2017-18 RAPM/RPM/etc. Thread

[dhsilv2]

Looked at RAPM on the first page... saw that Harden is at 53, and Lebron at present is at 78. More reasons for not putting much stock into these metrics.

This is the tiering structure for RAPM amongst allstars/well known players (top 22)

Curry

Embiid
Butler

Kemba Walker
Giannis
Oladipo

AD
Klay
Westbrook
Derozan
Draymond

Beal
Jokic
Aldridge
CP3

Towns
Lowry
Harden
Horford
Porzingis
Wall
Durant

Legitimately tells me nothing

It's vanilla RAPM and it isn't a player ranking system.

It is often used heavily in player rankings, and as an ultimate trump card. I have seen these numbers being used this season. So yes, if people use it more sensibly, then I'd have no issue.

Prior informed RAPM is much more commonly used. Vanilla isn't that commonly used, and generally it isn't until year end that people really push it.

[clyde21]

Yea, I don't even understand discussing a stat that we don't know the details of it.

another stat used to try and prop curry up.. dont worry about it

What stat should we be using?

[NO-KG-AI]

The height isn't consistent at all though. Forget the fact that height doesn't really matter in terms of impact, but some guys are listed in shoes, some guys are listed in bare feet, some guys are embellished, some guys aren't documented by anyone but their own team or agent, and some guys have grown since their height was listed.

It's the stupidest thing to add into a statistic that is aiming to measure impact that I ever saw. Why not just throw weight in there too?

Bigs impact defense more than smalls.

Therefore we can rank defense solely based on height. Guys that are 6'4 are better than the guys that are 6'3. Wow, this is easy.

Also, don't forget those guys that are 7' with a 6'10 standing reach, and how elite they are as defenders.

[therealbig3]

Meh, a lot of this just seems like trying to dismiss a stat that paints a picture you don't like.

If JE wanted to prop up certain star players and messes with the numbers to get the results he wants, why is Harden elite this year, but wasn't elite in 16 or 17, when his narrative was pretty much exactly the same as it is now? Why isn't Durant anywhere close to the top? Why is it consistently LeBron and Curry that are near the top?

No, it's not a perfect stat, but I don't really see the issue with it. It's jiving with everything I'm seeing so far.

[Dr Spaceman]

The height isn't consistent at all though. Forget the fact that height doesn't really matter in terms of impact, but some guys are listed in shoes, some guys are listed in bare feet, some guys are embellished, some guys aren't documented by anyone but their own team or agent, and some guys have grown since their height was listed.

It's the stupidest thing to add into a statistic that is aiming to measure impact that I ever saw. Why not just throw weight in there too?

It’s fair to bring this up as a point, but I think there’s something that’s not super clear.

From everything I’ve read, this is an evidence-based statistic. Meaning the formula he has today is the result of countless “experiments” to find the right combination of variables to give the best and most consistent measure of impact he can.

Put simply: he includes height as a variable in defense because it’s shown that taller players consistently are better at defense.

Put in more complicated terms: adding the height coefficient made the curve line more closely predict future team point differential results.

So if you’re criticizing him because you think he haphazardly threw that in there because of intuition, well that’s not how it works. If t didn’t give the stat better results, it wouldn’t be in there.

I recommend reading some of the early threads of ver on APBR. It helps a lot.

[The-Power]

The height isn't consistent at all though. Forget the fact that height doesn't really matter in terms of impact, but some guys are listed in shoes, some guys are listed in bare feet, some guys are embellished, some guys aren't documented by anyone but their own team or agent, and some guys have grown since their height was listed.

It's the stupidest thing to add into a statistic that is aiming to measure impact that I ever saw. Why not just throw weight in there too?

Bigs impact defense more than smalls.

I might misremember that, but wasn't height just used for the offensive/defensive splits and not for overall RPM?

[Pillendreher]

The height isn't consistent at all though. Forget the fact that height doesn't really matter in terms of impact, but some guys are listed in shoes, some guys are listed in bare feet, some guys are embellished, some guys aren't documented by anyone but their own team or agent, and some guys have grown since their height was listed.

It's the stupidest thing to add into a statistic that is aiming to measure impact that I ever saw. Why not just throw weight in there too?

