4 NCAA Tournament Infographics To Prep You For The Final Four | Stat Geek Idol

posted in NCAA Basketball, NCAA Tournament, Stat Geek Idol

This is a Final Four submission in our inaugural Stat Geek Idol contest. It was conceived of and written by Gregory Matthews of Stats In The Wild (@StatsInTheWild). For more from Greg check out his posts from previous rounds.
Sweet 16 entryMeasuring The Madness: 2012 Tournament Off To Craziest Start Ever
Round of 64 entry
Predicting The Sweet Sixteen Using A Classification Tree

So far in this tournament, I have used a classification tree to predict Sweet Sixteen teams and developed a measure of madness for the tournament. These analyses were done on the team and tournament level, respectively. So I decided it would be appropriate for the Final Four to present a player-level analysis.

With the Stat Geek Idol National Championship on the line, I’m going to present the use of multidimensional scaling to visually represent the players who have played in the Final Four over the past 6 years.

First, let’s do two things: recap the tournament as a whole so far and look back at the season of each of the Final Four teams.

Tournament Recap

(click to enlarge)

It’s clear from the large amounts of gray and white in this year’s Cinderella plot, that this year’s Final Four has returned to a relatively normal level of madness. The Final Four will feature one number 1, two 2’s and a 4 seed. Not exactly an 8 vs 11 like last year or even a 5 vs 5 from two years ago.

(click to enlarge)

Through the Final Four, the madness coefficient is 26.43, indicating a modest amount of madness similar to 2001 (29.72) or, more recently, 2010 (28.06). In fact, 2001 bears a striking resemblance to this year’s tournament in terms of madness by round. 2001 started out with many first round upsets, including 15 seed Hampton over Iowa State; this year there were TWO 15 seed first round upsets (I may never get over how incredible this is — until a 16 beats a 1). The Final Four in 2001 featured teams with seeds 1, 1, 2, and 3; 2012 features teams with seeds of 1, 2, 2, 4.

So, while we may have gotten a taste of the madness in the second round (nee first round), the teams at the top ruined the madness for every Cinderella-lover out there. Hopefully, however, some of these top teams can help me erase the memory of the 53-41 national championship “basketball” game from last year in which UConn “prevailed” as they limped towards a National Championship. (Note: Teaching UConn students the day after UConn wins a National Championship after a season spent telling those students how bad UConn was that year (9-9 in the Big East!)……..was not fun.) So hopefully the lack of madness in the Final Four this year will result in a championship game that is actually worth watching.

Season Recap

(click to enlarge)

Using the TeamRankings predictive power ratings over the course of the season, I’ve plotted each team in the Final Four as a time series of their power rating, while indicating relevant games throughout the season, including regular season match-ups of the Final Four teams.

From the plot you can see that Ohio State started out the season as the highest rated of the four remaining teams. Kentucky briefly overtook Ohio State early in the season after a big win over Kansas. Ohio State regained the top spot among these four teams and greatly increased their power rating with a big win over Duke. Ohio State would come crashing back to reality with its first loss of the season to Kansas. Kentucky and Ohio State then battled very closely in the power ratings through the middle of the season with Kentucky eventually pulling away as the regular season neared its end.

Kansas was the lowest rated of these four teams for much of the early part of the season. They briefly surpassed Louisville after a victory over Ohio State, and eventually passed Louisville for good near the end of 2011 (even after a bad loss to Davidson on December 19). They remained largely steady for the rest of the season and briefly passed Ohio State towards the end of February, eventually ending the season just below Ohio State in the power ratings.

Louisville has had the most volatile season of the bunch. For a brief period of time in November, they had a higher power rating than Kentucky. Then they slowly began a downward spiral ending with a terrible loss (by 31 points!) at Providence who finished the season just 4-14 in the Big East. Clearly, this is when Louisville coach Rick Pitino grew angry. Over the next several weeks, Louisville steadily climbed in the power ratings culminating in a Big East Tournament Championship. Louisville had the largest gain in power rating in the weeks before the NCAA tournament of any of the four teams that remain. Louisville, like Ohio State, hopes to ride this momentum to avenge their regular season loss to Kentucky and earn a spot in the National Championship.

