Sunday, March 11, 2012

Picking a Bracket: My Strategy (Stats, Pomeroy, LRMC, & Experts) - Part 1

Check out my 2014 updates HERE

Check out my 2013 updates HERE

Everyone has their own method for picking a bracket, whether it be based on jersey colors, mascots or extreme in-depth statistical analysis.  I by no means will put myself in a category with the likes of a Ken Pomeroy or anyone with a statistics background, but I like seeing what the numbers can tell us. There are a few key strategies and data points I use that have been helpful when picking my bracket, so I'm going to outline them for you here. 

Top 10 Stats Rank
It was after Florida's 2006 championship when I was looking over the team statistics of the regular season and noticed that Florida had led the nation in FT%.  I then took a look back at a couple previous winners (UNC 2005 and UConn 2004) and they also had led the nation in a statistical category as well (Note, when I first discovered this, it was using the stat categories ESPN had listed, so I have maintained this as my list for consistency.).  I have gone back as far as 2002 and over the past 10 years, the national champion has been in the top 10 of at least one of the the listed categories.  Until recently, the majority of National Champions were in the top 10 of at least a couple categories.

Here is a chart of where they ranked, but note I only included their rank if it was in the top 10.
Sheet10

YearChampionPts/GmReb/GmFG%FT%3Pt%Ast/GmStl/GmBlk/Gm

2011Connecticut10

2010Duke9

2009UNC222

2008Kansas10327

2007Florida110

2006Florida1

2005UNC111

2004Connecticut1861

2003Syracuse72

2002Maryland48926

Once the bracket has been set, we'll revisit which teams are the top 10 in these categories.

Ken Pomeroy + LRMC + Elite 8
I have been following the stats and methods of both of these for quite some time now.  Pomeroy's offensive and defensive efficiency (OE & DE) stats continue to become increasingly popular and the LRMC model was highlighted on CNN a few years ago.  I have built profiles of the Elite 8 teams to help narrow down a pool of possible teams in which to consider for making it to the Elite 8.  Unfortunately, I only have Pomeroy's pre-tournament stats and the LRMC ranking going back three years.

Focusing on Pomeroy: Over the past three years the teams who reached the Elite 8 and improved their offensive or defensive efficiency ranking, did so with an average of +9 OE and +13 DE.  Applying those to historical post-tournament rankings for estimates, here are some Elite 8 profile highlights since 2003:

  • Every team has had at least an OE or DE rank less than 60
  • 71 of 72 teams had a pre-tournament OE less than 60
  • 65 of 72 teams had a pre-tournament DE less than 60
  • Pre-tournament stats over the past three years (2009 - 2011):
    • National Champions Averages: 8 OE and 23 DE
    • Final Four Teams Averages: 26 OE and 35 DE
    • Elite Eight Teams Averages: 25 OE and 30 DE
    • 16 teams (66%) featured an OE and DE equal or less than 40
Focusing on LRMC: Dr. Joel Sokol, an associate professor at Georgia Tech, has developed a predictive model for the NCAA tournament.  These are some highlights of the Elite 8 teams from the past three years:
  • 20 of 24 teams had ranking lower than 25
  • 10 of 12 Final Four teams had rankings lower than 25
  • 22 of 24 teams had ranking lower than 30
  • National Champions Average: 7
  • Final Four Teams Average: 20
  • Elite Eight Teams Average: 17
We'll revisit the list of potential Elite 8 teams after each of these respective sites make updates following the games today.

Expert Picks
Many years ago I came across an article in an in-flight magazine about picking the Oscars.  It was a study by Harvard (I think) and it basically said that when the majority of people made the same pick, it usually ended up being the correct pick. Of course now I can't find the article to reference, so you'll just have to take my word for it.   All that said, I figured I would try to apply the theory for bracket picks.  In short, it's basically like aggregating the knowledge of the top college basketball minds of ESPN, SI, Yahoo! and CBS.  This has proven to be the most helpful for guidance in some of the 8/9, 7/10 and 5/12 match-ups.

Once the expert brackets start to pop up, I'll be tabulating their picks and provide links to their brackets.

No comments:

Post a Comment