Discussing Other Teams' Moves - Part 6

[Rafael122]

It makes me wonder instead of blowing all the cap space, would we have been better off tacking more $ on Wall's deal like Houston did with Harden.

[Dat2U]

... I'm not saying the Bulls are necessarily a winning team, but they'd have been much worse if they didn't sign players to multi-year deals this offseason.

Oh but you should -- they could flirt with 50 wins. I wonder whether people really appreciate how good Rondo is.

Why, because he's fifth among PGs in WP48? Right behind Jorge Gutierrez. Boy, someone ought to sign that Gutierrez guy. And maybe we should dump Wall because he's only 18th among PGs, behind T.J. McConnell & Pablo Prigioni. You think we could trick Philly into giving us a pick for McConnell? That would be highway robbery in your view, wouldn't it? Then we could sign Prigioni and be set at PG.

The over-dependence on any one stat makes for a dicey game of analysis. I personally like the RPM stat, but it wouldn't dare be my sole analysis of determining who's good or not. I'd look at on/off numbers, BPM, WS48, data from the NBA site, eye test ,etc.

[payitforward]

Oh but you should -- they could flirt with 50 wins. I wonder whether people really appreciate how good Rondo is.

Why, because he's fifth among PGs in WP48? Right behind Jorge Gutierrez. Boy, someone ought to sign that Gutierrez guy. And maybe we should dump Wall because he's only 18th among PGs, behind T.J. McConnell & Pablo Prigioni. You think we could trick Philly into giving us a pick for McConnell? That would be highway robbery in your view, wouldn't it?

The over-dependence on any one stat makes for a dicey game of analysis. I personally like the RPM stat, but it wouldn't dare be my sole analysis of determining who's good or not. I'd look at on/off numbers, BPM, WS48, data from the NBA site, eye test ,etc.

Oh Dat... spare me, ok? Gutierrez played 63 minutes! I assume that any lens through which you look at him those 63 minutes must look good; so what? Prigioni is a heady old man who played 800 minutes; he'd be a good backup for some teams. Doesn't help build your future, but a contending team might like to have him for 800 minutes over lets say Trey Burke? T.J. McConnell has been d*mn good, especially for a kid who went undrafted -- even though he was also very good in college. I'd sure rather have him than Trey Burke! And if you weren't ginning up a pointless and baseless argument, I bet you would too.

Here are the starting pure point guards w/ 2000+ minutes who Wp48 had a better season than John Wall: in order -- Stephen Curry, Russ Westbrook, Chris Paul, Rajon Rondo, Ricky Rubio, Kyle Lowry, Kemba Walker, and George Hill. The further down the list you go, the smaller the difference of course: Hill and Wall were within anyone's idea of a "margin of error."

I assume you have no grief with that list -- except of course for your prejudice against Rondo. & perhaps a similar one against Rubio. Or do you think WP48 has made any other mistakes? Did Wall have a better season than Curry? Westbrook? Lowry? Walker? No? Then maybe it really isn't the work of the devil?

As to "over-dependence on any one stat" being bad -- there are only 2 problems with that idea, aren't there? The first is that once you say "over-dependence" you no longer are making an empirical statement of any kind. Over-dependence on anything is bad by definition. The second is that WP48 is not "one stat." Other than those two little problems, your remark is penetrating.

Btw, that you "personally like the RPM stat" reminds me -- I asked you to tell me whether you had any idea of how it's calculated and if so, to explain how it is. You didn't answer that question. I'll ask it again: do you know how RPM is calculated, DAT? If not, please do tell why you "personally like" it, will you?

But, no, you won't explain -- because you don't know how it's calculated, do you? So... why do you like it? Because it agrees with your opinions when you happen to look at it? That's not usually thought to be much of a measure of anything is it? That something's right because it agrees with me?

Finally -- I'd say most basketball analysts regard Rajon Rondo as one of the best PGs to come into the league in the last 10 years, and most of those analysts have never heard of WP48. Nor did I bring it up.

To tell the truth, I don't think a high opinion of Rondo's career as a PG is a particularly controversial position. And one thing is sure: whether you like him or not doesn't weigh much in the scale of judgment of Rajon Rondo, does it?

Yup, you can bait me into an argument, Dat, little though I want it. What you can't do is win that argument. Once again, how is RPM calculated?

