Efficiency and Your 2011 NCAA Tournament Bracket

Many people have many different reasons for picking their favorite day of the year. Some people like their birthdays because it's all about them. Some people like Christmas and the spirit of giving. Others like Thanksgiving because of the perfect combination of food, family, football and being able to pass out drunk in front of everybody without anybody caring.

While Thanksgiving is and always will remain my number one day of the year, a very close second is the first day of the NCAA tournament. I have taken the first Thursday and Friday of the tournament off from work every year since I entered the workforce. Now that I have the package from DirecTV, I sit on my couch, clean bracket in my lap and spicy Bloody Mary in my hand, and bask in the glory of the single greatest sports event in the world.

I have many goals in life. I want to run communications at a major Fortune 500 company. I want to avoid getting a divorce and raising kids who hate me. I would like to retire before I reach 75.

And I want a god-damned perfect bracket.

As far as being a sports fan, the perfect bracket is the ultimate Holy Grail, even greater than a perfect season in fantasy football. To finish Championship Monday with not a single X on your sheet ... wow. Just wow. That's all I can say.

But I'm not just one to want and dream. I'm convinced that if I crunch enough numbers and look at the stat sheets long enough, I'll figure out just the right mixture of numbers to get me there. Of course there is always going to be luck involved, but for luck to make you perfect, you have to get most of the way on your own.

One of the stats I was interested in exploring was offensive and defensive efficiency and which created a more effective advantage in a given game. Using the data provided so generously by Ken Pomeroy on KenPom.com, I found that neither really created enough of an advantage to use for predictive purposes. Teams with an advantage in offensive efficiency have about a 60 percent win percentage (58-38) in the first round over the past three years, while teams with a defensive efficiency advantage have a 66 percent win percentage (63-32).

If you combine both offensive and defensive efficiency ratings, you get a higher success rate. Teams with both an offensive efficiency advantage and defensive efficiency average have an 80 percent win rate (36-9) over the past three tournaments. Of course that wouldn't have helped you pick Northern Iowa over Kansas last March, but not many stats would have.

To take it a step further, I decided to calculate the efficiency split for each team, subtracting the defensive efficiency (lower the better) from the offensive efficiency (higher the better). The reason here was to identify the truly dominant teams, splitting them out from the all-offense-no-defense teams with high offensive efficiency ratings but equally high defensive efficiency ratings, and grind-it-out clock killers who will have great defensive efficiency ratings but sub-standard offensive efficiency ratings.

The efficiency split turns out to be a pretty good barometer of success. If you just used it as a single stat, you would have had a 73.96 percent success rate in the first round of the past three years. That's not great, but it's better than almost all other single-stat success rates. (Scoring margin is the best single-stat predictor at nearly 74.74 percent.)

Where efficiency split turns out to be perhaps even more valuable is in determining the potential for a deep tournament run. Of the 69 teams over the past three years to come into the tournament with an efficiency split at 10.0 or lower, only two made it into the Sweet 16 (Villanova in 2008 and Arizona in 2009, both as 12 seeds who faced 13 seeds in the second round) and none made it into the Elite Eight. Fifty-one of the 69 went out in the first round, and 14 of the 17 teams to come in with an efficiency split under 5.0 lost in their first game.

On the other side, teams with a 20+ efficiency split have a much higher ceiling. Only one team in the past three years with a 20+ efficiency rating has lost in the first round (Utah State last year under the Texas A&M rule*). The other 16 teams with 20+ efficiency splits includes the last three national champions, 2008 runner-up Memphis, three other Final Four teams, and two Elite Eight teams. So when you're looking at potential Final Four teams, these should be the first teams you look at.

(*The Texas A&M Rule: No matter what the stats might say, don't take a team of over-achieving short guys from a mid-major conference over the Aggies. In 2008 and 2009, BYU ruled the stat sheet over A&M and lost both times. In 2010, it was Utah State with the numerical dominance, but a first-round loss. If A&M ends up in one of those match-ups again this year, consider yourself warned. They've also lost their second-round games each of those years, so keep that in mind, as well.)

Looking toward the 2011 tournament, it's still way too early to actually start using the numbers. Because of the wide disparity in non-conference schedules, several teams have astronomical splits that are likely to shrink considerably over the next few months (Ohio State is not likely to finish at +40). But if you figure that teams in the BCS conferences will have their splits drop as they face tougher competition, while teams in non-BCS conferences will likely see theirs maintain or even rise, it is worth noting that 11-1 Minnesota's efficiency split is a paltry 13.4, good for only 69th (nestled comfortably between Colorado and UTEP), while BYU's 11th-best 27.3 bodes well for a next-level run. (Lots of reasons to like BYU this year, particularly their similar growth pattern to the 2010 tournament Cornell and Butler squads.)

Of course it's always possible I'll do all this research and still wipe out to some Mary in the office pool who picked squads based on uniform color, but I don't care. One of these years I'm going to nail it. And when they interview me like they did that Autistic kid last March, and they ask me how I did it, I'm going to have a damn good answer.