Looking forward to how Joerger will play WCS

[Kingster]

So Willie can stay in front of a PG, or a SG, or SF, but he can't stay in front of a PF....Right!

No, the accurate statement is: In general, Willie could not stay in front of PGs, SGs SFs, or stretch PFs in his rookie year of the NBA; he was much more comfortable guarding centers or traditional power forward types. It is also accurate to say that for some strange reason the myth was perpetuated that he could guard those aforementioned positions, which appears to be the hype of the draft (and probably his agent), and the wishful thinking of some fans, not a realistic assessment of his performance on the court in his rookie season.

 

[Kingster]

IMO, this was a mis-read of what happened, caused by the commentators and the fans after the game.

Ryan Anderson destroyed WCS in games................. but he also destroyed every Kings player that tried to guard him, and it wasn;t just because of Karl's incompetent defense.
There were plays where Willie was in Ryan's face, and he still swished the shot.
Ryan Anderson played better against us last year than I've seen any stretch 4 play in recent memory. Guy was automatic, and unstoppable.

At a certain point, you have to just hand it to elite athletes and hope they don't keep doing that against ya.
I'm so sick of the current NBA atmosphere that concludes that your team (or guy) is doing something wrong when they don;t stop the best players in the world.
Newsflash - in the current NBA offenses, it is impossible to prevent 3-pt shooters from getting the shot off.
You have to give up SOME shots.
And Karl chose to give up the 3 ptr.
All season.
Even when the Kings lost single-handedly to it, time and time again.
Even when the guy driving the lane wasn't going to shoot it, or sucked at shooting it.
Even when our big men were dominant over their big men - they STILL would "challenge the drive" and give up the open 3, all game long, all season long.

IfAtfirst, it's irrelevant that Anderson killed every other Kings player. It's not news that other Kings players can't guard. In fact, it's expected. What was unexpected by many is that WCS couldn't guard stretch 4s (and let's not even broach the subject of his ability to guard PGs, SGs, or SFs). Also, I'm not basing my argument solely on the fact WCS couldn't guard Anderson. Anderson and Millsap are just two very prominent stretch 4s among many during the course of the season that WCS couldn't guard. And regarding Karl strategems, I'm not basing my opinion on Willie's pick and roll coverage or his switches or other actions on the floor that involve coordination with Kings players; I'm basing it on his inability to guard a guy in space, especially when such guy was isolated on him. Overall, the hype of WCS didn't match his performance on the court.

 

[Gilles]

WCS' defensive footwork and concentration are nowhere near his mobility and length. He could get away with it in college most of the time, because of physical domination and more compact game. Not gonna cut it in the NBA.
Still your teammates not executing defensive scheme properly will make you look very bad as well. That's the nature of the NBA - too much skill.

 

[biltalent]

2% worse is significant... and there's nothing impressive about it- let's take a look at a not-so-short-list of other PF/C rookies from this class (and WCS was advertised here as the best suitable big in defending smaller guys) that are doing better:

Bobby Portis: opponents shoot 10.6% worse than average on shoots from 15 ft out

Myles Turner: opponents shoot 4.3% worse than average on shoots from 15 ft out

Kristaps Porzingis: opponents shoot 3.1% worse than average on shoots from 15 ft out

Chris McCullough: opponents shoot 4.1% worse than average on shoots from 15 ft out

Larry Nance JR: opponents shoot 3% worse than average on shoots from 15 ft out

Trey Lyles: opponents shoot 2.8% worse than average on shoots from 15 ft out

Nemanja Bjelica: opponents shoot 1.6% worse than average on shoots from 15 ft out

KAT: opponents shoot 0.6% better than average on shoots from 15 ft out

Nikola Jokic: opponents shoot 1% better than average on shoots from 15 ft out


So I wouldn't call WCS impressive in that respect, in fact he is practically the same at protecting outside shoots as Okafor and Kaminsky:

WCS: opponents shoot 2% better than average on shoots from 15 ft out

Jahlil Okafor: opponents shoot 2.1% better than average on shoots from 15 ft out

Frank Kaminsky: opponents shoot 2.5% better than average on shoots from 15 ft out

That's pretty terrible, again I'm not saying he can't fix that- but right now claiming that he is good has no basis in truth.

