Predicting NCAA Tournament Outcomes Based On Historical Results

   Written by Rick Matthews

For the first time ever, the entire Division I NCAA Tournament will be held in Indiana. To mark this momentous occasion for the state, the Sports Innovation Institute wanted to help interested Hoosiers fill out their bracket for March Madness this year. This post is part of a 4-part series that is analyzed and written by the SII Data Analyst team to help you get ahead of the curve and have a leg up in your bracket pool using the power of data analysis. This post will focus on the historical statistics, but stay tuned before the round of 64 begins for posts on 2021 conference strength in the tournament, 2021 first-round matchups by offensive/defensive efficiency, and teams who have done the best and worst at beating the spread in 2021.

 

Reminder: While we can use data as a resource to create educated guesses on potential outcomes, this does NOT guarantee a particular result. Please do not take any tournament analysis as fact. 

 

One of the most fun parts of the tournament is strategically selecting which upsets will happen and where. Looking back into the tournament’s history, there has only been one year (2007) where there has been fewer than 8 upsets throughout the tournament, given that there is more than a one seed difference between the two competing teams (e.g. a 2 beating a 1 seed is not considered an upset, but a 3 beating a 1 is considered an upset, and so on). The NCAA has already done a fantastic write-up on this topic and I suggest that you read the full article here, but here are some of the highlights:

• On average, the number of upsets that happen every round are halved. You probably should not be picking more upsets in the second round than in the first round, but if you are predicting 10 upsets in the first round, you should probably predict 4-6 upsets in the next round, and so on.
• A 10 beating a 7, an 11 beating a 6 and a 12 beating a 5 have incredibly similar upset frequency rates, with only a 3.5% difference between them. It is entirely plausible looking at historical data that more 12 seeds could beat 5s rather than 10s beat 7s.
  • Full First Round Upset Rates:
    • 10 over 7: 39.3% 
    • 11 over 6: 37.1%
    • 12 over 5: 35.7%
    • 13 over 4: 20.7%
    • 14 over 3: 15.0%
    • 15 over 2: 5.7%
    • 16 over 1: 0.7%
• Since the tournament field expanded to 64 in 1985, only eight 15 seeds have won against 2 seeds. However, four of those eight 15 seeds have happened in the last 10 years.  
• Roughly once every other year, an 8/9 seed knocks off a 1 seed in the Round of 32. The last tournament that was held in 2019 did NOT have an 8/9 seed beating a 1 seed. 

Next, let’s talk about conferences. Schools come and go throughout conferences and conference strength likes to fluctuate, so it is hard to draw conclusions of the 2021 conference by looking at results from the 80s. Therefore, I pulled the results of all conferences with at least one Final Four appearance from the last decade to see how they stacked up against each other.

The ACC and Big East immediately jump out as the top dogs, holding 80% of all championships since 2010. If you’re looking for a winner based on historical data, I would look at the top teams from these two conferences like Villanova and Virginia. Picking a winner is only a small part of filling out your bracket however. Looking at the Final Four data, 20% of all Final Four teams have come from the Big Ten, followed closely by the Big East and SEC. Don’t be afraid to take most if not all of your Final Four teams from this conference, especially considering how strong they have been throughout this year. The final major stat here is winning percentage, and the ACC, SEC, and Big Ten are the only major conferences sitting comfortably over a 60% winrate. Teams from these conferences seem to always show up, so feel confident picking them to win their games. 

 

These conferences are typically the powerhouses, but who are the conferences that are likely to cause some trouble as double-digit seeds? I collected some data on how many wins conferences have gotten from their double-digit seeds. 

Looking at this data, I immediately notice how strong the A10 is when they are given a double-digit seed. VCU has done a lot of heavy lifting for the conference, including a Final Four run as an 11 seed in 2011, but even if you were to exclude VCU, the conference still sits near the top of the table thanks to other solid performances by teams like Dayton, Richmond, and La Salle. A10 teams have done historically well when counted out early, so if you’re looking for potential upset specials, I would tend to favor the A10 teams. As of today, Joe Lunardi has VCU and St. Bonaventure currently in as 10 seeds in his current prediction, with Saint Louis being in the First Four Out. I like taking one of these potential teams in the Round of 64 and given VCU’s rich history of overperforming in the tournament this decade, I think that they probably have the best chance to do it. 

Another item that stood out to me was how disappointing the Big 12 has been as a double digit seed, racking up only 1 win in 10 years as a Power 5 Conference (shoutout to 2013 Iowa State). You might be better off picking the top ranked Big 12 schools like Baylor and Texas to win some games rather than some Big 12 bubble teams like Oklahoma. 

One final note is how well Xavier and Syracuse have performed last decade as a double-digit team. Both are bubble teams this year, so if they can get in, watch out! 

 

Unfortunately, my final section will not be long to write as I wanted to focus on the Indiana-specific teams that would be participating in this special Indiana edition of the tournament. Right now, it looks as if Purdue is the only in-state team that will participate this year. Purdue has made the last five tournaments but have only gotten to the Elite 8 once (2019). Matt Painter will be bringing in a very young team and will be playing with fans in the stands for the first time all season so it is hard to say what will happen, but when Purdue is a 4 seed or higher, they have always at least reached the Sweet 16 since 2012. When Purdue is lower than that, it spells trouble, so keep that in mind when the bracket releases. 

 

I hope you enjoyed these insights and have taken something from them, but remember that the SII Bracket Analysis has just begun! We have plenty more to come and they will be heavily focused on the current season, so stay tuned to watch us break down individual matchups and what the tournament could have in store after the brackets are released.   

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The Sports Innovation Institute is leveraging the expertise of faculty and students across the IUPUI campus​, the Sports Innovation Institute blends academic thinking with an entrepreneurial spirit to create new products and research that help solve important problems in the sports industry.