Monday, November 2, 2009

They Grow Up So Fast

We didn't invent tempo-free stats. Neither did John Gasaway, Ken Pomeroy, or even Dean Oliver. Best we can tell, some guy who won a few games in North Carolina did. The fact that tempo-free stats aren't new should not come as a surprise. They're not terribly difficult to understand, and most are pretty easy to calculate as well. And moreover, they seem to illuminate productive basketball moreso than the traditional "per game" stats. So of course coaches (who tend to understand basketball pretty well) know about these things. Maybe that's why the tempo-free revolution that was to be televised has been met with a collective "ho hum" from the industry. Some of these stats already litter your local head coaching office, and probably did before Dean Oliver ever wrote his book. And this nugget, which I hope to illustrate in this post, is already something very well-known among coaches:

"The best thing about freshmen is that they become sophomores."

Al McGuire said that, and he was absolutely correct. As much as freshmen titillate with potential, they spend even more time frustrating coaches with uneven play. Freshmen commit more turnovers, miss more shots, are more timid in the offense, rebound less, etc. In general, freshmen are the worst class of player in college basketball.

But everyone reading this blog already knows that. The more interesting question is how much further back are freshmen relative to sophomores, juniors, and seniors. Or, put differently, how much we can expect freshmen to improve as sophomores? And how much we can expect sophomores to improve as juniors? And so forth with juniors to seniors.

And that's how I spent my summer (ok, I had an intern helping. And by "intern" I mean "college-aged family member who slept on my couch all summer.") -- looking at the statistical profiles of every BCS conference player who first stepped on campus between 2000 and 2005 (i.e., the last group's senior year was last season), with some additional filters:

No players with less than 10% Minute Percentage as a Freshman: In this study, the freshman year is the baseline number. We're measuring change, and that measurement is no good if you can't get a reliable relative point. Ten percent is arbitrary, admittedly.

Omit successive years with less than 10% Minute Percentage: Again, I didn't want small samples polluting the data.

Omit one-and-done players: Again, the measurement is change. A player has to give me at least one change data point to be considered.

No transfers: By and large, transfers not only change schools, but significantly, levels of competition. The most common transfer was from a BCS school to a low-level conference school. When the numbers improve, it's impossible to separate player development from the change in the competition level.

Conference games only: My affinity for conference game numbers is well-documented in this space. The idea is to try our best to hold the level of competition constant.


The big, overarching conclusion is this: a player shows the most improvement between his freshman and sophomore seasons than he does any other offseason. In fact, the freshman offseason improvement is, on average, greater than the improvement between a player's sophomore season and his senior season. That's not to say every player follows this pattern. There are lots and lots of exceptions, and this is no hard-and-fast rule. It's just a remark about the averages. And frankly, that's all we do around here, play the averages.

Pretty graphs below (going left-to-right is freshmen to sophomore to junior to senior):

The pattern is clear in all of the graphs, and perhaps most striking in the Shot% and TO% graphs, where a logarithmic growth & decline are shown (Keep in mind, this doesn't contradict Pomeroy's earlier work on shot percentage - that range is showing a pretty narrow band between 18% and 20.2%).

And now, to bring on home to why our readers should care. For that, I bring you the 2008-09 Fighting Illini. This team won 24 games after going 16-19 the year before. Yes, a lot of that was bad luck (or "bad DeChellis" if you prefer), but the Illini also lost 1/3 of the minutes from that 16-19 team. It's not like everyone came back. Another key piece in the Illini's resurgence is that 48% of the minutes were played by sophomores. In other words, 48% of the minutes were played by guys who figured to improve the most. And they did.

So what does that mean for this season? Well, nothing definitive, but we have started factoring this into our analysis of the teams for this season. And this graph should prove handy for others who want to do so as well:

(Note: In this table, I counted Alex Legion as a sophomore for the upcoming season, even though that's not technically accurate. Laval Lucas-Perry played in 5 games at Arizona, Legion played in 6 at Kentucky. It didn't make sense to treat them differently.)

