Short answer, no. (But really, who could have predicted this year’s outcome?)
For the past few years, Adobe has leveraged their enterprise analytics solution, Adobe Analytics, to build a tool in an attempt to use machine learning and predict the outcome of the basketball games throughout the March Madness tournament.
Last year, Adobe Analytics correctly predicted the 2018 March Madness winner, and brackets made with Analytics in 2018 finished in the top 98th percentile. Adobe also correctly predicted 50% of the Sweet 16, Elite 8, and Final 4.
The software has many factors to determine it’s prediction between two teams. When using the interactive tool on their website, it looks like they consider factors including; win/loss ratio, field goal %, turnovers per game, point differential a game, and a few others.
But how did the machine learning software fair this year? Let’s take a look at their bracket.
Adobe Analytics Hack the Bracket Results (2019)
Using traditional scoring methods, Adobe’s bracket scored 120 points of the maximum possible 190. Unfortunately, that would have put Adobe well outside Yahoo’s top 50 and well outside the top 100 of ESPN’s leaderboard.
Although Adobe’s bracket wasn’t among the best in the world, it still did pretty well despite not predicting the National Champion, Virginia. Some of the highlights include; successfully predicting 15/16 of the second round matchups and 5/8 of the regional semifinals. And compared to ESPN’s “People’s Bracket” (A bracket that aggregates all of the bracket submissions and determines a matchup winner by the average answer), Adobe placed 36 points higher than the average bracket.
Thanks to Adobe’s machine learning Analytics platform, if you followed their predictions, you would have been much better off than the average Joe, but still not perfect. Following the advice of Adobe Analytics predictions, you wouldn’t have won Warren Buffets annual March Madness competition, you wouldn’t have been recognized by ESPN or Yahoo, but at least you would have beaten everyone in the adHOME office.