March Madness: Can Big Data Prevent a Busted Bracket?

Shawn Krest Writer, Movable Media

While high-tech game simulators, crowdsourcing and advanced analytics can help fans pick their NCAA brackets, perfection remains elusive even for the most dedicated data nerds.

Despite the influence of big data and plenty of analytical tools to help make informed decisions, NCAA March Madness Tournament fans don’t appear to be getting any closer to picking a perfect bracket.

For some, so much data could be devastating.

“It’s going to be carnage,” ESPN’s Jay Bilas, a former player, said of the prospects of picking this year’s winners. “There are going to be landmines everywhere.”

Granted, the odds were never going to be easy. With 63 winners to pick in most brackets (the full tournament plays 67 games, but the opening four are usually not included in the sheets fans fill out), the odds of guessing correctly are one in 9.2 quintillion. Maybe even worse.

march madness basketball

From a game simulator that plays games possession by possession, to a group that picks sports wins as efficiently as it predicts Oscar wins to a data nerd with a surprising success rate, tech-savvy fans have sound digital tools to help improve their chances for picking winners.

Still, the perfect bracket doesn’t appear to be on the horizon.

Strength in Numbers

Basketball is a team game, so it should come as no surprise that one option for picking winners of NCAA games involves crowdsourcing with other fans.

UNU depends on large groups of users, which it calls “the swarm” to pick results.

In a head-to-head showdown with college basketball experts earlier this season, the swarm came out victorious. UNU’s group picked the winners of 35 college games, against the point spread and were victorious in 20 of them. CBS‘s Gary Parrish picked the same games and won just 13.

basketball team

UNU’s swarm is equally adept at picking other winners as well. In a football-betting challenge, using simulated money, the swarm would have made a profit of 34 percent betting on college bowl games, while ESPN’s on-air experts would have run a loss of 24 percent.

On Oscar night, the swarm picked 76 percent of winners correctly in all categories, beating film critics from Variety, Rotten Tomatoes, Rolling Stone, USA Today and the LA Times by at least 10 percentage points.

Advanced Stats from a Data Nerd

When can a weatherman in Utah help fans with their brackets? When it’s Ken Pomeroy. A former meteorologist with the U.S. government, Pomeroy quit his job to forecast college basketball instead.

Pomeroy is one of the foremost authorities in the field of advanced basketball metrics, and his subscription site is a must-have for serious followers of basketball.

basketball shot

Pomeroy rates all 351 college basketball teams, all season long, and he has history dating back to 2003. He also looks at advanced measures, such as tempo-free statistics—meaning that a team who allows a low number of points may not really be good at defense; they might just play really slowly.

Pomeroy also looks at strength of schedule, efficiency on offense and defense, and even a team’s luck, to help give the squad a true ranking. has already been providing ratings and predictions for all the conference tournaments around the country, and he’ll shift into high gear once the NCAA bracket is out.

Game Sim Knows All

There are several programs that attempt to simulate the matchups of the NCAA Tournament, with varying levels of success. For more than two decades, Jeff Sagarin has used his statistical ratings of teams and Monte Carlo simulation techniques to play the NCAA tournament 68 million times, keeping track of who wins most often.

If a fan needs help with an individual game, however, Jon Pence’s Game Sim is worth a try. The tool, part of his website, is one of the more accurate simulators in the country, picking not just winners, but also final scores with uncanny accuracy.

The key, Pence explained, is that his program simulates the entire game with shots, points and rebounds, not just the final result.

“Game Sim takes the stats and strength of schedule for two teams and plays out a game, possession-by-possession,” Pence said. “Each possession it determines what percentage of time a team shoots, turns the ball over or goes to the free throw line. It then plays out each play breaking things down a bit more.”

basketball team

In a situation where a player shoots, Game Sim looks at the player and team most likely to get the rebound. There’s also randomization involved, so each simulation doesn’t end the same way, but Pence maintains it’s not rocket science.

“There really aren’t any complex equations in Game Sim,” he said. “Instead, it’s more along the lines of just making sure things happen the right percentage of the time. The fascinating thing is how well it predicts the score of games.”

Pence has been using Game Sim for the last five years and says it picks the correct winner 75 to 80 percent of the time. About one third of the time, it’s within five points of the team’s actual score for the game. Against Vegas point spreads, it has a 53-percent success rate, “which is considered profitable, I’m told,” he said.

The Trouble with Turnover

Part of the reason fans rely on tech tools to pick the wins is increased turnover in the sport: The best players only stick around for one season.

Witness last season’s NCAA champion Duke Blue Devils, who lost their starting center (Jahlil Okafor), point guard (Tyus Jones) and small forward (Justise Winslow) to the NBA Draft after one season in college. This year, Duke has replaced them with another crew of freshmen who have scored 48 percent of the team’s points this year.

With so many new players on the best teams, and the lack of experience in playing in the tournament, the stage is set for even more upsets than in previous years. As Bilas explained, if the “influx of talent” to a freshman-dependent team like Kentucky or Duke isn’t up to the same level as previous seasons, the favorites come back to the pack.

basketball bracket

This year, in particular, Bilas said there doesn’t appear to be a team head and shoulders above the others. There are many good teams but not one that dominates the competition.

So fans have to decide whether the hot-shot freshmen on a big-name team are going to be able to beat out the age and experience of the smaller schools, who are often loaded with seniors and other veteran players.

Speaking of Luck…

Even armed with the most-advanced stats around and with Game Sim and the swarm on their side, fans shouldn’t count on getting very far with their bracket intact this March.

“They don’t call it March Madness for nothing!” Pence said. “Whether it’s a player getting hot from [landing a three-pointer], an injury or maybe someone is just having a bad day because their girlfriend broke up with them, these are college kids that can do amazing things or amazingly boneheaded things on any given play.

“We can play the averages, but I think we all know what happens in March is rarely average.”

Share This Article

Related Topics

Sports Entertainment

Read This Next

Read Full Story