[payitforward]

Dat -- take the above in good humor, please. And, if I bring up WP48 please feel free to tell me why, in the case where I use it, you don't think it's helpful. Otherwise, and especially when I haven't mentioned it and really wasn't thinking about it at all (i.e. most of the time!), maybe you could respond to what I've written with a thought -- you've got plenty of those -- rather than a silly attempt at ridicule.

Btw, I did love your idea that Philly would give us a pick for McConnell, whom they already have -- make that happen!

[nuposse04]

Ironically, there is something humorous about someone asking another to take their condescending BS with a grain of salt.

[bondom34]

Hey pif, just saw this and wanted to post. Some big issues w/ WP. And Rondo's been pretty awful for a few years:

https://ascreamingcomesacrossthecourt.wordpress.com/2012/01/09/how-wins-produced-fails-in-being-the-magical-assessor-of-player-value/

[montestewart]

Hey pif, just saw this and wanted to post. Some big issues w/ WP. And Rondo's been pretty awful for a few years:

https://ascreamingcomesacrossthecourt.wordpress.com/2012/01/09/how-wins-produced-fails-in-being-the-magical-assessor-of-player-value/

Author clearly doesn't understand WP48, which takes into account everything from GED score to softness of toilet paper. It is all encompassing.

[AFM]

Hey pif, just saw this and wanted to post. Some big issues w/ WP. And Rondo's been pretty awful for a few years:

https://ascreamingcomesacrossthecourt.wordpress.com/2012/01/09/how-wins-produced-fails-in-being-the-magical-assessor-of-player-value/

Author clearly doesn't understand WP48, which takes into account everything from GED score to softness of toilet paper. It is all encompassing.

Hehe, come on man, it doesn't do that and you know it!

[montestewart]

Hey pif, just saw this and wanted to post. Some big issues w/ WP. And Rondo's been pretty awful for a few years:

https://ascreamingcomesacrossthecourt.wordpress.com/2012/01/09/how-wins-produced-fails-in-being-the-magical-assessor-of-player-value/

Author clearly doesn't understand WP48, which takes into account everything from GED score to softness of toilet paper. It is all encompassing.

Hehe, come on man, it doesn't do that and you know it!

I'm trying to make up for Nivek's absence.

[AFM]

Author clearly doesn't understand WP48, which takes into account everything from GED score to softness of toilet paper. It is all encompassing.

Hehe, come on man, it doesn't do that and you know it!

I'm trying to make up for Nivek's absence.

I'll help you out:

"In my stuff..."

"...which is barely Replacement Level"

[dckingsfan]

Toronto had a nice hole at backup PF - and they got Sullinger - sigh.

[payitforward]

Hey pif, just saw this and wanted to post. Some big issues w/ WP. And Rondo's been pretty awful for a few years:

https://ascreamingcomesacrossthecourt.wordpress.com/2012/01/09/how-wins-produced-fails-in-being-the-magical-assessor-of-player-value/

I'm not nearly the WP fan-boy Dat seems to think I am. In fact, I'm not sure where he got the idea that I was.

Justin makes some good points in that blog post. But there are some logical and factual gaffes there as well. I'm more interested in the Wizards than I am in this issue, so I don't want to respond at length, and especially not to have a long thread on a subject like this.

Essentially, what WP does is to use box score stats and a lot of clever regressions on them (done with SAS or some other fancy statistical software) to create a metric that correlates as closely as possible to team wins-losses and distributes responsibility for that record correctly across players.

At the team level it's perfectly simple to know if you are right. Add up all the WP for all players on every team and see how the teams rank. The closer that result correlates to their rank by win-loss record, the better the metric. It's important to point out that you don't assess something like WP by asking how close it is to reality -- you just ask if it's closer to reality than any other tool. WP correlates substantially better than PER, for example, that makes it a better metric. Period.

The objections to WP all occur at the level of the individual rather than the team -- i.e. does it distribute credit correctly. The people who work on WP -- sports economists with Phds and a lot of ability and experience in statistical analysis -- care about whether they're right and work hard to make sure the metric does distribute credit correctly, but it's certainly not 100% right (just better than any other such tool -- all you can be).

