I feel better knowing that Okafor obviously tanked Duke's 3 losses.

Here's what you make of it.
Okafor's effect on the team is neutral at best and devastatingly negative at worst.

We have to play him in order to keep the recruiting train going, but K should gradually start cutting back his minutes under the guise of resting him to keep the ankle healthy.

DukeCrow said:

Here's what you make of it.
Okafor's effect on the team is neutral at best and devastatingly negative at worst.

We have to play him in order to keep the recruiting train going, but K should gradually start cutting back his minutes under the guise of resting him to keep the ankle healthy.

Click to expand...

Do you think this is only somewhat unlikely, or the LEAST likely thing EVER?

Plus minus and minutes played information is much more readily accessible than per possession information with/without a condition, so it's understandable to use plus minus per 40 minutes, and there's some subjectiveness in counting possessions that isn't involved in simply counting minutes. For example, if Duke is running out the clock in the final possession of a blowout and it involves Cook just dribbling a ball in place without taking a shot, this does not seem to be appropriate to count as a possession to me, for these purposes at least. That doesn't even register on a play by play record of the game. As another example, if Duke scores or misses a shot with a few seconds left in the 1st half and the game clock runs out on the other team without the other team doing anything competitive with the ball (they don't even get a full court shot off), this does not seem appropriate to count as an opponent possession to me.

The biggest downside of using plus minus per 40 minutes is the fact that significantly more possessions with Okafor occur while Duke is on offense, since K has been subbing him out for some defensive possessions. This doesn't just happen at the end of games or halves; throughout the games, K seemed likelier to sub in Okafor to begin his stint on offense, and K seemed likelier to sub out Okafor to begin his bench stint on defense. Okafor's plus minus is inflated simply because he plays more on offense than defense. This was most drastic in the VT game, in which the non-Okafors played nearly twice as many possessions on defense than on offense. Obviously, a poor plus minus should not be damning for the non-Okafors in a game where they play nearly twice as much on defense than on offense, and per possession accounts for this while per minute does not.

That said, the fact that Okafor has been a slightly negative player in terms of plus minus in a certain subset of Duke games still kind of proves the whole point. Shouldn't Duke with Okafor be better than Duke without Okafor, by almost any metric and for almost any reasonably large subset of games? That DBR post is saying that even if you limit the data to Duke's wins (for whatever reason) and you disregard the plus minus inflation for Okafor, Duke is still slightly better without Okafor. Doesn't really say anything to me.

So Rasheed really was addition by subtraction. K is so clever.

Tyus/Grayson/Matt/Justise/Marshall. So obvious.

Justise's impact on the defense stands out like crazy and looks like what it should be based on eye test, and that's with a little bad luck on FT defense holding him down.

The Okafor problems on both ends look heavily 3pt related, and I'm not sure what that says beyond being bad luck. What happens if the 3pt% without Okafor is corrected from 51% to 34%? I'm getting 1.100 without Okafor for Duke's offense, basically representing the offensive efficiency for the Nokafor offense if we assume the entire huge gap in Duke's 3pt% is due to luck and no actual factors. With that assumption, there's basically no difference when we go from Okafor to Nokafor on offense, instead of a big dropoff.

On defense, doing the same thing, I get 1.066 for Nokafor's corrected defensive efficiency, though assuming the 3pt defense is completely luck is a stretch. There's basically no difference when we go from Okafor to Nokafor on defense then, and there's basically no difference overall for net efficiency.

If we assume the 3pt% variation is 100% luck, Okafor makes no net difference to Duke on either end or overall. Zack, can you check that math? I just took out turnover possessions and then multiplied stuff to get the point differentials.

I would not be quick to assume that offensive 3PT% variation with/without Jahlil is due to luck? The spacing and sets are different -- a lot more 1/5 ballscreen action. I think that's a legit relationship.

Yeah I think we got the same numbers. I mean, I'm not worried about it as far as ruining the observations or anything. The people who hate what the numbers say aren't going to be the same ones to say, "ah, we should be correcting as if 3s are entirely luck."

I don't think anyone will have any problem concluding that Winslow is our most valuable overall player based on these numbers and the eye test.

Need to manually enter Pitt at least, though, or else people will bitch.