I don't have much to say about yesterday's game. It was a big step back for everyone except Josh, who had an excellent offensive day (even if a lot of it came against Osby and Bowers, not Ibekwe and Gist). Duke can't linger on this one, as they'll need to be ready for a BC team looking to exact revenge and continue their hold on first place.
Here's the table from the Maryland game. All stats and leaderboards are updated.
| Player | O Poss. | Points | OPPP | D Poss. | Points | DPPP | |
| McRoberts | ON | 65 | 54 | 0.831 | 66 | 68 | 1.030 |
| OFF | 5 | 6 | 1.200 | 4 | 4 | 1.000 | |
| Scheyer | ON | 62 | 54 | 0.871 | 64 | 62 | 0.969 |
| OFF | 8 | 6 | 0.750 | 6 | 10 | 1.667 | |
| Nelson | ON | 34 | 30 | 0.882 | 34 | 39 | 1.147 |
| OFF | 36 | 30 | 0.833 | 36 | 33 | 0.917 | |
| Paulus | ON | 69 | 60 | 0.870 | 70 | 72 | 1.029 |
| OFF | 1 | 0 | 0.000 | 0 | 0 | 0.000 | |
| Henderson | ON | 26 | 21 | 0.808 | 22 | 27 | 1.227 |
| OFF | 44 | 39 | 0.886 | 48 | 45 | 0.938 | |
| McClure | ON | 40 | 33 | 0.825 | 42 | 36 | 0.857 |
| OFF | 30 | 27 | 0.900 | 28 | 36 | 1.286 | |
| Thomas | ON | 25 | 23 | 0.920 | 24 | 34 | 1.417 |
| OFF | 45 | 37 | 0.822 | 46 | 38 | 0.826 | |
| Zoubek | ON | 9 | 10 | 1.111 | 8 | 6 | 0.750 |
| OFF | 61 | 50 | 0.820 | 62 | 66 | 1.065 | |
| Pocius | ON | 19 | 15 | 0.789 | 20 | 16 | 0.800 |
| OFF | 51 | 45 | 0.882 | 50 | 56 | 1.120 | |
| Davidson | ON | 1 | 0 | 0.000 | 0 | 0 | 0.000 |
| OFF | 69 | 60 | 0.870 | 70 | 72 | 1.029 | |
| Duke Overall | 70 | 60 | 0.857 | 70 | 72 | 1.029 |
2 comments:
Hi,
I really enjoy your blog, and especially appreciate your points per possession data. I have a question about this data: at what point do you consider aggregate data significant? For example, I added up the totals for the players in Duke's four losses, and both Pocius and Zoubek have improved Duke's offense and defense when they're on the court. However, this is based on 26 and 38 possessions, respectively. Would you consider this significant? If not, how many possessions do you think would be enough to draw conclusions on a trend?
Thanks so much.
26 and 38 possessions is definitely not enough, particularly for low usage players (not just low usage in terms of minutes played, but in terms of possessions used when on the court). I actually think that the season stats on the whole are tricky for these two because they play in such short spurts, and performance in those spurts is random by nature. For example, if Scheyer can get an equally open three with Z in the post as with McRoberts, and he hits one of every two wide open looks, it's totally random chance which one he makes. If he hits it with Zoubek on the court, does that really mean that Z made the offense more efficient? Probably not. If he plays only in 3 possession spurts, there's so much randomness involved that it's tough to get statistical value even when you put 30-50 of those spurts together.
So there's no one round number at which I say "aha! it's now significant" because it depends overall on how the player is used. The most reliable numbers are for Thomas, Henderson, McClure, and Nelson. They're relatively high usage players who play extended stretches, but (unlike Scheyer, Paulus, and McRoberts) they also sit for extended stretches, so we have meaningful comparable numbers. But even for them, it takes about 5-6 games (150-300 possessions) before we can distill a real pattern.
Post a Comment