It’s fair to bring this up as a point, but I think there’s something that’s not super clear.

From everything I’ve read, this is an evidence-based statistic. Meaning the formula he has today is the result of countless “experiments” to find the right combination of variables to give the best and most consistent measure of impact he can.

Put simply: he includes height as a variable in defense because it’s shown that taller players consistently are better at defense.

Put in more complicated terms: adding the height coefficient made the curve line more closely predict future team point differential results.

So if you’re criticizing him because you think he haphazardly threw that in there because of intuition, well that’s not how it works. If t didn’t give the stat better results, it wouldn’t be in there.

I recommend reading some of the early threads of ver on APBR. It helps a lot.

Something I've always wondered about: Isn't this working backwards from a given conclusion/result? If you create a stat based on what you think it should show, isn't it completely redundant?

[Dr Spaceman]

The height isn't consistent at all though. Forget the fact that height doesn't really matter in terms of impact, but some guys are listed in shoes, some guys are listed in bare feet, some guys are embellished, some guys aren't documented by anyone but their own team or agent, and some guys have grown since their height was listed.

It's the stupidest thing to add into a statistic that is aiming to measure impact that I ever saw. Why not just throw weight in there too?

It’s fair to bring this up as a point, but I think there’s something that’s not super clear.

From everything I’ve read, this is an evidence-based statistic. Meaning the formula he has today is the result of countless “experiments” to find the right combination of variables to give the best and most consistent measure of impact he can.

Put simply: he includes height as a variable in defense because it’s shown that taller players consistently are better at defense.

Put in more complicated terms: adding the height coefficient made the curve line more closely predict future team point differential results.

So if you’re criticizing him because you think he haphazardly threw that in there because of intuition, well that’s not how it works. If t didn’t give the stat better results, it wouldn’t be in there.

I recommend reading some of the early threads of ver on APBR. It helps a lot.

Something I've always wondered about: Isn't this working backwards from a given conclusion/result? If you create a stat based on what you think it should show, isn't it completely redundant?

I might be misunderstanding your question here, but my point was that this is working backwards from how well the data fits with real life results. JE isn’t picking formulas because of any player result.

He’s using a technique called out of sample prediction. What that means is, if he has a stat that can pick out the point differential of each player, then it follows that it should be able to predict the actual results of the season when those players team up. When his model can do that within a reasonable margin, he has a winning formula. That’s the standard it’s measured against.

[Jaivl]

.

Something I've always wondered about: Isn't this working backwards from a given conclusion/result? If you create a stat based on what you think it should show, isn't it completely redundant?

As far as I know, it's a stat created to maximize prediction of future point differential -> wins. Wins are pretty much the ultimate goal in basketball, hence being the gold standard to aspire to if you want a one-number predictive metric.

I'm guessing the process for the prior was as follows:
- Select data from older season/seasons (for example: 2005)
- Select certain variables (for example: points per 100, steal%, height, free throws attempted), or combinations of them (height*3ptscored is a combination I've seen used)
- Try to "predict" the next season's wins (2006) with only those 2005 variables (I guess he used some kind of minute-prediction function there) using some kind of regression method. That outputs a series of coefficients that best "fit", or in this case "predict" the results. For example, (0.03*heightinches + 0.5*steal% - 0.2*ftattempted + 0.44*pp100).
An explicative stat (BPM) is usually designed to best fit a known RAPM sample, instead of trying to predict future events.
- Compare the "predictive power" of that formula vs actual 2006 results (R^2)
- Select other variables, repeat process, compare the R^2 with the other formulas, select the one that gives the best results.

If the prior is "biased" in favour of taller players it's because it happens that having height in the formula produces the best results. Nothing more.

[Pillendreher]

It’s fair to bring this up as a point, but I think there’s something that’s not super clear.

From everything I’ve read, this is an evidence-based statistic. Meaning the formula he has today is the result of countless “experiments” to find the right combination of variables to give the best and most consistent measure of impact he can.

Put simply: he includes height as a variable in defense because it’s shown that taller players consistently are better at defense.

Put in more complicated terms: adding the height coefficient made the curve line more closely predict future team point differential results.

