Groundhog Day: Can It Predict March Success?

Posted by rtmsf on February 2nd, 2011

Today is Groundhog Day, and with much of the nation under snow, ice and feeling a lot like the inside of a commercial freezer, it may seem very hard to believe, but Punxsatawney Phil actually predicted an early spring this year.  Whether his prediction about March comes to pass is debatable, but it got us thinking that the celebration of his day makes for a good assessment point to see if what we think we know at this point of the season passes muster in less than six weeks.  Generally speaking, do the teams ranked highly on February 2 do well in March?  How predictive are the major polls today compared with what ultimately happens next month?  We’re not sophisticated enough to run high-level regressions on this stuff, but we were able to eyeball some of the numbers and come to some basic conclusions below.

He Can Predict Weather, But Can He Predict Hoops?

The first thing we did was look at the AP, ESPN/USA Today and the RPI ratings on or about Groundhog Day for the last three seasons.  Notably absent are the Pomeroy ratings, but to our knowledge, he doesn’t keep historical daily archives available for public consumption.  So we’ll deal with what we have.  We then averaged the top sixteen teams using those three metrics and then compared them with their ultimate season outcomes (NCAA Tournament seed; Sweet Sixteen appearance; Final Four appearance).  It’s a rudimentary analysis, but as you can see, below, Groundhog Day seems to be a fair to good predictor of March outcomes.  Here are the last three years worth of data:

As you can see above, eleven of the sixteen teams ranked in the aggregate top sixteen ended up getting a top four NCAA seed, and a full half of these teams made it to the Sweet Sixteen.  This year was the only incident in the three-year window where a Final Four team (Butler) came from outside the Groundhog Day top 16.  On to 2008-09:

Again, the magic number is eleven of the sixteen teams earning a top four NCAA seed, and nearly all of those eleven moving on to the Sweet Sixteen.  Four of the remainder were Final Four teams that year.  Now, a look at 2007-08:

Looks pretty similar to the others, right?  Eleven teams as a top four NCAA seed, ten to the Sweet Sixteen and four of this group to the final weekend. 

The three-year totals show that in an average year we should expect roughly two-thirds of the top sixteen on Groundhog Day to remain in the upper echelon by Selection Sunday, and a very high number of those to also advance to the Sweet Sixteen.  Additionally, if you’re looking for a Cinderella to reach the Final Four, you may not want to go too far down the rankings at this point.  The only team outside the top sixteen to make it was Butler last season, and the Bulldogs were a near-miss, sitting at #23 (AP), #15 (ESPN/USA) and #18 (RPI) for an average ranking of 18.67.   

So what can we learn from this?  If recent history (and a small sample size) is any indication, we should be able to look at the below table of this year’s current top sixteen teams and safely conclude that most of them are in position at this point for a pretty good March.  The difficult part is figuring out which teams are due to tank, keeping in mind that only five teams in the last three years fell from the aggregate top sixteen to a #7 seed or below.  And the worst such drop was still only to a #9 (Butler again, in 2007-08) so it’s not like any of these teams are in danger of missing the Dance. 

The true lesson here might be to resist some of the johnny-come-latelies who get hot in February and make a run through their conference tourneys.  By and large, the season’s best teams are already settled by Groundhog Day; it’s usually just a matter of placement.  Going along with the numbers, if you were going to pick eleven of the above teams to make the Sweet Sixteen, who are the five that you’d leave out?  And who would you replace them with?  Here’s our list: Louisville, BYU, Kentucky, Texas A&M and Notre Dame would not play into the second weekend.  We’ll replace those five with: Syracuse, Wisconsin, Arizona, Wichita State and UNLV.  We’ll check back in seven weeks to see how we did.

rtmsf (3998 Posts)


Share this story

2 responses to “Groundhog Day: Can It Predict March Success?”

  1. rb says:

    Would actually like to see it for 5-10 years before this. If we remember 2008 and 2009 were very favorite heavy years (4 #1s making it in 2008 and very few upsets in 2009 if I remember right). Anyway, I know you guys are busy, but I’d be curious to see it.

  2. rtmsf says:

    RB – yeah, I wanted to go back five yrs, but I can’t get RPI ratings on specific dates (like Groundhog Day) during the season any further back than 2008, so I stopped there. I wasn’t comfortable in just using the human polls prior to that, although now that I think about it, they’re not all that dissimilar than the RPI at this point in the season.

Leave a Reply

Your email address will not be published. Required fields are marked *