Multidimensional Scaling & The Players

Multidimensional scaling (MDS) is a method used for visualizing the similarities and dissimilarities of observations when many variables are measured for each observation. In this case, the observations are the players who have appeared in a Final Four game since 2007 and the measured variables are their regular season statistics.

MDS accomplishes this visualization by measuring the distance between all points in the data set, in this case all of the players, and then plotting the players in two dimensions that best explain the distances between them. Players who are similar to each other are plotted near one another; players who are different appear farther away from each other.

(click to enlarge)

All of the players from previous years appear in gray in varying size depending on how far the player’s team advanced. Previous champions appear as the largest gray font (Joakim Noah, Kyle Singler, Kemba Walker), previous finalists appear in medium gray font (Matt Howard, Travis Walton), and all other former Final Four players appear in the smallest gray font (Da’Sean Butler, Jamie Skeen, Hasheem Thabeet). This year’s Final Four players appear in color. Kentucky is blue, Ohio State is red, Louisville is orange, and Kansas is green.

In general, the further to the right you move on each of the plots, the more points a player averaged during the regular season. For instance, Kemba Walker, who is the furthest player to the right, averaged 23.54 points per game in the regular season, which is the most of any player to appear in the Final Four since 2007. Alternatively, Mareese Speights averaged only 3.93 points per game during the regular season when Florida went to the Final Four in 2007 (He average 14.16 PPG the following year) and he appears on the left side of each graph.

If you focus on the upper right of each plot, you see players like Tyler Hansbrough (twice), Greg Oden, Kevin Love, and Hasheen Thabeet, all big men who made their living in the paint. Alternatively, in the lower right you see players like Lee Humphrey and Joey Rodriguez, who tended to shoot lots of three pointers. Finally, on the right about half way up, there are players like Kemba Walker, Kyle Singler, and Da’Sean Butler.

As for this year’s players, the one thing that immediately jumps out at me is that Louisville doesn’t seem to have a player in the Hansbrough region of the graph. Kansas has Thomas Robinson, Ohio State has Jared Sullinger (who I keep calling J.D. Sullinger, but this guy ain’t no phony…..), and Kentucky has Anthony Davis. I could go on pointing out things to notice about these graphs, but I’ll let you stare at them and draw your own conclusions.

A Personal Note From Greg … And A Dig On John Calipari

When I was in junior high in the mid-nineties, my father and I had season tickets to UMass basketball. We religiously attended every home game at the Mullins Center (except the St. Bonaventure game; we gave those to my cousin as a Christmas present every year.) I remember watching the likes of Lou Roe, Donta Bright, Dana Dingle, and current head coach, Derek Kellogg dominate the Atlantic Ten for several years. The Minutemen were, of course, led by current Kentucky coach John Calipari.

I remember worshiping Coach Calipari as a kid. When I grew up I wanted to be just like him, and now I have my chance! We’re both in the Final Four this year, and we both have a chance to win our first National Championship (I consider Stat Geek Idol to be the National Championship of Stat Geek Writing). This will also, then, be my first chance to have a Final Four erased from my record — because just like Mr. Calipari, the only thing I love more than making the Final Four is having to forfeit my appearance later. Maybe Stat Geek Idol can make my dreams come true!

I’d like to thank TeamRankings.com for running Stat Geek Idol, and I’d like to wish all of the other contestants in the Final Four good luck. Good luck.

Remember, for more from Greg check out his posts from previous rounds.
Sweet 16 entryMeasuring The Madness: 2012 Tournament Off To Craziest Start Ever
Round of 64 entry
Predicting The Sweet Sixteen Using A Classification Tree

And for more from the other contestants, explore the Stat Geek Idol category on the blog!