(Guest post by Shoeless Jack)
With the NCAA tournament starting on Thursday, there’s two main questions people are
preoccupied with: who is going to win the tournament, and what lower seeds project as possible
upset candidates? Fortunately, there are several analytical tools that can help predict these
outcomes, yet they are often overlooked by the casual fan.
Predicting the Champion
Let’s start with potential tournament winners. If you look back over each tournament this
century, you’ll see that there are certain characteristics that almost every champion has. An
overwhelming 21 of 23 of the past winners have ranked in the top 10 in adjusted offensive and
top 25 in defense according to KenPom.
I’m guessing that mentioning KenPom may have just lost a portion of the audience who checks
in with college basketball at the beginning of March to fill out their bracket. Simply put,
KenPom is an analytics site that uses an algorithm to identify not just how well teams perform,
but also take into consideration who they’re playing against.
To give you an idea, last season Kansas finished the season 6th in adjusted offense and 17th in
adjusted defense. The year before, Baylor was 2nd in offense and 22nd in defense. In fact,
Baylor was the worst defensive team to win the championship besides the two outliers in this
Who were the two outliers you may ask? I’ll give you a hint: they both came from the same
school, and were guided by star guards who went on a crazy postseason run.
That’s right, the UConn championship teams in 2011 and 2014 are the lone exceptions, but were
still elite defensively.
So with this simple data point in mind, a significant portion of the bracket is eliminated from
contention. At this point, the only team who ranks in the top 10 in offense and top 25 in defense
at KenPom is UConn, who is 6th on offense and 18th in defense.
This doesn’t mean that they’re the only team that can win. These numbers are liable to change
due to tournament performance. Houston is 11th in offense and 4th in defense, whereas Purdue is
10th in offense and 26th in defense. It is likely that both of these teams can fall into the 10/25
category if they make deep runs.
While Purdue and Houston project as possible 1 seeds to win, the other two don’t have as bright
futures. Kansas is 28th in offense and 7th in defense, while Alabama is 20th in offense and 3rd in
defense. It’s possible that they could improve enough offensively, but it’s unlikely.
So what that all means is that based on this one metric UConn, Purdue, and Houston are by far
the most likely teams to win the tournament.
There are other significant factors that most champion teams have, such as 2 NBA-level ball
handlers, and 5 or more top 100 recruits. Both UConn and Houston have NBA guards, but
Houston does not have more than 5 top 100 recruits. All this considered, UConn seems to be the
best choice to cut the nets in April.
So I’m not sure if you guys know this, but you can sort the data on barttorvik.com to show you
how teams have performed since certain dates. And oooh baby I’ve been sorting the data.
Since February 12th (a month before Selection Sunday), some interesting trends have appeared
that help predict underrated teams.
Drake is 8th in the country over the last month, Utah State is 15th, Texas A&M is 12th, and
Memphis is 9th. All of these teams have been overperforming relative to their seed, and project
as good upset picks.
Some “elite” teams are also coming into the tournament hot. Gonzaga is 1st, UConn is 2nd, and
Duke is 14th. Using the last month to track how teams are doing allows us to see who is peaking
at the right time.
I’ll wrap this up by reminding everyone that there are a few teams in the field who are horrible
defensively and because of that likely have a capped upside. Baylor is 104th in defense, despite being 2nd in offense. Similarly, Gonzaga is 1st on offense but 76th in defense. However, as I just mentioned, Gonzaga is coming in playing their best basketball, but they don’t have the defense required to win it all.
With all this said, it’s called March Madness for a reason, so I’m expecting the unexpected. But
the numbers say UConn is the most likely winner, followed by Houston.