While it has running clock, basketball is also a turn-based contest. One team gets the ball, then the other team gets the ball. Repeat.

Teams don?t get the same number of plate appearances by any measure. Teams always get the same number of innings. Likewise, teams always have roughly the same number of possessions.

??Possession theory? shouldn?t be controversial – it?s just a necessary tool for calculating a rate, rather than an actual number ? calculating batting average or slugging percentage rather than the number of hits.?

It?s a necessary tool for converting a time-based contest into a turn-based contest. This is done because current statistical methods struggle with the concept of a time-based game as most have been developed for a turn-based game, which is far more conducive to statistical analysis.

?I think you are defensive because you think it?s inherently an either-or argument.?

I dislike the concept because I feel it makes too many compromises in the name of producing simplified answers.

?And I think this has a lot to do with the fact that he just isn?t a great athlete.?

Curry isn?t a great athlete? He has more fluid movements than some guards and his quickness and leaping ability are quite good. For a man his size, he?s an athletic freak.

?I should say, fwiw, that I am definitely influenced in my analysis by the fact that Wins Produced tells me that Dalembert has been above average every year of his career, something WP tells us Curry has never been, and a bottle of Cristal says he won?t be next year.?

This is because you?re working within the framework of an arbitrary statistical construct. You can afford to make that sort of bet because you know Curry won?t beat Dalembert when the statistical comparison is geared towards Dalembert?s skills and against those of Curry. You might as well be betting on blocks.

?I also struggle mightily to believe Dalembert is a weaker defensive center.?

This is probably because Dalembert blocks shots and makes impressive plays from the weakside. In man situations, he gets abused and commits too many fouls, as you saw when he was matched against Curry. Of course, since few centers are capable of much of anything on offense, is on-ball defense really important or relevant? Does it matter that Curry treats him like a ragdoll?

?In sum, statistically speaking, I just dont see the argument for Curry over Dalembert. You can?t make one on past performance. It can only be on potential.?

The argument is quite easy to make, actually. You could argue that Curry dominated when they were matched up against one another. Alternatively, you could claim that Curry?s massive advantage in offensive categories outweighs the advantages Dalembert enjoys in defensive categories and turnovers.

At this piont, an analysis gets very complicated. You are right, it’s true that Curry’s 9.6 rebounds per 48 was well below his career mark of 10.4, And his mark in 05-06 was above it by the same margin at 11.1. So that is a huge drop off there with increased minutes played. It’s also true his turnovers skyrocketed.

There seems to be some sort of basketball quanta for Curry. It’s a consistent pattern. When one thing goes up, another falls. Last year his scoring and minutes went up, and his scoring efficiency stayed the same, but his rebounding and turnovers got much worse. A .4 difference per 40 in offensive rebounding is a real cost, relative to what he was offering the year before. No matter how you spread Curry around the court, it seems you end up with a below average player. And I think this has a lot to do with the fact that he just isn’t a great athlete.

Dalembert’s rebounding numbers, on a bad rebounding team, were actually worse than his career average last year, and significantly worse than what he posted the two years before that. So, I don’t ihink his numbers were inflated by being no a bad rebounding team. He is, has always been, and always will be a much better rebounder than Curry.

It is true that his TS% was at a career high, but he was only 1.3% above his previous high and 2% above his career average.

I should say, fwiw, that I am definitely influenced in my analysis by the fact that Wins Produced tells me that Dalembert has been above average every year of his career, something WP tells us Curry has never been, and a bottle of Cristal says he won’t be next year.

I also struggle mightily to believe Dalembert is a weaker defensive center. That just seems ludicrous to me from what I have seen from him in college and the pros. Curry has made the Knicks 5.3 points worse on defense both years he has played here! And he made all his Bulls team worse as well. That is definitely statistically significant. Dalembert has made his team better on defense every year in the league, although as his minutes increase his defensive impact seems to be decreasing. Last year it was only 1.8.

In sum, statistically speaking, I just dont see the argument for Curry over Dalembert. You can’t make one on past performance. It can only be on potential.

ftk, this is getting weird. I mean, you are sharp with numbers, you think about what statistics represent but you are allergic to the word “posession.”

I (briefly!) thought more about the baseball comparison. Plate appearances are a good analogy to possessions; if an inning were the right analogy, a basketball team would get the ball back after a score, like we do it in playground 3-on-3.

“Possession theory” shouldn’t be controversial – it’s just a necessary tool for calculating a rate, rather than an actual number — calculating batting average or slugging percentage rather than the number of hits.

I think you are defensive because you think it’s inherently an either-or argument. Not necessarily. Hits are a decent statistic; they tell you, among other things, whether a player was durable enough to play a lot of games, and whether his fielding skills (or home run power, or speed) were enough to keep him in the lineup. Measures of total production, including per-game stats, are perfectly valid and useful, e.g. telling you at a glance is this player a starter, a primary scorer, etc.