So if you’re criticizing him because you think he haphazardly threw that in there because of intuition, well that’s not how it works. If t didn’t give the stat better results, it wouldn’t be in there.

I recommend reading some of the early threads of ver on APBR. It helps a lot.

Something I've always wondered about: Isn't this working backwards from a given conclusion/result? If you create a stat based on what you think it should show, isn't it completely redundant?

I might be misunderstanding your question here, but my point was that this is working backwards from how well the data fits with real life results. JE isn’t picking formulas because of any player result.

He’s using a technique called out of sample prediction. What that means is, if he has a stat that can pick out the point differential of each player, then it follows that it should be able to predict the actual results of the season when those players team up. When his model can do that within a reasonable margin, he has a winning formula. That’s the standard it’s measured against.

.

Something I've always wondered about: Isn't this working backwards from a given conclusion/result? If you create a stat based on what you think it should show, isn't it completely redundant?

As far as I know, it's a stat created to maximize prediction of future point differential -> wins. Wins are pretty much the ultimate goal in basketball, hence being the gold standard to aspire to if you want a one-number predictive metric.

I'm guessing the process for the prior was as follows:
- Select data from older season/seasons (for example: 2005)
- Select certain variables (for example: points per 100, steal%, height, free throws attempted), or combinations of them (height*3ptscored is a combination I've seen used)
- Try to "predict" the next season's wins (2006) with only those 2005 variables (I guess he used some kind of minute-prediction function there) using some kind of regression method. That outputs a series of coefficients that best "fit", or in this case "predict" the results. For example, (0.03*heightinches + 0.5*steal% - 0.2*ftattempted + 0.44*pp100).
An explicative stat (BPM) is usually designed to best fit a known RAPM sample, instead of trying to predict future events.
- Compare the "predictive power" of that formula vs actual 2006 results (R^2)
- Select other variables, repeat process, compare the R^2 with the other formulas, select the one that gives the best results.

If the prior is "biased" in favour of taller players it's because it happens that having height in the formula produces the best results. Nothing more.

Oh, OK. That's a different approach then. I thought you meant that the stat might be evaluated by looking at it and adjusting it because you consider it off based on your own 'assupmtions'.

[bondom34]

The height isn't consistent at all though. Forget the fact that height doesn't really matter in terms of impact, but some guys are listed in shoes, some guys are listed in bare feet, some guys are embellished, some guys aren't documented by anyone but their own team or agent, and some guys have grown since their height was listed.

It's the stupidest thing to add into a statistic that is aiming to measure impact that I ever saw. Why not just throw weight in there too?

Bigs impact defense more than smalls.

Therefore we can rank defense solely based on height. Guys that are 6'4 are better than the guys that are 6'3. Wow, this is easy.

Also, don't forget those guys that are 7' with a 6'10 standing reach, and how elite they are as defenders.

I never said that, but OK. A big has an advantage in impacting defense, but doesn't solely impact it more due to height. It is a factor. Like in an equation.

[bondom34]

The height isn't consistent at all though. Forget the fact that height doesn't really matter in terms of impact, but some guys are listed in shoes, some guys are listed in bare feet, some guys are embellished, some guys aren't documented by anyone but their own team or agent, and some guys have grown since their height was listed.

It's the stupidest thing to add into a statistic that is aiming to measure impact that I ever saw. Why not just throw weight in there too?

Bigs impact defense more than smalls.

I might misremember that, but wasn't height just used for the offensive/defensive splits and not for overall RPM?

Not sure.

[NO-KG-AI]

Bigs impact defense more than smalls.

Therefore we can rank defense solely based on height. Guys that are 6'4 are better than the guys that are 6'3. Wow, this is easy.

Also, don't forget those guys that are 7' with a 6'10 standing reach, and how elite they are as defenders.

I never said that, but OK. A big has an advantage in impacting defense, but doesn't solely impact it more due to height. It is a factor. Like in an equation.

Big men have a higher ceiling defensively, but just being big doesn't automatically make you more impactful, or have an impact at all. This is automatically giving credit just for being tall, even though you could be the worst defender ever.

[bondom34]

Therefore we can rank defense solely based on height. Guys that are 6'4 are better than the guys that are 6'3. Wow, this is easy.

