With or Without You - How Does This Core Fit Together? - Page 2
Old 08-02-2013, 01:46 PM   #21 (permalink)
whatever

Senior Member
 
Join Date: Dec 2007
Location: toronto
Posts: 10,287
Representing:
Default

Quote:
Originally Posted by woodchuck View Post
nbawowy!, it's a new site built on that principle earlier this year, a lot of good info there.

I wish we could take a look inside Darryl Morey's or Marc Cuban's stats software. Judging by their occasional tweets, I think there are stats that do an impressive job looking at defense from various angles, but sadly they aren't in the public domain.
this site is amazing btw, I wonder if they plan to keep it free in the long run
moremilk is online now   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 01:48 PM   #22 (permalink)
the gat'll killya quicker, when I'm drunk off the liquor

The Mara sisters are hot!
 
Bill Haverchuck's Avatar
 
Join Date: Oct 2009
Posts: 10,467
Representing:
Default

+1 for all the interesting info and hard work. Thanks, Dan.

Re: Amir

I think this is basically what LX was getting at when he said that losing Amir - even in a trade for someone as good as Horford - would cause LX some anxiety (paraphrasing). Now, it still might be a stretch to say Amir is better than Horford, but he might actually be just as valuable as someone like Millsap.

I like what these stats show about the core. The unfortunate thing, for those of us who want to make the playoffs, is that our bench is still a huge question mark and could be horriawful.
Bill Haverchuck is offline   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 01:51 PM   #23 (permalink)
Is wondering when the Patman and JJ will be unleashed?

Membres supérieurs
 
DocHoliday99's Avatar
 
Join Date: Dec 2010
Location: Southern Alberta
Posts: 5,214
Representing:
Default

Quote:
Originally Posted by DanH View Post
You make a good point on Lowry - he was much improved in the second half, just like JV.

I went back and refined my 1st half versus 2nd half numbers (estimated teammates' minutes played, ORTG and DRTG before as half a season, now updated for actual 1st and 2nd half minutes played and RTGs as per that WOWY site - it is pretty darn useful), and did Lowry too.

See post for updated JV numbers, and I'll put Lowry's here as well:

Player - ORTG impact from KL 1st Half - DRTG impact from KL 1st Half - Net RTG impact from KL 1st Half
DD: +1.1 / +4.2 / -3.1
Amir: +11.3 / +6.0 / +5.3
JV: -1.9 / +2.1 / -4.0

Player - ORTG impact from KL 2nd Half - DRTG impact from KL 2nd Half - Net RTG impact from KL 2nd Half
DD: +2.8 / -8.3 / +11.1
Amir: +5.0 / -6.6 / +11.6
JV: -1.3 / -10.2 / +8.9
Curious, DD was supposed to be playing better during the 2nd half as well, I'm curious to find out if that is true or not. Observations say yes but eyes can't always be trustworthy.
DocHoliday99 is online now   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 01:51 PM   #24 (permalink)
the gat'll killya quicker, when I'm drunk off the liquor

The Mara sisters are hot!
 
Bill Haverchuck's Avatar
 
Join Date: Oct 2009
Posts: 10,467
Representing:
Default

Quote:
Originally Posted by moremilk View Post
also, great work, but (and I'm a major fan of numerical analysis in the nba) I have a lot of skepticism for any kind of so called advanced stats that are supposed to capture defensive impact and even overall impact. While pace or mpg adjusted basic stats and even things like PER are very reliable (even though PER is so often misused as a measure of a player's value), every advanced stat I've seen has at least a few if not many outliers. Like the warp based ranking of PGs that had parker way down the list despite being the best or second best PG last year.

Now the problem is that even a small number of outliers is enough to make the stat very unreliable. If you have to say, well Durant scores bad on this stat, but that's because so and so, how can you trust this stat for any other player when you can always make so and so arguments for them too.