I don't get why it's so weird to you- you really thought he has done an amazing job defending outside shoots? I remember multiple games where guys complained about that in the game threads.

Notice that I'm only talking about outside shoots, in general WCS was pretty average.
And he has been very good at defending the rim stopping guys there at 47.9%, which is among the top for rookies (Porzingis is slightly better with 47.7%- but it is a better result than KAT and Turner for example) and actaully better than both KK and DMC (slightly worse than last years DMC result- 47%) who are pretty good themselves.

I'm also not saying that he will never be a good defender in that respect- he certainly have a lot of tools that can help him with it, but this season he wasn't great at it.

I agree with you that the eye test says that Willie certainly room to improve guarding the perimeter. My point was more about your use of statistics that I don't agree with. Just because 37% is 2% lower than 39% it doesnt mean that difference is "real" / statistical significant on a significance level (f.e. 95%) or in anyway meaningful. You would also have to factor in the standard error and run a t-test for such a conclusion.

 

[Amit149]

I agree with you that the eye test says that Willie certainly room to improve guarding the perimeter. My point was more about your use of statistics that I don't agree with. Just because 37% is 2% lower than 39% it doesnt mean that difference is "real" / statistical significant on a significance level (f.e. 95%) or in anyway meaningful. You would also have to factor in the standard error and run a t-test for such a conclusion.

So the eye test is bad and the numbers are bad but you disagree with me because???

Again no team in the NBA has been worse than 1.7% in that respect, Stein is 2%- pretty much every rookie PF/C has done better outside of Kaminsky (2.5%) and Okafor (2.1%)... I don't see your point, I'm not sure there is one.

It's 2% from the average- not from being a good defender, you claimed that it was impressive than I guess Okafor is impressive and the Kings defense is stellar.

Marco Belinelli is a terrible defender, amongst league worst- opponents shot 3.6 better on him- does that count as significant? it's less than 2% away from WCS so I guess the difference between them has no statistical significant... that's ridiculous.

And again standard error how? this are stats from his entire season and it's 51.6% of the shoots he defended so it's not a small sample size and in any case- it's all the evidence we got and we're talking about THIS season.

The only thing I've argued is that he didn't guard outside shoots well- this stats are specifically about how well opponents shoot when he defended them- it's worse than other rookies, it's worse than team average- what more do you want? or is it just not valid to say non-positive things... let me guess that if the stats showed he was locking down guys on the perimeter you would have used that to show Kingster he's wrong.

 

[macadocious]

I'm just going to go ahead and say throw out every single defensive metric on Kings players last year. They played a switching sagging zone that constantly put them in awful matchups. The only way the defense could have been worse was to specifically line up big men on PGs.

Shelve this entire conversation until AT LEAST the halfway mark of this next season.

 

[Capt. Factorial]

So the eye test is bad and the numbers are bad but you disagree with me because???

Again no team in the NBA has been worse than 1.7% in that respect, Stein is 2%- pretty much every rookie PF/C has done better outside of Kaminsky (2.5%) and Okafor (2.1%)... I don't see your point, I'm not sure there is one.

It's 2% from the average- not from being a good defender, you claimed that it was impressive than I guess Okafor is impressive and the Kings defense is stellar.

Marco Belinelli is a terrible defender, amongst league worst- opponents shot 3.6 better on him- does that count as significant? it's less than 2% away from WCS so I guess the difference between them has no statistical significant... that's ridiculous.

And again standard error how? this are stats from his entire season and it's 51.6% of the shoots he defended so it's not a small sample size and in any case- it's all the evidence we got and we're talking about THIS season.