Then again, the list of top PGs in order of WP48 (Wins Produced for every 48 minutes the player's on the court) does seem more or less what you'd think it should be, doesn't it? Curry, Westbrook, Paul, etc. And in fact, ranking the outstanding players shouldn't be difficult for most such tools -- where it gets "dicey" as Dat describes it is the details of the long list of players where lots of players are separated by very small amounts that it's definitely unclear can be trusted as accurate -- it's questionable whether WP or any such tool is granular enough for the list to accurate at each individual position -- for every player on the list in exact order.

In fact, it's certain that it isn't -- no tool would be. And the WP ideologues (of which there are some to be sure) make their biggest mistake in imagining they can use it in that way.

OTOH, the presence of Rajon Rondo near the top of the list -- the part that's easiest to predict -- is an indicator that, no, he isn't awful. If you use EFF instead of WP, he's still right near the top. If you use EFF40, ditto.

How about PER? It's well known that the main source of inaccuracy in PER is that it rewards taking shots even at a low FG%. As Dat points out Rondo isn't a good shooter, and he doesn't shoot much; he takes 12 shots for every 20 John Wall takes. Hence, sure enough and no surprise, PER rates Rondo much lower than other metrics. But even PER rates him in the top 20% of PGs.

Now, if that lower ranking for Rondo makes you think PER must be more accurate as a metric, then I hope you think Kyrie Irving was the 6th best PG in the league overall last season, because that's where it ranks him. Btw, John Wall ranks about the same place whether using WP or PER.

I use WP, when I do use it, to think about guys we might or might not draft, trade for, or sign as a FA. In fact, I don't even use WP, I just use WS40, a raw box score roll up that's the rough equivalent of WP -- it's better than PER. Since I don't overrate small differences between players, and since I don't rely on any 1 year of this kind of number, I don't trouble myself to get a WP48, esp. not for college guys -- it's hard to get it.

There are plenty of other useful tools I'm sure -- why wouldn't there be? None of them are perfect predictors for any individual player. And ones that vary wildly from year to year (as I'm led to understand RPM does) are useless in deciding if you should acquire a guy: the metric doesn't predict his future, which is what you'd be acquiring!

Ok... back to regularly scheduled programming!

[queridiculo]

There was a lot of value to be had for teams that didn't panic and waited to let the market sort itself out.

Obviously the Wizards weren't one of those team. Not surprising really.

[bondom34]

Hey pif, just saw this and wanted to post. Some big issues w/ WP. And Rondo's been pretty awful for a few years:

https://ascreamingcomesacrossthecourt.wordpress.com/2012/01/09/how-wins-produced-fails-in-being-the-magical-assessor-of-player-value/

I'm not nearly the WP fan-boy Dat seems to think I am. In fact, I'm not sure where he got the idea that I was.

Justin makes some good points in that blog post. But there are some logical and factual gaffes there as well. I'm more interested in the Wizards than I am in this issue, so I don't want to respond at length, and especially not to have a long thread on a subject like this.

Essentially, what WP does is to use box score stats and a lot of clever regressions on them (done with SAS or some other fancy statistical software) to create a metric that correlates as closely as possible to team wins-losses and distributes responsibility for that record correctly across players.

At the team level it's perfectly simple to know if you are right. Add up all the WP for all players on every team and see how the teams rank. The closer that result correlates to their rank by win-loss record, the better the metric. It's important to point out that you don't assess something like WP by asking how close it is to reality -- you just ask if it's closer to reality than any other tool. WP correlates substantially better than PER, for example, that makes it a better metric. Period.

The objections to WP all occur at the level of the individual rather than the team -- i.e. does it distribute credit correctly. The people who work on WP -- sports economists with Phds and a lot of ability and experience in statistical analysis -- care about whether they're right and work hard to make sure the metric does distribute credit correctly, but it's certainly not 100% right (just better than any other such tool -- all you can be).

Then again, the list of top PGs in order of WP48 (Wins Produced for every 48 minutes the player's on the court) does seem more or less what you'd think it should be, doesn't it? Curry, Westbrook, Paul, etc. And in fact, ranking the outstanding players shouldn't be difficult for most such tools -- where it gets "dicey" as Dat describes it is the details of the long list of players where lots of players are separated by very small amounts that it's definitely unclear can be trusted as accurate -- it's questionable whether WP or any such tool is granular enough for the list to accurate at each individual position -- for every player on the list in exact order.