For example, your numbers above make acy the most impactful player on our team. Now, that's clearly not the case so nobody takes it seriously. At the same time, Amir has similar numbers and we say, wow, amir is really good. But that's only because we have a pre-conceived opinion that amir is really good and acy not so much.

The ultimate litmus test is if we were to show these stats to somebody who doesn't watch our games and has no idea who the players are, how will we be able to explain to them why we choose to trust amir's numbers but discount Acy's?
What?

Amir is playing more minutes, and against better competition. I think the sample size is why people react differently to Amir's numbers versus Acy's, not some preconceived notion about Amir. People look at that and realize that Acy is not playing with the other guys against starter's, he plays with them against bench players and it happens in garbage minutes, for the most part.
Bill Haverchuck is offline   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 01:58 PM   #25 (permalink)
Is wondering when the Patman and JJ will be unleashed?

Membres supérieurs
 
DocHoliday99's Avatar
 
Join Date: Dec 2010
Location: Southern Alberta
Posts: 5,214
Representing:
Default

Quote:
Originally Posted by Bill Haverchuck View Post
What?

Amir is playing more minutes, and against better competition. I think the sample size is why people react differently to Amir's numbers versus Acy's, not some preconceived notion about Amir. People look at that and realize that Acy is not playing with the other guys against starter's, he plays with them against bench players and it happens in garbage minutes, for the most part.
Excellent rebuttal. Not like Acy is up against the Bosh's and Duncan's and Gasol's of the league as Amir is
DocHoliday99 is online now   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 01:59 PM   #26 (permalink)
the gat'll killya quicker, when I'm drunk off the liquor

The Mara sisters are hot!
 
Bill Haverchuck's Avatar
 
Join Date: Oct 2009
Posts: 10,467
Representing:
Default

Quote:
Originally Posted by DanH View Post
Player - ORTG impact from Bargs - DRTG impact from Bargs - Net RTG impact from Bargs
DD: -0.7 / +3.5 / -4.2
Amir: +0.8 / +4.6 / -3.8
Lowry: -5.5 / +4.1 / -9.5
JV: -1.5 / +5.0 / -6.5
Gay: +0.4 / +7.8 / -7.4
And a big thanks for putting this one together, Dan. It may not be conclusive evidence about the "addition by subtraction" argument, but it certainly leans towards the theory having merit. Several discussions about this have come up, and a couple more will probably pop up, and at least now we have another useful reference point for assessing his negative impact.

Last edited by Bill Haverchuck; 08-02-2013 at 02:09 PM. Reason: spelling
Bill Haverchuck is offline   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 02:03 PM   #27 (permalink)
rb
let the games begin

Senior Member
 
rb's Avatar
 
Join Date: Jul 2009
Location: Vaughan
Posts: 958
Representing:
Default

awesome post dan, thanks.
rb is offline   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 02:06 PM   #28 (permalink)
Is wondering when the Patman and JJ will be unleashed?

Membres supérieurs
 
DocHoliday99's Avatar
 
Join Date: Dec 2010
Location: Southern Alberta
Posts: 5,214
Representing:
Default

Quote:
Originally Posted by Bill Haverchuck View Post
And a big thanks for putting this one together, Dan. It may not be conclusive evidence about the "addition by subtraction" argument, but it certainly leans towards the theory having merit. Several discussions about this have come up, and a couple more will probably pop up, and at least know we have another useful reference point for assessing his negative impact.
That and the team started winning in the second half when he wasn't playing. Between that and this? Hard to refute 'addition by subtraction' I would think.
DocHoliday99 is online now   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 02:13 PM   #29 (permalink)
sticking to his guns... is stuck to his guns...

Senior Member

 
DanH's Avatar
 
Join Date: Jul 2010
Location: Cambridge, ON
Posts: 5,120
Representing:
Default

Quote:
Originally Posted by Bill Haverchuck View Post
What?

