# Living Large in the Madness

There are 9 quintillion brackets possible this year. In fact, there are more if you add the 4 games played before Thursday. That number is so enormous it makes big look small. How big? If you could produce 1 billion unique brackets per second and never repeat any bracket you've made, it will take about 300 years to create them all. That's just this year.

So, the odds of a perfect bracket seem insurmountable. You have to find that 1 bracket in the 9 with 18 zeros after it options. True, but the odds aren't 1 in 9 quintillion. That assumes you have a 50/50 chance of picking a game. Historically, people pick brackets at about 70% accuracy. Some years are higher and some years lower.

If you assume you have 70% accuracy, then your odds of a perfect bracket drop to 1 in 5.7 billion. Keep in mind, that's over 50 times less likely than winning Powerball.

If you raise your accuracy by just 1%, your odds drop to 1 in 2.3 billion. That's why looking for that edge can help. Even if you aren't perfect, you just narrowed the odds a lot!

Can you find that Cinderella team that no one sees? Can you find the high seed that's ripe to fall apart?

With about 5,500 games between about 350 teams, where do you look? One place where people often look is the conference tournaments. Who did well and who didn't?

Be careful. Why? Here's what data back to 2002 showed us when I worked with data analysts at big data company Tresata where I'm serving as Chief Researcher over my sabbatical year.

Remember, at-large bids didn't win the conference tournament. While they didn't, 85% of these teams made it to Elite 8 or beyond. Further, if an at large bid made it to the finals, they always won.

Suppose you still want to look at the conference tournaments. If you discount the season and say that someone doing great in the tournament is important, then you believe, to some degree, the conference tournament play is predictive of March Madness. To test this, Davidson undergraduate Shane Macnamara and I used our bracket methods Colley and Massey (available at http://marchmathness.davidson.edu) to predict the week of conference tournaments and March Madness. After looking at the past 14 season, we found no linear correlation in these predictions. In other words, you might have a method that can predict March Madness that didn't predict the conference tournaments well, at all.

Is this a reflection of our ranking methods? Possibly. So we turned to ESPN's BPI rankings to see the correlation between their rankings prediction of the conference tournaments and March Madness. Unfortunately, the sample size is small for the BPI rankings, only the past four seasons. Here the season that BPI best predicted the conference tournaments performed very poorly in March Madness. The season that the BPI rankings did poor predicting the conference tournaments, the ranking predicted March Madness well.

Consider the teams who perform well in the NCAA tournament, these teams are overwhelmingly teams that have performed well throughout the entire season. These teams have earned an at-large bid to the NCAA tournament prior to the conference tournaments taking place. Does a Kansas, a Kentucky, a North Carolina need to win their respective conference tournament to reach the NCAA tournament? Rarely. But a Kansas State, a Mississippi State, a Florida State, often need to perform well in their conference tournament to reach the NCAA tournament. Can the toll a conference tournament takes on a team, three games in three days for the top seeds, hurt them when the NCAA tournament starts?

Amongst the past 14 NCAA champions, seven won their respective conference tournaments whereas nine won their regular season conference titles. Amongst those seasons there is one major outlier and that is the 2011 Connecticut Huskies. That team finished in 9th in the Big East regular season, went on to win their conference tournament and continue on to win the national championship. A better example that points to our thought is the 2012 Kentucky Wildcats, the Anthony Davis led beast of a team who only lost two games all season. One of the two games they lost was the SEC Conference Tournament Championship against Vanderbilt.

Want to pick a team that did weak all season but won the conference tournament? Choose carefully. You may have an outlier ready to break some brackets. And, you might just be that you are valuing success in the conference tournament when in fact we should devalue it. Whatever the case, the tournament will have luck and skill. So make your picks, hope to better your odds of a highly accurate bracket, and may the odds be ever in your favor. But, remember, it's called March Madness for a reason. Hang on to your seat!