The data shows that Willie was slightly worse than average at defending that shot this year. The question isn't whether he was slightly worse, the question is why? Was he worse because he's not a good defender on that shot, or was he worse because he just got unlucky? We want to answer that question because we want to know how WCS will do on defense in the future, so we want to get a grasp of what his underlying ability on D might be. And trying to answer that question comes down to sample sizes. You wouldn't want to call WCS the greatest defender ever if he defended one jump shot and it missed (0% against! Amazing!) any more than you'd want to call him the worst defender if he defended one jump shot and it hit (100% against! He's awful!).

Now, I don't know the actual sample sizes (perhaps someone does have those handy) but there were about 200,000 shots attempted in the NBA last year. Let's guess and say that 100,000 of those were 15+ foot jumpers, on which the average shooting percentage was 37%, so 37,000 went in. That's a good sample size. How did WCS do? Well, he played 1412 minutes, only about half of that should be on D so about 700 minutes of D, which could translate to about 2500 possessions or so, depending on how quickly teams shot. We already assume that about half the time teams shoot a 15+ foot jumper, so bring that down to 1200 of those jumpers (there will be some turnovers in those possessions) and assume WCS is responsible for defending at most 1/3 of those. So perhaps he defended 400 15+ foot jumpers this year, at 39% that would be 156 makes. That seems like too many attempts to me, but I'll roll with it.

If you run a z-test for difference of proportions on 156/400 vs. 37000/100000, the one-sided p-value is about 0.2, which means that there's a 20% chance that WCS's results would come out by random luck to be 39% or worse if his true "ability" is in fact 37% - dead on average. That's pretty marginal to make a solid statement. You could call it a trend, but if somebody wants to remain unconvinced on the strength of such a sample size, they're pretty well within their rights. I can tell you that a scientist would not make a conclusion on the basis of that data.

 

[Amit149]

The data shows that Willie was slightly worse than average at defending that shot this year. The question isn't whether he was slightly worse, the question is why? Was he worse because he's not a good defender on that shot, or was he worse because he just got unlucky? We want to answer that question because we want to know how WCS will do on defense in the future, so we want to get a grasp of what his underlying ability on D might be. And trying to answer that question comes down to sample sizes. You wouldn't want to call WCS the greatest defender ever if he defended one jump shot and it missed (0% against! Amazing!) any more than you'd want to call him the worst defender if he defended one jump shot and it hit (100% against! He's awful!).

Now, I don't know the actual sample sizes (perhaps someone does have those handy) but there were about 200,000 shots attempted in the NBA last year. Let's guess and say that 100,000 of those were 15+ foot jumpers, on which the average shooting percentage was 37%, so 37,000 went in. That's a good sample size. How did WCS do? Well, he played 1412 minutes, only about half of that should be on D so about 700 minutes of D, which could translate to about 2500 possessions or so, depending on how quickly teams shot. We already assume that about half the time teams shoot a 15+ foot jumper, so bring that down to 1200 of those jumpers (there will be some turnovers in those possessions) and assume WCS is responsible for defending at most 1/3 of those. So perhaps he defended 400 15+ foot jumpers this year, at 39% that would be 156 makes. That seems like too many attempts to me, but I'll roll with it.

If you run a z-test for difference of proportions on 156/400 vs. 37000/100000, the one-sided p-value is about 0.2, which means that there's a 20% chance that WCS's results would come out by random luck to be 39% or worse if his true "ability" is in fact 37% - dead on average. That's pretty marginal to make a solid statement. You could call it a trend, but if somebody wants to remain unconvinced on the strength of such a sample size, they're pretty well within their rights. I can tell you that a scientist would not make a conclusion on the basis of that data.

I get that, but again the claim was that he did a good job at that- I think the eye test says otherwise, I don't treat that stat as ultimate proof- but at the very least it supports that claim (and 80% that it does reflect the truth is pretty good if I understand correctly) furthermore even if I would concede that he might be average that's far from an accomplishment if you look at his fellow classmates.

The number of shots Willie defended is about 309 (4.9 attempts in 63 games) which isn't a tiny sample- and if you do think it is such than you can go ahead and discount everything he did this season.