Amir is playing more minutes, and against better competition. I think the sample size is why people react differently to Amir's numbers versus Acy's, not some preconceived notion about Amir. People look at that and realize that Acy is not playing with the other guys against starter's, he plays with them against bench players and it happens in garbage minutes, for the most part.
Yeah, this is my response too. Amir's sample size is huge, and he's played against all kinds of opposition. Acy's is miniscule (that's why I didn't include him, even though he is pretty remarkable in that SSS) and he only really played in garbage time.

Of course, people said the same thing about Amir's crazy plus-minus numbers when he first came here, and his have somehow held up in every scenario and no matter how large the sample got. So there's certainly a chance Acy really is that good - but there's not nearly enough proof of of that to make any predictions.
DanH is offline   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 02:21 PM   #30 (permalink)
whatever

Senior Member
 
Join Date: Dec 2007
Location: toronto
Posts: 10,287
Representing:
Default

Quote:
Originally Posted by Bill Haverchuck View Post
What?

Amir is playing more minutes, and against better competition. I think the sample size is why people react differently to Amir's numbers versus Acy's, not some preconceived notion about Amir. People look at that and realize that Acy is not playing with the other guys against starter's, he plays with them against bench players and it happens in garbage minutes, for the most part.
see, that's exactly what I mean. There is no place in that formula for things like minutes and competition. This is just something you bring up to justify your pre-conception. A formula is only good if it's reliable in itself or if any exceptions can be consistently explained using the same arguments.

For example, I would be ok with a formula that favours bench players over starters because that can become an automatic adjustment. As long as we never compare a bench guy with a starter, the numbers can be taken at face value.

Which is not the case here, because ed is clearly a better player than acy, and yet he has a rating of -4 vs acy's 17. Also, amir played half the season on the bench too, but that's a different matter.

Just to clarify - I am not disputing that amir is great for us, after all I was one of the small minority who advocated him as starter from day one last season. And I also liked Acy last season.
moremilk is online now   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 02:21 PM   #31 (permalink)
Is wondering when the Patman and JJ will be unleashed?

Membres supérieurs
 
DocHoliday99's Avatar
 
Join Date: Dec 2010
Location: Southern Alberta
Posts: 5,214
Representing:
Default

It just dawned on me why there is such a huge jump from the first half to the second half of the season, summed up in 1 word:

Chemistry
DocHoliday99 is online now   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 02:25 PM   #32 (permalink)
whatever

Senior Member
 
Join Date: Dec 2007
Location: toronto
Posts: 10,287
Representing:
Default

Quote:
Originally Posted by DocHoliday99 View Post
It just dawned on me why there is such a huge jump from the first half to the second half of the season, summed up in 1 word:

Chemistry
or maybe these three words: strength of schedule ...
moremilk is online now   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 02:27 PM   #33 (permalink)
.

Senior Member
 
carp's Avatar
 
Join Date: Jul 2009
Location: Toronto
Posts: 11,494
Representing:
Default

Quote:
Originally Posted by moremilk View Post
or maybe these three words: strength of schedule ...
I think it could be best explained in more than seven words....
carp is offline   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 02:33 PM   #34 (permalink)
sticking to his guns... is stuck to his guns...

Senior Member

 
DanH's Avatar
 
Join Date: Jul 2010
Location: Cambridge, ON
Posts: 5,120
Representing:
Default

Quote:
Originally Posted by moremilk View Post
or maybe these three words: strength of schedule ...
I've looked at the strength of schedule argument, and it just doesn't add up to me. I can do an expected results analysis if you like, but as I recall the expected wins for the tough part of our schedule versus the latter half was something along the lines of 2 or 3 wins less over the entire span.
DanH is offline   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 02:33 PM   #35 (permalink)
Is wondering when the Patman and JJ will be unleashed?

Membres supérieurs
 
DocHoliday99's Avatar
 
Join Date: Dec 2010
Location: Southern Alberta
Posts: 5,214
Representing:
Default

Quote:
Originally Posted by moremilk View Post
see, that's exactly what I mean. There is no place in that formula for things like minutes and competition. This is just something you bring up to justify your pre-conception. A formula is only good if it's reliable in itself or if any exceptions can be consistently explained using the same arguments.

