Ever since we remodeled the basement here at GK Central, we have had more room to work on advanced mathematics and take a look at some questions we’ve been meaning to address for a long time.
For example, our statistical model combs through historical data to find the traits common to the teams that pull off or succumb to big NCAA tournament upsets. But we’ve always wanted to find ways beyond citing the same old key metrics to get across how those teams play. And now, with help from our Giant Killer colleagues at Furman University, we have developed a new toy for doing just that: similarity. For any given team, our similarity method shows the five most comparable Giants or Killers from our historical database (which dates back to 2007), measured by what matters to Giants and Killers. Knowing a program’s most similar squads, you can get a better feel for its style -- and how it will fare.
The basic math behind this is pretty straightforward. (But feel free to skip this paragraph if you prefer to hide your vegetables under the mashed potatoes.) Suppose that on a graph, Point A is 3 inches away from Point B horizontally, and 4 inches away vertically. Using the Pythagorean theorem we all learned in high school, we can figure out pretty quickly that the distance from A to B along a straight line is 5 inches. In the same way, we can combine distances on any two metrics and see how “close” two teams are. For instance, Oklahoma plays at a quick tempo and has faced a tough schedule this season. If we look only at those two scales, the most similar teams to Oklahoma in 2016 -- other programs that have played at both a fast pace and against strong opponents -- would be UCLA, Wake Forest and Iowa State. Our similarity method extends this technique to measure the distance between teams while balancing all of the shooting, rebounding, turnover and pace statistics that define Giants and Killers. And then it looks for each team’s closest neighbors.
Now that the basic math lesson is out of the way, here’s what caught our eye once we crunched the numbers. First, to the Giants: