Wednesday, November 12, 2008

Duke 97, Georgia Southern 54

The Devils did a nice job yesterday illustrating that offense is about more than just making shots. Duke did not make shots exceptionally well against Georgia Southern - not from the field, and especially not from the line. Certainly it wasn't the kind of shooting performance one normally associates with a 97 point outburst in 77 possessions. The reason the offense was so efficient is because Duke completely controlled the offensive glass - 23 of the 43 available rebounds (that is, Duke had more offensive rebounds than the Eagles had defensive rebounds), including several boards off of missed free throws - and because, until the last 5-10 minutes of the game, Duke simply didn't turn the ball over - only 11 on the game as a whole, and I think with 10 to play in the game that number was 6. This gave the Devils many more opportunities than their opponents (only 2 more FGAs, plus a huge edge in FTAs), and allowed Duke to turn a pedestrian shooting night into an excellent offensive showing.

On the individual level, Singler once again had a fantastic game. He led the team in scoring, added in 8 boards, a couple of assists, etc. He looks ready to carry a healthy share of the offensive responsibility for the team this season. From the look of the box score, pretty much everyone played well. Williams turned the ball over a lot, but also managed to grab 11 boards in about 13 minutes. Scheyer's shot wasn't falling, but he dished out 7 assists against just 1 turnover. McClure, Plumlee, Zoubek, and Thomas were all very active on the offensive glass. All in all, it was a good night for Duke. At the very least, it calmed down the people who are inclined to start wringing their hands after beating Presbyterian by "only" 31.

Below are the two HD boxes from the 2ksports pod games. I just noticed that I didn't sort the GS one by +/- before uploading it, so it's slightly disorganized. My apologies. I'm also going to create a post to collect all of these boxes and sticky it over to the links on the right side of the page. As for the other stats, I'm in the process of updating my sheets to get everything ready for the season, and will hopefully update those by early next week. They're likely to be posted via embedded spreadsheets as well.

Presbyterian:


Georgia Southern:

4 comments:

Douglas said...

paul,

being new to, but excited about, these new advanced offensive stats you've brought us this year, i wonder if you can help provide us a little context.

for instance, what would one expect for an average starter in the NCAA? in the ACC? what would they be for an outstanding season, a la 2005 jj redick?

i'm sure some of this will fill in over the season, as we've had a few months to check the ACC leaderboards and compare the stats of players in the league as we become more familiar. but for now, how "good" is that 164 next to singler's name, for instance?

Paul Rugani said...

The 164 is very, very good. Unsustainably good, in fact. The "average" offensive rating for players is about the same as the average offensive rating for teams, so in the neighborhood of 100-102. The reason it's the same is that the "points produced" figure (used to calculate player offensive rating) is designed to come out essentially equal to the actual points the team scored. For example, the total points produced for the Presbyterian game came out to 79, and we scored 80. For the Georgia Southern game, it came out to 102, and we scored 97. So it will be roughly the same.

The range for player offensive rating is a little bigger than team offensive rating, particularly because lower usage players can more easily amass higher ratings. High usage players tend to have a lower ceiling. For example, Redick's senior year, his offensive rating was 120.2. He also used 29.2% of possessions (so very high usage). For players who used at least 28% of their team's possessions, he was second in the country (to Nick Fazekas). When you lower the threshold to all players who played at least 15 min/g, regardless of usage, Redick drops all the way down to 48th.

Pomeroy keeps all this information on his site, although it can be hard to find. Here's a link to the 2005-06 numbers: http://kenpom.com/leaders.php?c=ORtg&y=2006
And here's one for last season:http://kenpom.com/leaders.php?c=ORtg&y=2008

Last year Scheyer was 8th in the country with a 127.6, although his usage was rather low, at just 17.6%.

As for the difference between the ACC stats and the NCAA stats, on the season overall, ACC players will likely be higher - maybe an average of 104-106, which is roughly what the team's average overall. For conference-only play, the average will come down to around 100-101.

Douglas said...

awesome, thanks!

so i think i've got it:

(1) offensive rating is designed to center at 100

(2) the sum of every player's floor possessions across the team should approximate the total points the team scored times 5 (so for individual players is designed so that, on average, the players' individual floor possessions approximate the total points the team scored multiplied by the percentage of possessions that each player was on the floor)

(3) usage is designed to center at 20.

i'm assuming it must exist somewhere, but do you know of a plot online showing the distribution of offensive rating versus usage?

Paul Rugani said...

Sort of.

1) The rating isn't necessarily designed to center at 100. It's designed to mirror (in a way) the team's performance. Now, as a factual matter, the average team offensive rating is around 100, so that tends to be the center point for players as well. But if average team offensive performance just happened to be 110 one year, then offensive ratings for players would similarly center around 110.

2) The "floor" figure in the HD box is a different rating. It's individual scoring possessions divided by individual total possessions - so a measure of how successfully a player converts usage into points. So Singler's 55.81% floor rating for the Rhode Island game means that 55.81% of the possessions he used resulted in a score for Duke (and, correspondingly, that 44.19% did not). The individual total possessions and scoring possessions are not on the box score, nor are the individual points produced (used for offensive rating), although I guess I could include them all.

3) Yes. 20% across the board would be an even distribution, and the weighted average of usage should be 20%. Also, in theory, the sum of everyone's usage * possessions played should equal team possessions (with a little rounding). For example, if you do that math on the Georgia Southern game, it comes out to 76.94, and we had 77 offensive possessions. The URI game is off for some reason - I'll have to check why when I get home.