Average Fans take on NCAA Basketball Experts and Win

As the NCAA basketball season heats up, fans and bettors across the country are turning to experts at ESPN, Sports Illustrated and CBS Sports for information on how their favorite teams will do. Guys like CBS’s Gary Parrish can boast of hundreds of thousands of readers for their daily columns and betting advice.

But, are these experts really any better than average fans in predicting games? To find out, researchers at Unanimous AI have been tracking Parrish’s picks against a set of picks made by average fans. But, there’s a twist….

The fans are using a new form of of A.I. called Swarm Intelligence to boost their accuracy. It works by tapping the wisdom of online groups, combining the knowledge and insights of many sports fans into a single unified prediction.

No, this isn’t a poll or survey, but a dynamic system – a swarm – in which the participants push and pull in real-time, until they converge together on optimal solutions.  It’s enabled by a free software platform called UNU that allows people around the world to login and anonymously form intelligent swarms on any topic.

The image below shows a swarm making a prediction in real-time. Each magnet is controlled by a different user, the group quickly converging on a prediction.  But is it a good pick?  Well, the swarm predicted that Kansas would win by 11 and cover the 7 point spread. Incredibly, the swarm was accurate to within 1 point: Kansas won by 10.


Of course, one game means nothing.  How accurate is Swarm Intelligence over the long haul? So far this season,  researchers at Unanimous AI have asked swarms to pick a total of 36 NCAA games against the spread. Those picks were all recorded HERE and are compared to Gary Parrish’s expert predictions, available HERE.

How did the Swarm do against the expert?  In short, incredibly well. Here are the facts:

  • CBS Sports expert Garry Parish’s record is 13 wins and 23 losses. That’s just 36% accuracy
  • The Swarm’s record over the same games is 20 wins, 15 losses, and one push, or 57% accuracy

That means that, when average fans put their heads together, they did a bit better than the expert. If you’d placed your bets following Gary Parrish’s advice, you’d actually have lost money. If you bet with the Swarm, you’d have made a tidy little return.

But the swarm outpicking the experts is no fluke. Previous swarms crushed ESPN in making nearly perfect predictions about the NFL Playoffs, and beat 99% of bettors in betting on Prop Bets in each of the last two Super Bowls. The swarm generated a 34% gain betting on the major College Football bowl games, while the ESPN experts lost 24%!

Now we want to spread the wealth! We’re only 3 weeks into this experiment, and we’ll be continuing through March Madness. If you want to get our Weekly Picks NEWSLETTER of swarms predicting Against the Spread, just click the button.


This Post Has 2 Comments

  1. Pingback: March Madness: Can Big Data Prevent a Busted Bracket? - GenX i Social Buzz

  2. Pingback: March Madness: Can Big Data Prevent a Busted Bracket? – Synertree

Comments are closed.