Here’s the high-minded reason for today’s column: Major corporations like yours rely on business-intelligence software such as that provided by SAS to discover competitive advantages in an increasingly cutthroat business world. Here’s the real reason: I know that your time for such weighty work matters is limited today by a more pressing need to find competitive advantage in your cutthroat March Madness office pool.
Here’s the high-minded reason for today’s column: Major corporations like yours rely on business-intelligence software such as that provided by SAS to discover competitive advantages in an increasingly cutthroat business world.
Here’s the real reason: I know that your time for such weighty work matters is limited today by a more pressing need to find competitive advantage in your cutthroat March Madness office pool. The sooner we can complete your brackets the sooner your mind can return to the business of business.
First the background: Jay Coleman, an operations management professor at the University of North Florida, and Allen Lynch, an economics professor at Mercer University, have been making a name for themselves among basketball fans for the past dozen years by using SAS business-intelligence software to predict the field of participants in the annual NCAA tournament. They claim their model has correctly picked 94% of the at-large teams, or those schools that do not qualify by winning conference championships. You can read the nitty-gritty at www.dancecard.unf.edu and see how they did in divining the brackets announced yesterday.
So how might that 94% rate compare with that of any good sportswriter?
“Probably about the same,” Coleman concedes. “What we bring to the table is that we can be as accurate without consulting the expert and without really taking the time to watch basketball games and pore over statistics,” he says. “That’s one of the things that a model or a formula can do for you that in a business setting might actually allow you to free up managerial time.”
Right, right, business setting, managerial time . . . what else do they have that might actually allow us to free up a few of the dollars from our office pool? Well, this year for the first time Coleman and Lynch are turning their model loose on what really matters: who’s going to win the games.
“We’re going to give it a shot,” Coleman says. They’ve rejiggered the variables they use to pick the participants so as to produce “maximum likelihood estimation processes,” or in English, pick winners.
“One thing that is quite interesting is that the things that seem to matter on who’s actually going to win a game are not the same things that the [NCAA selection] committee looks at when it decides who’s going to get in the tournament,” Coleman says.
According to the academics’ analysis, the selection committee relies most heavily on a half-dozen factors, including a team’s Ratings Percentage Index (RPI) ranking, conference record, and record against Top 25 teams. But do not follow the committee’s lead when completing your brackets.
What does matter?
First up is the power, relatively speaking, of the conference from which a team arrives in the tournament.
That might seem painfully obvious to some, but the selection committee apparently does not consider relative conference strength in choosing tourney participants, according to the professors’ analysis. You should.
The second key to picking winners is to favor those teams that topped their conference regular-season standings, not those that won their post-season conference tournaments, as conventional wisdom often holds. “It’s not the hot team, it’s how did you do over the bigger sample size of the whole season,” Coleman says.
And then there’s this utterly strange X-factor: non-conference wins and losses.
“The weird thing here is the worse you do out of conference, the better your shot at winning a game in the tournament,” Coleman says. “It is strange and it may be spurious; we worry about that when we do quantitative analysis.”
What it might mean is that such teams are battle-hardened come tourney time.
There you have it: The software says you should pick teams from powerful conferences that won regular-season titles and lost their fair share of non-conference games. Now back to work.
Don’t blame me if they’re wrong. The address is buzz@nww.com.