Now this approach is reasonable but I wonder where all the doubts were when halfway through last season DMC was regarded a top NBA rim protector based on perhaps a smaller sample size (and for the record I think they were right to do so)...

You said that a scientist would not make a conclusion on the basis of that data- but would he ignore it completely? like you said it points to a trend and that's all the data we have so far- and if you ask to make a statement based on it that's where the data is (notice that when at first it was looked at as impressive people welcomed it with open arms).

My point was that there's nothing to base the claim that WCS is good at that on, I'm not saying he won't get better- he certainly have the tools, but I get the feeling people use stats here just when it confirms the opinion they already hold and label it as false whenever it clashes with the belief they already hold.

 

[biltalent]

So the eye test is bad and the numbers are bad but you disagree with me because???

Again no team in the NBA has been worse than 1.7% in that respect, Stein is 2%- pretty much every rookie PF/C has done better outside of Kaminsky (2.5%) and Okafor (2.1%)... I don't see your point, I'm not sure there is one.

It's 2% from the average- not from being a good defender, you claimed that it was impressive than I guess Okafor is impressive and the Kings defense is stellar.

Marco Belinelli is a terrible defender, amongst league worst- opponents shot 3.6 better on him- does that count as significant? it's less than 2% away from WCS so I guess the difference between them has no statistical significant... that's ridiculous.

And again standard error how? this are stats from his entire season and it's 51.6% of the shoots he defended so it's not a small sample size and in any case- it's all the evidence we got and we're talking about THIS season.

The only thing I've argued is that he didn't guard outside shoots well- this stats are specifically about how well opponents shoot when he defended them- it's worse than other rookies, it's worse than team average- what more do you want? or is it just not valid to say non-positive things... let me guess that if the stats showed he was locking down guys on the perimeter you would have used that to show Kingster he's wrong.

Ah no, I didnt want to imply that the eye test is bad. My point is that I agree with you that he needs to improve his ability to guard perimeter players. Thats what I saw watching games as well. Its just that the numbers you pointed out don't support that claim as the Capt pointed out. If we run the numbers he is not below average because 37% is not significant different from 39%.
Now, the question is, how do we interprete this finding? This is where its get tricky because its highly subjective. My point was that I didnt expect him to be above average because he is a rookie. Your point is that a number of other rookies were better. Both are valid points I think.

 

[IfAt1st]

I'm just going to go ahead and say throw out every single defensive metric on Kings players last year. They played a switching sagging zone that constantly put them in awful matchups. The only way the defense could have been worse was to specifically line up big men on PGs.

Shelve this entire conversation until AT LEAST the halfway mark of this next season.

macadocious is correct again.

I just can't understand why some people are actually looking at data from last year to determine if a player is good at defending outside shooting.

Did you guys not just see the historically-bad defensive scheme they were forced to perform, thanks to Karl?!
How can one player make up for a scheme that forces them to leave outside shooters open?

I guess some people will argue just about anything, but making conclusions about a player's ability to defend outside shots is one of the last statistical analysis I would have imagined being run from last season's numerous debacles.

 

[Capt. Factorial]

The number of shots Willie defended is about 309 (4.9 attempts in 63 games) which isn't a tiny sample- and if you do think it is such than you can go ahead and discount everything he did this season.

It's not a tiny sample, but the difference from average is a small difference, which makes it hard to say that WCS is definitely worse than average at defending the 15+ foot jumper. For example, given the 309 sample size, he'd have to have allowed about 42.5% shooting (relative to the 37% average) to reach the traditional p < 0.05 threshold for confidence.

You said that a scientist would not make a conclusion on the basis of that data- but would he ignore it completely?

Well, they wouldn't be able to publish it as a significant difference. Typically nobody would believe it as a real result without further verification.

My point was that there's nothing to base the claim that WCS is good at that on, I'm not saying he won't get better- he certainly have the tools, but I get the feeling people use stats here just when it confirms the opinion they already hold and label it as false whenever it clashes with the belief they already hold.

