Saturday, March 21, 2009

Pace and Turnovers

No great insight in this post, I'm afraid, just a data dump.

First, a look at Georgetown's raw pace, both in all games and in conference regular season games only, during the JT3 era:
Year All BE
2005 60.2 59.9
2006 59.0 58.6
2007 59.7 58.9
2008 62.1 62.3
2009 64.2 63.5

Next, back when John Gasaway was posting at Big Ten Wonk, he introduced a stat he called ePoss, or "effective possessions" which is simply a team's pace (possessions per 40 minutes) less the percentage of possessions on which they turn the ball over. Here's how those numbers have looked for the Hoyas:
        All games         Big East
Year ePoss TO Rate ePoss TO Rate
2005 47.0 21.9 46.6 22.3
2006 47.9 18.8 47.6 18.6
2007 46.6 22.0 45.6 22.6
2008 48.9 21.2 48.7 21.9
2009 50.3 21.6 49.5 22.2

From both sets of numbers, it's clear that the Hoyas played faster this year than they did last year, and that last year was faster than 2007. 2007 stands out, in ePoss terms, as a clear aberration in terms of being slow-even slower, in fact, than the nominally slower 2006.

If/when I come up with something interesting to say about these numbers, I'll put up a post accordingly. Other reader(s), feel free to come up with potentially interesting comments of your own.

UPDATE (3/26/09 2215 CT): Per request, I've updated the table with TO Rate for each year, broken down between all games and just BE conference play:

I'll put up a separate post with the numbers for Hoya opponents.


  1. Hey Tom,

    Could you add turnover percentage for each year to the ePoss table? My first reaction was that the 2006 team didn't turn the ball over much compared to other years.

    I haven't talked about it since early this season, but one thing I do want to add to the HD box score over the summer is average time per scoring possession, which I think would be even more useful in understanding what kind of pace a team wants to play.

    The underlying goal is really the same as Gasaway's, in that I'm trying to understand how long a team wants to run its offense each possession. I figured I'd just look at scoring possessions, since they can't include turnovers, obviously. I wonder if it would just be better to look at average time of non-turnover possessions?

  2. In that analysis, I think you'd want to slice and dice a few ways.

    Namely, there's an inherent bias in scoring possessions to shorter possessions. Fast breaks should score at a higher rate, for example.

    But let's also face it -- you shoot faster versus poorer defenses. If the defense is playing well, you're likely to take longer.

    This will affect the bias not only by including more possessions against the teams we scored more against, but also within a single game.

    It's not a causal thing -- shooting quicker doesn't necessarily mean it's easier to score.

    If you are comparing to other teams, I can see it. But I'm not sure it is going to say much about what pace you SHOULD play.

  3. Perhaps a bit simplistic, but how does the ePoss stat look for our opponents for the same period of time? My guess is that number went up significantly higher than the 0.8 number the hoyas increased during BE play year over year. Also, does an offensive rebound count as an additional possession? If not, that could mask the increase pretty well this year.

  4. The ePoss is interesting, but the floor percentage might be a little more informative on how efficiently a team is scoring. If the team is average, and most of the non-scoring possessions can be attributed to turnovers (as opposed to lack of rebounding for example), then you have a idea of what needs to be "fixed". Pomeroy pointed out several seasons ago that Georgetown had a very high turnover rate but scored very efficiently, minimizing the potential "problem". Are the turnovers "forced" (ie steals for example) or "unforced"? Another avenue to explore.

    PDRHoya99 -- offensive rebounds to not count as another possession; they are a continuation of the "current" possession.

  5. Sorry, I meant to check comments here but forgot-I'll add TO% when I get home. You're right, CO, the 2006 team had more ePoss because they turned the ball over less. More on why will probably be part of my TO post when I get to that.

    It wouldn't be too hard to run numbers for why a possession ended, at least in percentage terms, since that's just 1 - TO% (Stl%+"Unforced"%) - (DR/Poss) = made shot%. To get a good answer, I think you'd need more granularity in terms of made 3 v made 2 v made FT, but I'll see if I find something interesting. I can run this and ePoss for opponents during BE play as well. One interesting breakdown might be Opp TO% v. NCAA and non-NCAA teams.