BUT… there are many reasons many (most) people on this board prefer rate (i.e. possession-based) statistics. Some of them:

1. Per-minute stats stay fairly constant as minutes increase, or decrease. There is an element of truth that yes, players only get more minutes if they play well and earn more minutes. a) That’s a limitation of any stat — more so if you’re talking PPG; a player can’t score more unless he’s playing well enough to be in the lineup. b) If there were any significant tendency for “role” players to do worse when they get more minutes, it would be apparent, since coaches do in fact test guys who don’t pan out, players are pressed into larger roles by injury, etc.

2. Per #1, possession-based stats enable you to make comparisons between full and part-time players.

3. Possession based stats are more precise, and provide much more detail. Per-game stats tell you, above all, who gets playing time. If you only look at total points scored, the conversation is over. If you want to understand what factors went into scoring those points – which skills are most valuable, what types of offenses (or defenses) are most effective, etc., you have to analyze possessions.

Clearly we like to look at different things. I place significantly more emphasis on offensive statistics than you do and you value possession statistics more.

Just tell me how many more points Curry scores for the Knicks then Dalembert and the Sixers do. Then back out 1.2 points for Dalembert’s offensive rebounding, and the difference in turnovers committed, and tell me why it doesn’t make sense that Curry doesn’t actually have a huge impact offensively compared to a player like Dalembert.

And this is, again, totally leaving out defense, wehre Dalembert 2.9 more rebounds, .3 more steals, and 2.3 more blocks.

Dalembert is the better player by a large margin.

About the offensive rebound, the difference per 40 (which is what I based all my calculations on) is actually 0.8, which is worth about 0.7-0.8 points, so you are right that makes the offensive cotribution about equal, or at least within 0.2-0.3 points per 40, but that is assuming that the extra 0.8 offensive rebounds that Dalembert gets come from taking the rebound from the other team and not from a teammate. So I would value that 0.8 extra offensive rebounds per 40 at no more than an extra 0.5 points but still, I admit that is less than a one point difference.

But I think that Dalembert’s defensive contributions are exaggerated, he is a great weakside shot blocker but is a fairly poor man defender as reflected by the 17.0 PER he allowed opposing center’s to get this year. So while he is a much beter shot blocker than Curry, Curry is a better man defender and only allowed his man a 15.8 PER against him. Dalembert is at best an average defender, and at worst actually worse than Eddy.

Also while Dalembert averaged more rebounds than Curry he did so on one of the worst rebounding teams in the league and would most likely see his rebounding contribution diminish if he played on a great rebounding team like the Knicks.

With all that said I still like Dalembert and while I would not take him over Curry I still think he is an above average center. The big reason why I prefer Curry is that hehad a career high turnover rate and a career low free throw percentage and Dalembert had a career high in TS% and FT%, and the second lowest turnover rate of his career, yet Curry was still the better offensive player. If Curry can return to his previous form in regards to turnovers and free throw percentage the offensive differerence would not be close at all.

As I’ve already explained, if you’re interested in measuring each player’s impact on TS%, you can see what their contributions did for their team’s respective offenses.

If you wanted to see Dalembert matched up against Curry, we’ve already gone through that scenario as well. I think the numbers speak for themselves.

I understand that you’re interested in finding some way of proving Dalembert more valuable than Curry, via statistics. But manipulating data is an arbitrary fashion is not exactly a sound method for analysis.

I am not sure your numbers are right. I don’t understand them anyway. How does Curry get 1.4 less assists for instance? They have the same assist number. I do see what you are doing combining turnovers into the possession equation. Will run the numbers by hand per 48 and see what I think.

Also, does that 1 point advantage Curry has over Dalembert take into account the extra offensive rebound Dalembert would capture per 40 for the Knicks? That would result in an extra point and make their impact equal, right? So then you would have basically the same offensive impact, would leave you with a far superior player on the defensive, someone who rebounds better and blocks a lot more shots.

Isn’t that right?

Curry Total 23.5 Pos:
22.2 Pts 4.1 To 1.0 Ast
Dalembert 15.5 Pos + NY-Curry 8 Pos:
19.8 Pts 3.5 To 2.4 Ast

So in 23.5 Possesions Curry scores 2.4 more points, and gets 1.4 less assists, while causing 0.6 more turnovers. (since an assist is worth about one point over a simple pass and the negative value of turnovers is already built into his used possesions) Curry has a +1 point advantage over Dalembert(with other NY players picking up the slack).

Curry vs the average NY player
Curry 23.5 Pos:
22.2 Pts 4.1 To 1.0 Ast
NY-Curry 23.5 Pos:
16.5 Pts 3.0 To 4.0 Ast
Again Curry is better than the average Knick player. He scores 5.7 more points and has 3.0 less assists while having 1.1 more turnovers. That gives Curry a +2.7 advantage over the average Knick.

As you can see Curry makes up for his turnovers by converting a much higher percentage of his shots. Curry is a good offensive player and if he can get his turnover rate and free throw percentage back to where it was in Chicago he will be a great one and his dominance over the average Knick or a center like Dalembert will be even more pronounced.