As the graph shows, Illinois figures to be at the other end of the spectrum this season. Not surprisingly, Indiana laps the field. And maybe the people ranking Michigan and Minnesota (#15 and #18, respectively, in the initial coaches' poll) aren't so crazy after all.


Anonymous said...

Oops, posted to the wrong thread. Meant to post comment on greater relative improvement in big men as sophomores and not the Wisconsin preview thread.

chiefjd said...

extremely interesting post! is there any correlation between % of returning minutes and change in conference record?

Anonymous said...

Much thanks to you and your in-house intern for providing empirical support for the proposition that college players generally improve the most between freshman and sophomore years.

I wonder if you broke it down by position: guards and wings vs. big men. My guess is that there is a significant difference, and that big men experience an even bigger improvement than guards and wings after their freshmen years.

Big Ten examples last year: JaJuan Johnson, Andrew Jones, Mike Tisdale and Mike Davis. On the other hand, E'twaun Moore, Manny Harris, Kalin Lucas, Jon Diebler, Taylor Battle, Michael Thompson all improved as sophomores, but except for Diebler it wasn't as dramatic as the improvement in the sophomore big men above.

If this is true (and I think there are good reasons related to the bigger adjustment from high school play for big men), one would expect the biggest improvements this year from Minnesota's Sampson and Iverson, Northwestern's Mirkovic, Rowley, Shurna and Curletti and Indiana's Pritchard rather than from Michigan's Zach Novak and Stu Douglass, Indiana's Verdell Jones, Iowa's Gatens or Purdue's Jackson.

Anonymous said...

Hi, relatively new to the blog, but really appreciate it. you've done some great work.

One request: any chance, since you've apparently gone through the trouble of putting all the teams rosters into excel, that you could post that file? I'd love it if starters were somehow tagged, but just getting the data would be great.

I ask b/c, as a Michigan fan, I remain concerned about our lack of interior talent. But my impression from reading the previews is, despite the apparent uptick in the Big Ten's talent level, that in fact this specific weakness may not be as fatal as I once feared. It seems few teams will be able to trot out two classic post players, w/ Purdue and Wisc being notable exceptions. Anyway, could you discuss? Or post data so I can draw my own conclusions?

Thanks tons!

Josh said...

Anonymous (the last one),

shoot us an email at bigtengeeks AT gmail DOT com, and we'll be happy to send you the spreadsheet.

Off the top of my head, I think the following teams will start 2 players who probably are defined as post players:

Penn State

You mention Purdue, but I think Johnson is the only post starter there. And guys like Damian Johnson and Christian Watford are probably borderline, but they have a lot of diversity to their game.

Michigan will have to worry about the frontcourt, but they had to do that last year too, and this year they have a healthy 7-0 Ben Cronin to add. So if anything, things have gotten slightly better on that front. But I can't say it's a strength.

greyCat said...

If John Beilein's history at West Virginia is a guide, lack of talented (ie -- athletic rebounders and shot blockers) post players should not be a handicap to Michigan's prospects this season. Beilein employed a 1-3-1 zone with the Mountaineers that drove shooters either into the corners (low percentage shot areas) or lured them into the interior seams where the handler could be double and triple teamed. On offense his WVU teams were notoriously poor rebounders, often ranking in the lower quarter of the conference in both offensive and defensive rebounding. His #4s (especially) and #5s could as easily step out to the arc to take a 3 as they would back down a defender in the low post. The statistical profiles of his teams more resembled those of Georgetown (Princeton Offense), than they did those of the other conference teams. But they won, and won consistently. Very paradoxical considering the physical nature of Big East play.

Terrific piece BT Geeks. I enjoyed this one very much.

B. Lerner said...

This is very nice.

I did a related analysis last month - looking at the relationship between possessions used and offensive rating for successive years - but this was a bit flawed since I didn't segregate by player year.

It looks here, and also from some data presented by Dan Hanner in his Notre Dame preview in the new College Basketball Prospectus, that most freshman improve tremendously through their junior season, but the difference between juniors and seniors is marginal.

At least, that's my take away

greyCat said...

Saw your preview pieces over at Hoya Prospectus. I guess Dante Cunningham proves the exception to the rule!