In fact, it's certain that it isn't -- no tool would be. And the WP ideologues (of which there are some to be sure) make their biggest mistake in imagining they can use it in that way.

OTOH, the presence of Rajon Rondo near the top of the list -- the part that's easiest to predict -- is an indicator that, no, he isn't awful. If you use EFF instead of WP, he's still right near the top. If you use EFF40, ditto.

How about PER? It's well known that the main source of inaccuracy in PER is that it rewards taking shots even at a low FG%. As Dat points out Rondo isn't a good shooter, and he doesn't shoot much; he takes 12 shots for every 20 John Wall takes. Hence, sure enough and no surprise, PER rates Rondo much lower than other metrics. But even PER rates him in the top 20% of PGs.

Now, if that lower ranking for Rondo makes you think PER must be more accurate as a metric, then I hope you think Kyrie Irving was the 6th best PG in the league overall last season, because that's where it ranks him. Btw, John Wall ranks about the same place whether using WP or PER.

I use WP, when I do use it, to think about guys we might or might not draft, trade for, or sign as a FA. In fact, I don't even use WP, I just use WS40, a raw box score roll up that's the rough equivalent of WP -- it's better than PER. Since I don't overrate small differences between players, and since I don't rely on any 1 year of this kind of number, I don't trouble myself to get a WP48, esp. not for college guys -- it's hard to get it.

There are plenty of other useful tools I'm sure -- why wouldn't there be? None of them are perfect predictors for any individual player. And ones that vary wildly from year to year (as I'm led to understand RPM does) are useless in deciding if you should acquire a guy: the metric doesn't predict his future, which is what you'd be acquiring!

Ok... back to regularly scheduled programming!

The problem is the bolded isn't away to parse credit among players, the box score can't capture that. WP notoriously overrates rebounds (see Deandre Jordan for MVP as a good idea from the writers) and isn't nearly as good a measure as true PM data. Rondo rates well because his boxs core numbers look great, problem is they don't relate to wins.

The boxscore was not created to measure a player's ability,only his stats. Hence boxscore metrics have much greater weakness in measuring real ability. PER overrates volume scoring, WP rebounds and other stats, WS others still. The biggest problem is none relate well in predictive ability, the metrics which have been shown to predict future success are plus/minus based.

Also,heres an old thread on WP issues too.

viewtopic.php?t=1301834

It's got a wide variety of flaws, ranging from
- assuming that values of certain team statistics are the same on the player level
- failing to realize that defenders that make offensive players miss shots should get some of the credit (and not just the guy who got the defensive rebound). They made some changes to this but probably not enough
- failing to understand that your metric should be able to predict well, and not just correlate with data from the past

That being said, I don't think PER is that much better as a metric, as it has it's own flaws (arbitrary weights, for one). WS also uses arbitrary weights, but at least these are closer to what the 'good' BoxScore metrics (i.e. ASPM) have found they should be

Nothing is perfect, but WP is generally pretty poor at everything, PER not much better to me. Alright....carry on.

[Dat2U]

Oh Dat... spare me, ok? Gutierrez played 63 minutes! I assume that any lens through which you look at him those 63 minutes must look good; so what? Prigioni is a heady old man who played 800 minutes; he'd be a good backup for some teams. Doesn't help build your future, but a contending team might like to have him for 800 minutes over lets say Trey Burke? T.J. McConnell has been d*mn good, especially for a kid who went undrafted -- even though he was also very good in college. I'd sure rather have him than Trey Burke! And if you weren't ginning up a pointless and baseless argument, I bet you would too.

Yes, I'd take McConnell over Burke in a heartbeat, who wouldn't? Same with Prigioni over Burke, because Burke IMO is brutal and basically non-playable. So my question is, would you trade Wall for McConnell and an unprotected 1st round pick. I assume CCJ would. But how bout you?

Here are the starting pure point guards w/ 2000+ minutes who Wp48 had a better season than John Wall: in order -- Stephen Curry, Russ Westbrook, Chris Paul, Rajon Rondo, Ricky Rubio, Kyle Lowry, Kemba Walker, and George Hill. The further down the list you go, the smaller the difference of course: Hill and Wall were within anyone's idea of a "margin of error."