Well, there's certainly nothing in WCS allowing 39% on 15+ foot jumpers that suggests that he's better than average. It's not quite enough to come to a solid conclusion that he's worse than average, though. Maybe he was, maybe he wasn't. And, as you say, he might well get better - it's not like he has reached the end of his career and we can assume that he has plateaued. I can't really begrudge the wait-and-see crowd - if nothing else, WCS was drafted as high as #6 specifically because his toolset suggests he will be a good NBA defender pretty much across the board once he reaches his prime.

But you're right in the contention that ignoring statistics that tell "the wrong story" isn't terribly intellectually honest. That doesn't you can't make a reasonable argument that the statistic may not be meaningful, though.

 

[Amit149]

It's not a tiny sample, but the difference from average is a small difference, which makes it hard to say that WCS is definitely worse than average at defending the 15+ foot jumper. For example, given the 309 sample size, he'd have to have allowed about 42.5% shooting (relative to the 37% average) to reach the traditional p < 0.05 threshold for confidence.



Well, they wouldn't be able to publish it as a significant difference. Typically nobody would believe it as a real result without further verification.



Well, there's certainly nothing in WCS allowing 39% on 15+ foot jumpers that suggests that he's better than average. It's not quite enough to come to a solid conclusion that he's worse than average, though. Maybe he was, maybe he wasn't. And, as you say, he might well get better - it's not like he has reached the end of his career and we can assume that he has plateaued. I can't really begrudge the wait-and-see crowd - if nothing else, WCS was drafted as high as #6 specifically because his toolset suggests he will be a good NBA defender pretty much across the board once he reaches his prime.

But you're right in the contention that ignoring statistics that tell "the wrong story" isn't terribly intellectually honest. That doesn't you can't make a reasonable argument that the statistic may not be meaningful, though.

A few thoughts that came into my head and I'm asking out of curiosity since you obviously know more about it than I do:

1. What if I'll change the argument- can I make a claim that he is worse than the average rookie PF/C for example? or even the general C? in that case his sample will be bigger as it relates to the overall sample.
And a follow up question (which you probably took into account)- the difference isn't really 2% it's just over 5%- because that's the difference between 37 and 39 in precentages.

2. Can't I make the exact same claim on the other side? since the average doesn't really mean anything (unlike say using dice) and it is only use as a reference point- aren't the odds the same for him being average and for him being 4% worse?
I know it won't effect validity but if there is nothing I missed it could serve as a argument for it since it seems to me there's an idea if it's wrong it will be wrong in favor of him.

3. If we'll hold all stats to scientific demands won't it destroy our ability to use them? since we are talking here about men and not controlled environment and there is a ton of variables (injuries, travel, mental state, growth, training, aging, off court behaviour, luck...) that change from game to game and especially between season and season- could I really make a scientific claim that Cousins will score more points than the average player next year for example?

I would think that you could throw out all stats and all other measuring tools for that matter if that's the case- which makes sense because I doubt you can make a scientifically accurate prediction of the future about a game played by human beings with so many variables... so the claim is not only against the stats I brought but also against any source of information.

BTW just to make that point- here is the full list of players with more than 300 attempts and precentage worse than 42.5%:

Jeff Green.

 

[Bricklayer]

I am not even remotely worried about WCS's perimeter defense. Rookies always struggle on defense., Even ones who have good defensive coaches. He'll need to figure out how to play the stretch 4's, but I'd bet dollars to doughnuts the new coaching staff does a better job helping him sort that out than the old one did. Physically he's nearly flawless for that work. Just needs experience and a theory/plan of attack.

 

[King Baller]

I'm just going to go ahead and say throw out every single defensive metric on Kings players last year. They played a switching sagging zone that constantly put them in awful matchups. The only way the defense could have been worse was to specifically line up big men on PGs.

Shelve this entire conversation until AT LEAST the halfway mark of this next season.

I think macadocious has it right here. I'm not getting too worked up about the results from the Karl coached Kings.

