Tuesday, November 20, 2007

News: Mid-week Roundup

Thanksgiving week has been slow, but here are a few items:

  • Speaking of Louisville, the loss of David Padgett for at least 10 weeks will be a big test for the Cardinal players and Coach Pitino. Padgett quietly lead the nation in Offensive Efficiency last year (134.3, using 17.8% possessions in 58.9% minutes). To put him in Hoya context, Padgett's offensive value last year was roughly equivalent to the 2006 version of Darrel Owens, obviously with a different skill-set. Defensively-challenged Derrick Caracter and offensively-limited Terrance Farley will be nominally asked to take up the slack, but I think it's much more likely that Pitino will revert to what he knows best, namely going small and having his team take an enormous number of 3-pt shots.
  • While the amount of fawning I normally direct toward Ken Pomeroy and John Gasaway can be sickening, I have to point out a couple of posts on their website:
    • Gasaway finally posted his Big East Preview (Overview, Part 1, Part 2). You can read it for yourselves, but I just wanted to note that his main stipulations for Hoya success this season (improved def. rebounding and fewer turnovers) gibe with my own view. Both are better so far, but it's only 2 games in.
    • Pomeroy posted a strange article today about rating coaching performance; the premise of the article (that he can quantitatively evaluate coaching) is based upon the idea that all things regress toward the mean over time. Specifically, bad teams get better and good teams get worse the next season, so that "teams get pulled towards the .500 mark over time." This is based upon work Dean Oliver did with NBA teams, and Pomeroy applies this to college teams (with respect to their conference foes, not all of Div I). This strikes me as counter-intuitive for a couple of reasons. Certain teams are historically good (e.g. Duke, UNC, Kansas) or bad (e.g. Rutgers, Florida State, Colorado), at least in the short- to mid-term, and use this as a recruiting advantage within their conference. More importantly, the high turnover of talent in college seems likely to swamp almost any other effect. To this end, Dan Hanner (HP's new favorite blogger) has put together his own system ranking coaches, which is more interested in incoming talent and post-season results. It provides a nice alternative take to Pomeroy, who hopefully will come back to this idea of coaching evaluation again. Edited to add: Sure enough, Hanner chimes in on Pomeroy's article.
Finally, Ray Floriani is back. He sent me an e-mail yesterday with his tempo-free look at the Memphis-UConn 2k Classic Final; G'town will see both teams relatively early in the season.

NEW YORK CITY – The 2K Classic final turned out to be a gut check for both teams. Memphis stopped UCONN 81-70 in a hard fought contest at Madison Square Garden. The third ranked Tigers sprinted out of the gate and built an 18-3 lead. Slowly, possession by possession, UCONN battled back. The Huskies took a 41-40 lead into the locker room at the half.

The second half saw a tightly contested affair with the Tigers gradually pulling away over the final eight minutes.

  • eFG%: Neither team opted for threes (Memphis 3 of 13 and UCONN 1 of 5). The significance here lies in two point domination as Memphis, between size and penetration, was effective in the paint.
  • OR%: Both clubs hit the boards hard but allowing 48% of the possible rebounds to be gathered by the offense was just too much for UCONN to overcome.
  • TO Rate: Close and no shock as both teams were more than willing to push the ball.
  • FT Rate: UCONN’s ability to get to the line greatly aided the Huskies first half comeback. It also contributed to coach Cal’s first half technical. Basically, UCONN began running their offense and boxing out on the boards. These two factors contributed greatly to their ability to draw fouls. Actually this figure, though high, keeps in line with the three UCONN games prior to the final. Over those first three games the Huskies ft rate was 36%, which shows a good ability to get shots, or rebounding position, that can get them to the line.

Pace 80

Eff. 87.5

eFG% 40.6%
TO% 21.3%
OR% 18.8%
FT Rate

Stats courtesy of KenPom.com


  1. When I read Pomeroy's article the first three questions that came to mind were:

    1. Did Pomeroy read Hanner's analysis from last spring, and are they in fact attempting to measure the same thing?

    2. Regression to the mean may be useful when players and coaches work with relatively fewer time constraints (length of season, practices, eligibility, volatility within playing and coaching ranks...), but is it appropriate in a D1 environment when annual turnover is, relative to the pros, much higher for both coaches and players?

    3. When will the Hoya Prospectus run the numbers for the Big East coaches?

    Forging a consensus on what makes a coach great is further complicated by the sheer breadth of programs and conferences in D1 basketball give rise to a multitude of coaching circumstances. A winning season in one environment may be as worthy of mention as a conference championship in another. I suspect that is why fans rarely move beyond the basic metrics -- number of wins, wins per season, winning percentage, number of national championships, appearances in the Final Four, etc.

    The task of identifying a great coach is akin to Potter Stewart's description of pornography -- can't define it, but I'll know it when I see it.

    Have a good Thanksgiving.

  2. Mmmm . . . pornography.

    What were we talking about again? Oh yeah, Pomeroy and coach rankings.

    I didn't feel all that great about running down KenPom's work, since I certainly don't have something better to present on my site (and no, I don't plan on running his model on Big East coaches).

    My concern is really that he's using something developed by Oliver to look at the NBA and apply it to college. One thing that the NBA has going for it that forces regression towards the mean is the draft, where bad teams are rewarded with (likely) more talented young players than good teams. No such mechanism exists in college.

    Hanner and a Pitt(?) blogger, who's post I can't find, both made the point last year that McD AAs (i.e. who has the most talent) are among the best predictors of tournament success, and, likely, performance as a whole. So, to me, any coaching evaluation tool for college would have to account for talent variability year-to-year.

    Happy Thanksgiving back.