I assume you have no grief with that list -- except of course for your prejudice against Rondo. & perhaps a similar one against Rubio. Or do you think WP48 has made any other mistakes? Did Wall have a better season than Curry? Westbrook? Lowry? Walker? No? Then maybe it really isn't the work of the devil?

I'd say the top PGs are Curry, Westbrook, Paul & Lowry. Rubio had a very nice year and is a helluva defender. I don't have a problem saying Rubio outplayed Wall last year, especially on the defensive side of the ball. I'd say Walker and Wall had comparable years. You can make an argument & give it to Walker by a hair. So that makes Wall 7th. I disagree with Rondo who's usefulness has always been wildly overrated, even in his prime. He was surrounded by 3 HOFs in Boston! Since then his defense has fell off a cliff and offensively he seems stat hungry in his search for assists. He's not been the greatest teammate either. I liked George Hill for years and viewed him as very underrated but his defense IMO slipped last year (which is why Indy was willing to deal him). In past years, I might have rated him with Wall in terms of impact, but not last year.

As to "over-dependence on any one stat" being bad -- there are only 2 problems with that idea, aren't there? The first is that once you say "over-dependence" you no longer are making an empirical statement of any kind. Over-dependence on anything is bad by definition. The second is that WP48 is not "one stat." Other than those two little problems, your remark is penetrating.

Ok, WP48 is a rollup of stats into one number. It's still a poor measurement. All of your previous analysis appears to match WP48 to a tee. When you said Rondo was really good, I automatically knew what that meant, he rated well in the WP48 metric.

Btw, that you "personally like the RPM stat" reminds me -- I asked you to tell me whether you had any idea of how it's calculated and if so, to explain how it is. You didn't answer that question. I'll ask it again: do you know how RPM is calculated, DAT? If not, please do tell why you "personally like" it, will you?

But, no, you won't explain -- because you don't know how it's calculated, do you? So... why do you like it? Because it agrees with your opinions when you happen to look at it? That's not usually thought to be much of a measure of anything is it? That something's right because it agrees with me?

RPM is a proprietary stat purchased by ESPN. So no, we don't know all the details of it. I respect the creator of the stat, Jeremias Englemann a good deal and he's known as one of the best analytics guys in the business. RPM is supposed to be an improvement on xRAPM and other APM-type metrics which I find quite useful and while not perfect do a pretty good job historically of measuring on-court effectiveness. There's a 90% correlation b/w RPM & RAPM as well as a strong relationship with WS48 because a good deal of it is based off the box-score information... and yes, I find it does often match well with my own visual analysis. I thought Beal & Neal couldn't defend a chair last year and it validated that opinion. I thought Hump was a terrible option at starting PF and it validated that opinion. I thought Markieff Morris improved after settling in a bit and it showed in RPM but no I don't think it's a perfect stat. Merely a good one.

Finally -- I'd say most basketball analysts regard Rajon Rondo as one of the best PGs to come into the league in the last 10 years, and most of those analysts have never heard of WP48. Nor did I bring it up.

To tell the truth, I don't think a high opinion of Rondo's career as a PG is a particularly controversial position. And one thing is sure: whether you like him or not doesn't weigh much in the scale of judgment of Rajon Rondo, does it?

Most basketball analysis is superficial. Like saying Derrick Rose helps the Knicks and can take the pressure off Melo by being a 2nd star... even though by ALL advanced measures, he was one of the worst starters in the entire league at any position. I hope your not relying on it to validate your opinion. That would be so unlike you.

[dckingsfan]

Wow, I am going to go hide in the politics thread

[closg00]

Toronto had a nice hole at backup PF - and they got Sullinger - sigh.

1-year, $6 million = smart signing

[montestewart]

Wow, I am going to go hide in the politics thread

Seriously, when PIF and Dat go at it, they make Trump-Ginsburg look like a children's tea party

[long suffrin' boulez fan]

Toronto had a nice hole at backup PF - and they got Sullinger - sigh.

1-year, $6 million = smart signing

Jordan Hill. Two years, 4 mil per.

It's a sad day when the Twolves start out maneuvering us.

[dckingsfan]

Toronto had a nice hole at backup PF - and they got Sullinger - sigh.

1-year, $6 million = smart signing

Jordan Hill. Two years, 4 mil per.

It's a sad day when the Twolves start out maneuvering us.

UGH!