 

[Capt. Factorial]

A few thoughts that came into my head and I'm asking out of curiosity since you obviously know more about it than I do:

1. What if I'll change the argument- can I make a claim that he is worse than the average rookie PF/C for example? or even the general C? in that case his sample will be bigger as it relates to the overall sample.

From a statistical point of view, "evening up" the sample sizes doesn't actually help - in fact making any sample size smaller means that the differences between the two results has to be bigger in order to be statistically significant. I think what would be of more impact by looking only at rookie PF/Cs for instance, is that they may actually be worse than the league average of 37% when defending 15+ foot jumpers, both because they are rookies and because they are not typically used to defend on the perimeter. Compared to a smaller sample of big men, particularly rookie big men, WCS could well be better at his value of 39%. But by reducing the sample size, the difference would have to be even larger to be statistically significant.

And a follow up question (which you probably took into account)- the difference isn't really 2% it's just over 5%- because that's the difference between 37 and 39 in precentages.

I recognize that, but the value of 2% doesn't actually play in to the calculations for a z-test. The actual ratios and sample sizes are all that matter (e.g. 120/309 vs. 37000/100000).

2. Can't I make the exact same claim on the other side? since the average doesn't really mean anything (unlike say using dice) and it is only use as a reference point- aren't the odds the same for him being average and for him being 4% worse?
I know it won't effect validity but if there is nothing I missed it could serve as a argument for it since it seems to me there's an idea if it's wrong it will be wrong in favor of him.

I'm not sure I follow what you're asking. I think you're asking the question of why we should assume that WCS's true percentage allowed is the league average 37%, rather than say 39% or even 41%. If that's what you're getting at, I see what you're saying, but it's kind of coming at it from the wrong direction. The reason is that we have a specific question here, and that question is: Is WCS a worse than average defender against 15+ jumpers? We're comparing him to the average because that's what we're interested in - we don't want to assume that he has any particular value of percentage against. We could ask other questions, of course, such as is WCS's percentage allowed statistically better than 45%? (Answer: yes) or is WCS's percentage allowed statistically worse than 35%? (Answer: no. Almost, but not quite at p < 0.05).

As for the partial question of whether the odds are the same that he is average or 4% worse than average...generally, no. The underlying distribution of something like that is usually Gaussian (a "bell curve") where average is most likely and the further you are from average the less likely you'll find someone with that value. For instance if that question were reworded ("aren't the odds the same that WCS allows 37% and that WCS allows 100%?") I suspect you'd intuitively say that the odds that WCS is such a poor defender that he allows 100% is very unlikely. BUT, that's not actually very relevant, either way. We don't actually use knowledge of the underlying distribution of percentage allowed when we do the statistical test.

3. If we'll hold all stats to scientific demands won't it destroy our ability to use them? since we are talking here about men and not controlled environment and there is a ton of variables (injuries, travel, mental state, growth, training, aging, off court behaviour, luck...) that change from game to game and especially between season and season- could I really make a scientific claim that Cousins will score more points than the average player next year for example?

I think one thing that one finds when they play around with stats in terms of asking questions about statistical significance is that we can't be really rock-solid certain about most sports stats as we'd like to think. I mean, we don't really know who is going to lead the league in scoring next year (Curry? LeBron? Harden? Somebody else?) and sportsbooks take advantage of that uncertainty. But some things are easier than others. I'd suspect (without trying to run any numbers) that you could make a very good case for Cousins to have scored statistically more than average this year, especially if you're looking at per-game numbers. But there's a big difference there...figure a team scores about 100 points and plays about 10 guys, so the average player scores about 10 points more or less. Cousins was around, what, 25 PPG or so? That would be 250% of average - it's not hard to imagine that the data will show he's better than average. On the other hand, the defensive percentage against stats are much tighter. WCS was in ratio terms about 5% worse than average. And you need to have a REALLY big sample to be confident that 5% worse than average really is worse than average rather than maybe just bad luck.