For example, I would be ok with a formula that favours bench players over starters because that can become an automatic adjustment. As long as we never compare a bench guy with a starter, the numbers can be taken at face value.

Which is not the case here, because ed is clearly a better player than acy, and yet he has a rating of -4 vs acy's 17. Also, amir played half the season on the bench too, but that's a different matter.

Just to clarify - I am not disputing that amir is great for us, after all I was one of the small minority who advocated him as starter from day one last season. And I also liked Acy last season.
Love Acy as well, I would love to see him as a back-up and not a 3rd stringer.

As for the stats you make a valid point in so far that you would need to understand the competition the player is facing. If you do, then the stats have something of value to offer - and if you're looking at them you probably understand that already. I think I've read where DanH and Bill and a couple others talk about that, coupling the stats with observations as they are unreliable on their own.

Last edited by DocHoliday99; 08-02-2013 at 02:36 PM.
DocHoliday99 is online now   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 02:35 PM   #36 (permalink)
Is wondering when the Patman and JJ will be unleashed?

Membres supérieurs
 
DocHoliday99's Avatar
 
Join Date: Dec 2010
Location: Southern Alberta
Posts: 5,214
Representing:
Default

Quote:
Originally Posted by moremilk View Post
or maybe these three words: strength of schedule ...
I was only half serious lol but honestly, their play was much improved in the second half regardless of their schedule.
DocHoliday99 is online now   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 02:41 PM   #37 (permalink)
the gat'll killya quicker, when I'm drunk off the liquor

The Mara sisters are hot!
 
Bill Haverchuck's Avatar
 
Join Date: Oct 2009
Posts: 10,467
Representing:
Default

Quote:
Originally Posted by moremilk View Post
see, that's exactly what I mean. There is no place in that formula for things like minutes and competition. This is just something you bring up to justify your pre-conception. A formula is only good if it's reliable in itself or if any exceptions can be consistently explained using the same arguments.
Except in this case, the difficulty of competition and minutes heavily favour Amir, so it's not about preconceived notions. Amir genuinely has the harder task of playing in minutes where the opposition is trying to win the game, and perfoming at a high level consistently.

If we tried to make excuses for Ed, that would suggest some preconceived notions since he also played some minutes against quality opponents and doesn't look as valuable as we may have thought.

We're talking about the "core" of the team, which involves other players getting significant minutes. So the context is always starters with guys playing significant minutes against other starters; that's why JV and Bargnani or Ed can included although they were not fulltime starters. They still played significant minutes against them.

Quote:
For example, I would be ok with a formula that favours bench players over starters because that can become an automatic adjustment. As long as we never compare a bench guy with a starter, the numbers can be taken at face value.
None of us were wanting to compare a bench guy with a starter. We were looking at Amir in the context of what he does with the core, who are largely playing against units that have 3-5 opposing starters on the court; that's higher level competition.

You brought up Acy. None of us would ever have thought to do that, because it's not a good way to look at it.

Quote:
Which is not the case here, because ed is clearly a better player than acy, and yet he has a rating of -4 vs acy's 17. Also, amir played half the season on the bench too, but that's a different matter.
Right, but people, like you and I, who watch enough ball, would know to say, hold up, Ed played a different type of minutes, so we wouldn't except the comparison as convincing. Now, an Ed versus Amir comparison, on the other hand would make some sense. And we might look at that if, say, a team came calling to make a trade and wanted one of them included in the deal. This type of stat could be useful in that sense. I think it adds some worthwhile insight.

"You can have Ed! You're not touching Amir!"

Quote:
Just to clarify - I am not disputing that amir is great for us, after all I was one of the small minority who advocated him as starter from day one last season. And I also liked Acy last season.
Noted.
Bill Haverchuck is offline   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 02:46 PM   #38 (permalink)
sticking to his guns... is stuck to his guns...

