A little while ago I extracted three point, two point, free throw and total field goal numbers for each team in the Big Ten from 2002-2011. I sat on the data as I watched the end of the regular season, and then conference tournaments, finish. One of the things I enjoy thinking about is the role of luck in college basketball, but in all sports. In baseball, there's been advances that help strip away luck/good fortune/whatever you'd prefer to call it. For those that don't know, I'm a baseball junkie as well, and have written for a few different outlets (including penning two articles for the Detroit Tigers annual that was put out recently by Maple Street Press. Go buy it!).
One of things that strikes me as 'lucky' or fortunate or whatnot in college basketball is three point shooting. We can all agree that Jon Diebler is a very good outside shooter. Can we also agree that Jon Diebler isn't really good enough to be a true-talent 50% shooter from three-point land, as well?
One of the banes of baseball analysis -- and sports analysis in general -- is teasing out true talents from observed performance. Sure, Ohio State's team three point percentage of 41.3% goes down as the best among Big Ten teams since 2002, but are they really a team good enough to hit 4 of every 10 long balls? Doubtful.
That's where regression comes in.
Tom Tango, consultant to nearly every NHL team, a few MLB teams, author, and noted advanced baseball statistician blogger laid out regression very well in a post on his blog here. His post was on NBA free throw percentage and how much skill exists in that (a lot). I may appear to somewhat know what I'm talking about mathematically, but it's really more of a smoke screen. I didn't possess the knowledge to do something that turns out to be relatively simple, provided you've got a spreadsheet handy. Tom's post sparked my mind which prompted me to grab the shooting data for the Big Ten teams as far back as I could get them. It also sparked Matt Snyder, author of To Maroon and Gold, a blog on the Calvin Knights -- a D3 school in Michigan -- who ran a regression to estimate true talent team three point defense. (You may also know Matt from his wonderful baseball writing at Motor City Bengals on the Tigers.)
So moving onto the actual data. I'm using full-season data I took from ESPN for the 2001-02 season through today. I've followed the post Matt laid out and I'm taking a look at how skilled a team is at shooting the three ball.
First up, the top 10 three point shooting teams in the last 10 years:
If anyone's seen Ohio State in the last two months, you would've guessed they were phenomenal from three-point land. I don't know what's more surprising: their high percentage of made three's or the fact that they still shot so well and shot more three than all but three other teams in the top ten.
Now, these are what are in the record books as tangible evidence to what did occur. But, I submit to you that these aren't their true talents. In fact, in looking at this data, three-point shooting has to be the most heavily regressed of the three basic shooting functions (three point, two point, and free throw).
But how much? Well, in any given season you should regress a teams three-point exploits about 64% towards the league mean. To put that into context, you should regress a teams two-point shooting about 29% toward the league mean and a team's free throw shooting about 26% towards the league mean.
Even further, the team that has taken the most three point attempts in my data set was the Michigan Wolverines in 2008-09. That squad jacked up 859 three's -- just one more than the 2009-10 Northwestern Wildcats. No one else even attempted more than 760 three's in a given season out of the 110 team seasons that I have.
So the 2009 and 2010 Michigan and Northwestern teams are clear outliers giving more attempts in this data set. And we still have to regress them 54% towards the league mean. To get to the 50% regression barrier, a team would have to attempt around 1010 three's.
Now that we've established the level of regression necessary, how far are we dropping the great Buckeye's shooting this year? Quite a bit. Here are the top 15 teams in regressed three-point shooting (r3P%):
Well, this years Ohio State team is still the best three point shooting squad in the last ten years, but the gap has closed. Now, they are marginally better than the great 2005 Illini squad that featured Deron WIlliams, Dee Brown and Luther Head. (Those three shot .364, .434 and .410 from three point range, respectively.)
Our 2003-04 Spartans squad ranks fifth best in the conference in the last ten years. That team was, of course, led by Paul Davis' 16 points, 6 rebounds, 2 assist, 1.5 steals mark. Chris Hill was the second leading scorer knocking down 45% of his treys (84-for-185).
The 2001-02 club ranks seventh. Not that you need reminder, but Chris Hill lit it up from outside and Marcus Taylor led the team in scoring at over 16 per game and dishing out over 5 assists per contest.
What teams were docked the most and the least in this regression? Glad you asked. The five leaders and trailers are:
This years OSU and the 2003-04 MSU team were docked the most. The current iteration of the Iowa Hawkeyes were actually "unlucky" from three this year. Their actual three point shooting improves from .314 to .340. But, the 2002-03 Hawkeye squad featuring Chauncey Leslie and Glen Worley as their leading scorers under performed by about 4.4%.
Here's the entire 110 team seasons in a handy google doc if you'd like to peruse at your own leisure. It has team, year, three pointers made (3PM), three pointers attempted (3PA), three point percentage (3P%), regressed three point percentage (r3P%) and the difference between the regressed three point percentage and the actual three point percentage (Delta). It is sorted by regressed three point percentage.
I plan on looking at two point, three throws, and then flipping it to the defensive side of things, as well as individual regressed shooting.