But I don't think that we necessarily HAVE to hold stats to statistical significance. It's perfectly valid to say, hey, WCS was worse than average on defending 15+ foot jumpers this year. It's a small effect, but that's what happened. At the same time, if somebody says they don't think that's necessarily real and points to it not being statistically significant, well, they're right too. There's a good chance that WCS's slightly worse than average 15+ foot J defensive numbers this year were just the result of bad luck. "He was worse than average this year" is right. "There's not very convincing evidence that he's going to be worse than average next year" is also right.

I could say a lot more, but I've probably said way more than enough!

 

[Amit149]

I think what would be of more impact by looking only at rookie PF/Cs for instance, is that they may actually be worse than the league average of 37% when defending 15+ foot jumpers, both because they are rookies and because they are not typically used to defend on the perimeter. Compared to a smaller sample of big men, particularly rookie big men, WCS could well be better at his value of 39%. But by reducing the sample size, the difference would have to be even larger to be statistically significant.

Thanks for the answers

WCS won't be better than other rookie PF/C since the only guys doing worse than him that played meaningful minutes in that respect are Okafor and Kaminsky...

But I don't think that we necessarily HAVE to hold stats to statistical significance. It's perfectly valid to say, hey, WCS was worse than average on defending 15+ foot jumpers this year. It's a small effect, but that's what happened. At the same time, if somebody says they don't think that's necessarily real and points to it not being statistically significant, well, they're right too. There's a good chance that WCS's slightly worse than average 15+ foot J defensive numbers this year were just the result of bad luck. "He was worse than average this year" is right. "There's not very convincing evidence that he's going to be worse than average next year" is also right.

I agree with that and agreed with that to start with so most of that was pointless (atleast for you, I learned some new stuff).

I never said I think we can project his next year results from this year, less because of the option of "bad luck" and more because he is a rookie and I think he has a very good chance of being better at it since he doesn't lack the tools needed (unlike a guy like Kaminsky for example)- I just objected to the claim he was really impressive this year in that respect.

 

[VF21]

From a statistical point of view, "evening up" the sample sizes doesn't actually help - in fact making any sample size smaller means that the differences between the two results has to be bigger in order to be statistically significant. I think what would be of more impact by looking only at rookie PF/Cs for instance, is that they may actually be worse than the league average of 37% when defending 15+ foot jumpers, both because they are rookies and because they are not typically used to defend on the perimeter. Compared to a smaller sample of big men, particularly rookie big men, WCS could well be better at his value of 39%. But by reducing the sample size, the difference would have to be even larger to be statistically significant.



I recognize that, but the value of 2% doesn't actually play in to the calculations for a z-test. The actual ratios and sample sizes are all that matter (e.g. 120/309 vs. 37000/100000).



I'm not sure I follow what you're asking. I think you're asking the question of why we should assume that WCS's true percentage allowed is the league average 37%, rather than say 39% or even 41%. If that's what you're getting at, I see what you're saying, but it's kind of coming at it from the wrong direction. The reason is that we have a specific question here, and that question is: Is WCS a worse than average defender against 15+ jumpers? We're comparing him to the average because that's what we're interested in - we don't want to assume that he has any particular value of percentage against. We could ask other questions, of course, such as is WCS's percentage allowed statistically better than 45%? (Answer: yes) or is WCS's percentage allowed statistically worse than 35%? (Answer: no. Almost, but not quite at p < 0.05).

As for the partial question of whether the odds are the same that he is average or 4% worse than average...generally, no. The underlying distribution of something like that is usually Gaussian (a "bell curve") where average is most likely and the further you are from average the less likely you'll find someone with that value. For instance if that question were reworded ("aren't the odds the same that WCS allows 37% and that WCS allows 100%?") I suspect you'd intuitively say that the odds that WCS is such a poor defender that he allows 100% is very unlikely. BUT, that's not actually very relevant, either way. We don't actually use knowledge of the underlying distribution of percentage allowed when we do the statistical test.