Senior Member

 
DanH's Avatar
 
Join Date: Jul 2010
Location: Cambridge, ON
Posts: 5,120
Representing:
Default

Quote:
Originally Posted by moremilk View Post
see, that's exactly what I mean. There is no place in that formula for things like minutes and competition. This is just something you bring up to justify your pre-conception. A formula is only good if it's reliable in itself or if any exceptions can be consistently explained using the same arguments.

For example, I would be ok with a formula that favours bench players over starters because that can become an automatic adjustment. As long as we never compare a bench guy with a starter, the numbers can be taken at face value.

Which is not the case here, because ed is clearly a better player than acy, and yet he has a rating of -4 vs acy's 17. Also, amir played half the season on the bench too, but that's a different matter.

Just to clarify - I am not disputing that amir is great for us, after all I was one of the small minority who advocated him as starter from day one last season. And I also liked Acy last season.
If you're looking for a perfect formula that will work without exceptions like small sample size, you'll be looking for a long time.

My goal was certainly not to create a catch-all stat - every attempt at that is very flawed. I was just trying to show another way to look at players' impacts - particularly in terms of how lineups and pairings play together.

And by the way, there certainly is room for minutes and opposition in that formula. I could muliply the per possession rating by total possessions played, which would give a raw net impact. That would be a good indicator of players who are both effective and used properly. Of course, that would work best in conjunction with the per-possession format, so that you can also evaluate players who are being underused (or overused).

As for the opposition stats, I'd love to include them, but I just don't know where to find the data. Especially in a lineup sortable format to use in conjunction with this stuff. I'd love to use it.

Now, one area where WOWY is really helpful is the idea that a player typically plays against the same level of competition - this is most true for players with a set role. For example, Gay, playing his starter's minutes, will usually match up with other starter level players. As such, Johnson's impact on Gay's WOWY ratings would be defined by the level of opposition Gay typically faces, rather than the level of opposition Johnson usually faces. It's not perfect, but it helps buffer against opposition effects much better than just looking at net on-off data as I have in the past.
DanH is offline   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 02:49 PM   #39 (permalink)
hopeful the Raps eliminate the Wiz from the post season

Senior Member
 
MBailey85's Avatar
 
Join Date: Feb 2012
Location: Toronto,Ontario
Posts: 2,445
Representing:
Default

Great work Dan,obviously! With that post,you are officially a legend...

DD averages the most minutes and plays ever game year in and out for a pretty bad or sub par team. Its gonna be easy for him to look like the weak link,just as some believe his stats are inflated because he plays on a poor team. Its a tricky thing. The most important numbers are the win column,with the core of this unit having more time together I think they'll be a bit better than last year.

Again,great stuff Dan...broke it down to a science!
MBailey85 is offline   Boss Key Wife Key Reply With Quote
Old 08-02-2013, 03:28 PM   #40 (permalink)
Is wondering when the Patman and JJ will be unleashed?

Membres supérieurs
 
DocHoliday99's Avatar
 
Join Date: Dec 2010
Location: Southern Alberta
Posts: 5,214
Representing:
Default

Quote:
Originally Posted by DanH View Post
As for the opposition stats, I'd love to include them, but I just don't know where to find the data. Especially in a lineup sortable format to use in conjunction with this stuff. I'd love to use it.
Question, since the site is by date, can't one just pull each individual's data on the day they play each other? Not the best approach but one could extrapolate a rudimentary projection, by position of how opposing players are effective against each other. Like comparing Bosh to KG when they play against each other. Again it would be rudimentary as there are switch offs and injuries what not but with some extra 'digging' those causalities to the projection could be reduced if not eliminated.
DocHoliday99 is online now   Boss Key Wife Key Reply With Quote
Reply


Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


All times are GMT -4. The time now is 12:04 PM.


Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2015, Jelsoft Enterprises Ltd.
Copyright RaptorsForum.com 2005-2011

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24