Brackets Busted But TIBCO Basketball Predictions Score Big 

TIBCO Spotfire
Reading Time: 2 minutes

Were you among the 20 million to witness the Kansas Jayhawks come from behind and defeat the University of North Carolina Tar Heels? Though Kansas had a slight edge going into the final, this year’s NCAA men’s basketball tournament was notable for the number of upsets. In fact, Kansas was the only #1 seed to even crack the Final Four. 

How did your bracket fare? 

For those who joined TIBCO’s “Metrics of Data Madness” webinar prior to the start of the tournament, congratulations, your bracket performed far better than most. The TIBCO data science team ended in the 77th percentile with its forecasted bracket. The model did not predict Kansas would win but instead had them losing out in the Final Four. Overall, however, and just like every year we’ve participated, the TIBCO bracket beat the crowd–and the house. 

Building a Winning Bracket

The Metrics of Data Madness

The odds of choosing a perfect NCAA Men’s bracket are 1 in 9,200,000,000,000,000,000. Yes, that’s one in 9.2 quintillions. Such improbable odds mean even those who follow basketball closely are effectively no better at selecting a perfect bracket from those who have never watched a single game. But if you want to beat the odds, heed the advice of the TIBCO data science team. It’s not luck, it’s science. 

The science behind building a winning bracket is a multi-step process. The TIBCO data science team leverages multiple publicly available data sets, including historical data from past tournaments and individual team performances. They use TIBCO® Data Science and TIBCO Spotfire® to identify data patterns and perform data exploration of the historical data. Factors such as draft ratings of individual players and even the distance a team must travel for its next game are also built into the model.  

The TIBCO data science team simulates potential matchups and explores multiple scenarios using various methods, including classification and regression techniques, random forest modeling, and other embedded data science workflows. 

Finally, the TIBCO team creates an algorithm to make their NCAA tournament predictions. This algorithm is tested against the historical data to see if it accurately predicts past winners and losers. This interactive matchup simulator reveals the top predictors for the tournament.

The Power of Prediction

Every year, the TIBCO model significantly outperforms against the 70 million or so brackets submitted. And just like with bracket predictions, data science also gives you a leg up in your business. TIBCO’s industry-leading data analytics software can help you make smarter, faster, and better-informed decisions about your business. 

To apply TIBCO analytics and data science solutions for greater business success, download a complimentary trial of TIBCO Spotfire analytics today. Winning is always fun. 

And for more stories on how business leaders are overcoming today’s most pressing digital challenges with innovative analytics and data management solutions, join us at the TIBCO Analytics Forum on June 13-15, 2022. It’s a totally free, all-digital event.

Try TIBCO Spotfire free for 30 days.