I think one thing that one finds when they play around with stats in terms of asking questions about statistical significance is that we can't be really rock-solid certain about most sports stats as we'd like to think. I mean, we don't really know who is going to lead the league in scoring next year (Curry? LeBron? Harden? Somebody else?) and sportsbooks take advantage of that uncertainty. But some things are easier than others. I'd suspect (without trying to run any numbers) that you could make a very good case for Cousins to have scored statistically more than average this year, especially if you're looking at per-game numbers. But there's a big difference there...figure a team scores about 100 points and plays about 10 guys, so the average player scores about 10 points more or less. Cousins was around, what, 25 PPG or so? That would be 250% of average - it's not hard to imagine that the data will show he's better than average. On the other hand, the defensive percentage against stats are much tighter. WCS was in ratio terms about 5% worse than average. And you need to have a REALLY big sample to be confident that 5% worse than average really is worse than average rather than maybe just bad luck.

But I don't think that we necessarily HAVE to hold stats to statistical significance. It's perfectly valid to say, hey, WCS was worse than average on defending 15+ foot jumpers this year. It's a small effect, but that's what happened. At the same time, if somebody says they don't think that's necessarily real and points to it not being statistically significant, well, they're right too. There's a good chance that WCS's slightly worse than average 15+ foot J defensive numbers this year were just the result of bad luck. "He was worse than average this year" is right. "There's not very convincing evidence that he's going to be worse than average next year" is also right.

I could say a lot more, but I've probably said way more than enough!

Okay, you just transported me back in time to a lazy May afternoon in Mr. Tolomei's Advanced Math class at Highlands High School. I found myself starting to nod off, much as I did more than once in his class.

The only part of that I understood was the very last sentence.

 

[Capt. Factorial]

The only part of that I understood was the very last sentence.

It was probably the most important sentence, so at least you didn't miss anything much.

 

[bajaden]

Okay, you just transported me back in time to a lazy May afternoon in Mr. Tolomei's Advanced Math class at Highlands High School. I found myself starting to nod off, much as I did more than once in his class.

The only part of that I understood was the very last sentence.

Hey, I have to sit next to this guy at summer league and listen to him. My eyes are usually crossed by the end of a game. The Capt is one smart dude, I refuse to argue with him. I find beer is a better opeion.

 

[bajaden]

Thanks for the answers

WCS won't be better than other rookie PF/C since the only guys doing worse than him that played meaningful minutes in that respect are Okafor and Kaminsky...



I agree with that and agreed with that to start with so most of that was pointless (atleast for you, I learned some new stuff).

I never said I think we can project his next year results from this year, less because of the option of "bad luck" and more because he is a rookie and I think he has a very good chance of being better at it since he doesn't lack the tools needed (unlike a guy like Kaminsky for example)- I just objected to the claim he was really impressive this year in that respect.

Look, I'm not a stat guy, so I'll leave that to you and the Capt. I go on visuals, and personally, I think that can be complicated as well. But I won't go into that. I thought Kaminsky had a good season, but I would never compare Kaminsky to Willie. Two entirely different kind of players. Kaminsky will never be the defender that Willie will be, and it's likely the Willie will never be the post player, or passer that Kaminsky will be. But in general, Willie has a better chance of being an overall better player than Kaminsky. Doesn't mean he will, but that the potential is there.

Probably the best comparison to Wille would be Myles Turner, who came with a similar skill set. Willie is the better athlete but right now, Turner has the better offensive game away from the basket. What the future holds, who knows for sure. They played similar minutes. Turner 22:47 minutes per game, and Willie, 21:23 minutes per game. Turner averaged 10 ppg, and Willie averaged 7 ppg. Of course Turner took almost 200 more shots than Willie, so the points scored differential is explainable. Turner's overall shooting percentage was 49%, and if he hadn't taken a bunch of three point shots, it would have been higher. Willie's overall shooting percentage was 56%, which is where it should be when most of your shots are at or near the basket. All in all, stat wise, they had similar seasons. By the way, for those interested, Turner shot under 30% from the three. He's a better shooter than that, and I expect it will be better next season.