The NCAA Tournament: Where Matchups Matter (Part 1) | Stat Geek Idol

***NOTE*** This is a winning Round of 64 entry in our inaugural Stat Geek Idol contest. This article was conceived of and written by Martin Manley of Upon Further Review. The opinions or predictions expressed below do not represent the views of TeamRankings.com, and are solely those of the author.

For some time, I have been trying to figure out how to apply team statistical characteristics to the NCAA tournament. It’s not good enough to simply say Team X has been better throughout the season than Team Y. Therefore, Team X should go farther in the Big Dance. It’s more complicated than that.

Here is the crux of the issue. Suppose Team X is the best in the nation in field goal percentage while Team Y is the best in the nation at creating steals. If one were to give both categories equal weight, they wouldn’t know which team was better. But, we all know any team excelling in FG% is more important to winning than excelling in steals.

Given there are many statistical categories, it’s critical to recognize which ones are the most correlative to winning. Once that has been determined, it’s hypothetically possible to determine which teams are set to do well in the tournament.

But, even then, there is the question of matchups. It’s the age-old question. Do strengths for one team get negated by the strengths of its opponent?

What Stats Are Most Important For Regular Season Success?

The 10 major statistical categories are FG%, 3FG%, FT%, Rebound Margin, Turnover Margin, Assists/FGM, Ast/TO ratio, Steals per game, Blocks per game, and of course, Defensive FG%.

I decided that to truly determine the value of each category, I would take a look at the top-5 teams in the final AP poll over the last 10 seasons. That’s 50 team seasons. By looking at all 10 categories for all 50 team seasons, I should be able to determine which ones correlate best to winning.

The top table in the below image shows the rankings for each of the top-5 teams for 2012.

(click to enlarge)

The order of the categories above is shown (left to right) by the most correlative to winning. In other words, Assist/TO ratio is the most correlative with Offensive and Defensive FG% being second and third.

This is, of course, only one year and only five teams. Thus, the sample size is far too small to depend upon.

Therefore, the bottom table shows the ten categories and the average ranking for the top-5 teams for the last 10 years. There are approximately 330 teams per season, so you can see that every category is correlated to winning to some degree. To view all ten years for the top-5 in each year, see my blog.

It’s pretty clear that Assist/Turnover ratio is the most important metric. One would assume this to be the case because ATO likely represents how efficient is a particular offense.

The main difference between 2012 and the combined 10-year period is Missouri. You will notice the Tigers are the exception, rather than the rule on a number of categories. Thus, when one of five teams, MU’s stats skew the correlation, but when one of 50, far less so.

How Can We Use This Info?

There may be a great deal that can be done with this type of information. Most rating systems (if not all) are predicated upon the margin of the game, the opponent and where the game was played. That’s pretty basic, but it’s also pretty encompassing.

In one sense, if Team A beat Team B by 20 points on the road, does it really matter how they accomplished it? Do we care which statistical categories they did well in and which they did not?

In terms of a power rating system, perhaps it doesn’t matter. But I’m more interested in how these categories affect the NCAA tournament. Simple power rating is fine, but is it enough?

Almost everyone, including myself, would agree the tournament is about matchups. What has not been definitively addressed in my opinion is what constitutes a matchup “problem” and which of those are the most important.

My goal in this series is to discover, if possible, what causes upsets and whether those upsets are the result of unique and identifiable matchup issues.

Probably the single most hands-on sporting event in the United States is the NCAA tournament. More people fill out the brackets than anything I can think of. We’ve all been told a hundred times that just picking the favorites could never win because it never happens that way. So, you must pick upsets to have a shot.

That’s not true, of course. Whereas the odds of being perfect or even close to perfect are astronomical if you pick all the favorites, it’s even more astronomical if you try to pick upsets. There will be upsets, but your odds of knowing them ahead of time are too slim to calculate.

Nevertheless, it’s widely assumed that there will be lots of people – especially people that don’t pay a lot of attention to the sport otherwise, that will be filling out brackets and they will just mark the higher seed in all (or almost all) cases. So, to be different and to be able to boast after the fact that you picked this and that team when nobody else did… you have to pick some upsets. Besides, that’s what makes it fun.

This series is designed to try to figure out how to do that using the major statistical categories that are widely available. By the way, most of these categories were taken from the NCAA directly, but I also utilized Ken Pomeroy’s data base.

In Part II, I’ll be looking at teams that have made the Final 4 in the NCAA tournament and try to determine if there is something unique about them as opposed to the top-5 teams in the regular season.