Also, don't forget those guys that are 7' with a 6'10 standing reach, and how elite they are as defenders.

I never said that, but OK. A big has an advantage in impacting defense, but doesn't solely impact it more due to height. It is a factor. Like in an equation.

Big men have a higher ceiling defensively, but just being big doesn't automatically make you more impactful, or have an impact at all. This is automatically giving credit just for being tall, even though you could be the worst defender ever.

Agree there, but in honesty there is some (minor) credit for just being tall. A guy who's 7 feet with slow feet and poor shot blocking and quickness is likely a better defender than one who's 6 feet, based on size alone. Just being a physical presence is often deterrent.

[Jaivl]

Big men have a higher ceiling defensively, but just being big doesn't automatically make you more impactful, or have an impact at all. This is automatically giving credit just for being tall, even though you could be the worst defender ever.

PPG is automatically giving cretit just for shooting, even though you could be the worst scorer ever. Oh noes.

[dhsilv2]

Therefore we can rank defense solely based on height. Guys that are 6'4 are better than the guys that are 6'3. Wow, this is easy.

Also, don't forget those guys that are 7' with a 6'10 standing reach, and how elite they are as defenders.

I never said that, but OK. A big has an advantage in impacting defense, but doesn't solely impact it more due to height. It is a factor. Like in an equation.

Big men have a higher ceiling defensively, but just being big doesn't automatically make you more impactful, or have an impact at all. This is automatically giving credit just for being tall, even though you could be the worst defender ever.

By year end most of the prior especially a minor piece is going to work itself out.

It still figured out 7 footer KAT was an awful defender (-1.41) despite being one of the better defensive rebounders in the league, despite blocking a lot of shots, and despite this "tall prior". Now without all that, I'd assume RAPM without a box score prior etc would have scored him lower, but the stat isn't grossly blundering.

[NO-KG-AI]

Big men have a higher ceiling defensively, but just being big doesn't automatically make you more impactful, or have an impact at all. This is automatically giving credit just for being tall, even though you could be the worst defender ever.

PPG is automatically giving cretit just for shooting, even though you could be the worst scorer ever. Oh noes.

Not even close to the same thing.

[Texas Chuck]

Big men have a higher ceiling defensively, but just being big doesn't automatically make you more impactful, or have an impact at all. This is automatically giving credit just for being tall, even though you could be the worst defender ever.

PPG is automatically giving cretit just for shooting, even though you could be the worst scorer ever. Oh noes.

Not even close to the same thing.

it actually kind of is.

But let's ignore that since its bothering you. Don't these formulas usually give you credit for attempting 3 pt shots even if you aren't good at making them? With the idea that just taking 3's helps spacing? If we are okay with that notion, we should be okay with the height for defense one.


I actually think there are too many "subjective" decisions being made and then selling the end product as "objective", but I don't really want to be labeled the anti +/- guy despite my continue belief that none of these versions do nearly as good a job isolating individual player impact as they claim they do.

[NO-KG-AI]

PPG is automatically giving cretit just for shooting, even though you could be the worst scorer ever. Oh noes.

Not even close to the same thing.

it actually kind of is.

But let's ignore that since its bothering you. Don't these formulas usually give you credit for attempting 3 pt shots even if you aren't good at making them? With the idea that just taking 3's helps spacing? If we are okay with that notion, we should be okay with the height for defense one.


I actually think there are too many "subjective" decisions being made and then selling the end product as "objective", but I don't really want to be labeled the anti +/- guy despite my continue belief that none of these versions do nearly as good a job isolating individual player impact as they claim they do.

If it's giving credit for taking and missing 3's, that's just as bad. I'm not okay with anything like that.

Also, it's not the same. PPG doesn't give credit for just taking shots, because they still actually have to go in, and PPG isn't a measure of impact, it's just how many points you scored. This is trying to assign how much lift you give a team, and gives you bonus points for a physical trait that may or may not have any impact at all for a particular player, and isn't even a correct measure for MOST NBA players.

[TheSuzerain]

I think there's some weird stuff going on with RPM this year. It looks more constrained at the top and bottom (pulled towards 0) compared to previous seasons.

Anybody?