In the modern era of the NCAA tournament, you might think that projecting games based on a little artificial intelligence would be useless. After all, algorithms that drive projections require good data, and good data comes from robust sample sizes.
So what could a computer tell that our eyes couldn't about Kentucky, led by another band of likely one-and-dones? That's not a shot at the Cats -- ripping John Calipari for recruiting, well, is like ripping your mechanic for never trying out duct tape -- it's a shot at the reality of small samples. Consider that Anthony Davis, the best player on the team and a lock to be taken No. 1 overall in the NBA draft, hasn't just played one year of college hoops as a post player, he's only played two years of basketball as one. He was 6-foot-3 as a high school junior and is quite literally learning about playing on the block (including swats) on the fly. What can a computer tell us about a team that is growing before our eyes?
Well, a lot more than you'd think, apparently. Kentucky isn't the only squad relying on youth, but our Elite Eight picks came in at 7-of-8, and our Final Four picks came in at 3-for-4. Only Michigan State (and its vets, go figure) led us astray, and perhaps more of that has to do with Louisville. The model had a tough time with the Cardinals, perhaps because it would be like trying to divine some truths from three teams. Seriously. The Cardinals started the season hot, were utterly dismal for a good stretch of the regular season and have since caught fire late in the year. Sort of like that Connecticut Huskies squad of last year, no?
Where the computer has the edge is in considering four models (see above table). And some are more predictive than you might assume. With that in mind, let's see where the computer likes things headed into the Final Four.