PredictionMachine.com is a website that plays the game 50,000 times before it’s actually played. For the NCAA Tournament, the “Predictalator,” a computer simulator plays through every play of every game, to not only provide picks and predictions for each of the first round games, but to actually play through the entire field and give the likelihood of any team making it to any level of the bracket.
As General Manager of PredictionMachine.com, this is my eighth tournament applying similar technology to the Madness. In the previous seven seasons, as of Selection Sunday, this methodology has accurately forecast the eventual champion five times. For the 2011 NCAA Tournament, PredictionMachine.com likes Ohio State to win the title more often than anyone else, yet just 12.7% of the time – leaving the field pretty open for other teams like Duke (11.5% likely to win the championship), Pitt (10.3%), Notre Dame (10.1%), Kansas (8.8%), Syracuse (5.7%), Purdue (5.4%) or Kentucky (4.7%).
Paul Bessire joined Murph and Mac on KNBR in San Francisco to talk about the Predictalator, first round upsets, each region, and the most likely champion.
On if it is true that the machine had a “pretty good” NFL Playoff season (and how that translates into college basketball):
“Yeah, we did all right. We went 11-0 against-the-spread with our predictions on the site. It’s hard to do better than that. Actually, going back to last year, are 12-0. We launched the site for the Super Bowl and I think we talked to you guys for that last year. The mantra is that we play the game 50,000 times before it’s actually played. Especially around this time in the season – we haven’t done nearly quite as well during the regular season in the NBA, where it’s tough to figure out motivation, etc. – but when it comes to playoff time or the NCAA tournament like we are talking about here, the technology is very successful because everyone is trying their hardest to win, we have neutral courts so we don’t have to too much to worry about there and, beyond that, we have a full season’s worth of data on every player. So something like what we saw with the NFL playoffs, we can keep going here. And actually, it was around this time of year, when we had our first great success, hitting over 60% of our games in the tournament against the spread, picking every game.”
On how this year’s most likely champion compares to previous:
“Some of those years, it was 30-35% of the time or so that we were predicting as the likelihood of the top champion winning, which is crazy to think about now. Our top champion this year is only about 13%. That’s the least likely, most likely champion we have ever seen. I think that speaks a little bit to the parity to the top 16 seeds. All those 1-4 seeds in this year’s tournament are interchangeable to a good degree. Outside of those 16 teams, very little teams have a shot. But in those top 16 teams, just about anybody could win.”
On the depth in East region:
“Ohio State is our number one team, but just at about 13% though. If they were specifically in the Southeast region, which is by far the easiest region for us and ultimately a situation where Pitt is just kind of handed a walk into the Final Four – making it almost 40% of the time, if Ohio State were in the region against those teams, we’d have them winning the championship almost twice as much – about 24% of the time. They would have to go up against Kentucky, a top 5-6 team in the country. North Carolina is obviously very good. Seven of the top 20 teams in our final power rankings ended up in the East. That’s just brutal to see that Ohio State, the number one overall seed, gets that draw.”
On the other number one seeds:
“Duke is our second most likely champion. They have a very straight-forward region, very defensively oriented, something where we see them getting to the Final Four about one out of every three times. Notre Dame is the one non-one seed that we would expect to be the team that makes it to the Final Four, beating the one seed in that region. Notre Dame gets to the Final Four about 26% of the time, while Kansas gets there about 22%, so there is not much difference. The Southwest region is very top-heavy with Kansas, Purdue, Notre Dame, maybe even a Georgetown or Louisville. Any of those teams could represent that region in the Final Four. And lastly, I mentioned Pitt. They have as much chance as anybody to make it to the Final Four. After that, it’s fairly bleak for anybody coming out of the Southeast. That region specifically only wins the championship about 17% of the time. If you think of it (with no dominant teams), it should be about one out of four. That region is much weaker than the rest.”
On the best team no one is talking about that the computer loves:
“It’s Kentucky. I think it’s a travesty to put them in that region. If Kentucky were in the Southeast region, they would be our Final Four participant. If they were in any other region, they would be much more likely to make the Final Four than they are right now. Kentucky is definitely the strongest four (seed) – one of the top 5-6 teams in the nation. Usually with Calipari, you concerned with a young team that turns the ball over and shoots free throws poorly. This is actually a team that shoots about 72% from the free throw line and is one of the better teams in terms of efficiency with turning the ball over… And if you like double-digit seeds that might be able to make a run, we like Gonzaga and Marquette as elevens over sixes. And even the Missouri team could go deep. It’s almost like all three of the eleven seeds that are already into the Thursday and Friday games are much better than their seeds.”
On the difficulty predicting regular season NBA:
“It’s (player) apathy really. I have to acknowledge it. It’s fascinating to me. Going back to the SEC tournament with Kentucky the other day, we assume that Doron Lamb would be hurt and that DeAndre Liggins wouldn’t be at 100% and I got frustrated because they both ended up playing, playing well and that threw off our pick. Well that happens 3-4 times a night in the NBA, where somebody we assume is going to play doesn’t or vice versa. Add that into the clear factor that some players play harder in certain games than others. It’s difficult. That being said, when it comes to looking at series in the NBA, we do well, hitting 11 of 12 last year. And when looking at individual games we should do well in college basketball this time of year. It makes a lot of sense, but I still wish we could be better in the NBA.”
And on how the Predictalator would fare against Watson from Jeopardy:
“If it’s just picking games in college basketball or at least the NFL or something else that we have talked about on the program, I think we are in good shape. Outside of that, we have no chance.”