, despite them having wins over Kenpom’s 8th ranked Michigan State, 18th ranked Clemson, 26th ranked Ole Miss, and the 35th ranked UConn Huskies, who by the way are the 2 time defending champs.

This is not a critique of models like Kenpom. It’s just an entirely different approach. While others are looking at more advanced and efficiency based statistics, I tried to look at things from a more straightforward perspective.

Basically, based on the last 20 tournaments, I believe a championship team can answer yes to all of the following questions:

Can you win 6 games in a row?

Can you win consistently?

Do you beat every bad team in your path?

Can you handle a moderately difficult schedule?

Can you beat the best teams when you have the opportunity?

Can you beat tournament caliber teams more often than not?

Can you get quality scoring outputs from multiple players?

What teams just missed out on hitting all the requirements?