Advanced metrics guys!

[AHF]

12 hours ago, Peoriabird said:

So the 9 minutes W. Mathews played and the 2 minutes G. Mathews played against Orlando lost the game even though W. Mathews was +11 and G Mathews was -1 vs the starters who all were double digit - plus/minus.  This is amazing

If you want an actual conversation on this topic I'm here.  If you want to play games like this I'm not.  

Bottomline:  The team is worse without Hunter because we are employing existing players in roles to which they are not as well suited and we are playing guys who stink.  

13 hours ago, thecampster said:

You mean Wesley Matthew's right? Meth Bogi has had some quality moments.

I mean both.  Neither has been any good and if we are playing Meth Bogi that is really bad news for the team.  Not saying that he can't have a few good nights but just doesn't do enough on both sides of the ball to warrant those minutes.  It is a desperation play to cover for our injured starter.

[Peoriabird]

5 hours ago, AHF said:

If you want an actual conversation on this topic I'm here.  If you want to play games like this I'm not.  

Bottomline:  The team is worse without Hunter because we are employing existing players in roles to which they are not as well suited and we are playing guys who stink.  

I mean both.  Neither has been any good and if we are playing Meth Bogi that is really bad news for the team.  Not saying that he can't have a few good nights but just doesn't do enough on both sides of the ball to warrant those minutes.  It is a desperation play to cover for our injured starter.

The Hawks are not losing because W Mathews and G. Mathews play 5 minutes per game. The bad plus minuses in most games lately lie with the starters.

[TheNorthCydeRises]

6 minutes ago, Peoriabird said:

The Hawks are not losing because of W Mathews and G. Mathews play 5 minutes per game. The bad plus minuses in most games lately lie with the starters.

The Trae - DJ - Bey - JJ - Capela lineup has been very bad.

You are correct on this. 

[Peoriabird]

7 minutes ago, TheNorthCydeRises said:

The Trae - DJ - Bey - JJ - Capela lineup has been very bad.

You are correct on this. 

Thank you!!! So let's apply our advanced metrics and sort this out instead of playing games like @AHF wants to do.

[AHF]

2 minutes ago, Peoriabird said:

The Hawks are not losing because of W Mathews and G. Mathews play 5 minutes per game. The bad plus minuses in most games lately lie with the starters.

This is why I have a hard time we these discussions.  I post comprehensive numbers and  you don't bother to read those posts.  Then you come here and pretend like last game was a normal game for minutes played by our bad players.  But let's assume you are coming at this in good faith and get into the numbers.

Our 9-17 players have been absolute garbage this year giving us a net 0.0 Win Shares and many with negative BPM and other metrics. 

We are better when they don't play and the minutes go instead to people like Hunter or JJ.  But when Hunter misses time they play more and it hurts the team. 

So a key question is sow much  do they play when Hunter is out?  Is it the 5 minutes per game you suggest in your post above?

No.  These 9-17 players play an average of 32.4 mpg when Hunter is out.  That is a 54% increase in minutes to our worst players when Hunter is out versus games when Hunter plays.  

Do you think replacing a big chunk of Hunter's 0.057 WS/48 minutes with a big old 0 WS/48 is better?  Of course not.  Is it a good thing when you ask Bey to be your primary wing defender because the team's primary wing defender is out?  Of course not.

That is why the premise of this thread misses the point entirely.  The advanced metrics predict that when Hunter is out and you increase minutes to these players that the team will be worse.  And low and behold we are worse when he misses time and guys get pushed into roles that are bad for them (ala Bey) and you end up playing the 9-17 guys more minutes.

 

[Gray Mule]

I know nothing.  Let me guess.

If our starters play against our opponents starters, then the (+) (-) means how well the five players on the floor did.  When we send in one of the bench players to give a starter a breather, his (+) (-) would depend on who is on the floor with him, both on the four Hawks and the five opponents.

Therefore, most bench players have a lot less time played in each game and it's much harder to figure their stats because you just have to look very closely to discover who the other nine players were who played the same time that he did.  

I was kinda confused when I started this post.  Now I am completely up the creek without a paddle.  Help!

[Peoriabird]

Last 6 games and i picked these games because JJ played in these games

Team                                   +/- G. Mathews                               +/- Trae Young

Bulls                                             -1                                                     -3

Kings                                           +8                                                    -11

Wizards                                       +4                                                    -13

OKC                                             -8                                                     -9

Pacers                                         -9                                                     -28

Magic                                          -1                                                      -10

Note, I could have picked any starter and the result would have been similar so do we really want to blame G. Mathews for our recent team performances?   Let have a serious discussion to figure out what is wrong with the numbers if they indeed are favorable to our main starters

 

[AHF]

On 1/9/2024 at 5:12 PM, Peoriabird said:

Last 6 games and i picked these games because JJ played in these games

Team                                   +/- G. Mathews                               +/- Trae Young

Bulls                                             -1                                                     -3

Kings                                           +8                                                    -11

Wizards                                       +4                                                    -13

OKC                                             -8                                                     -9

Pacers                                         -9                                                     -28

Magic                                          -1                                                      -10

Note, I could have picked any starter and the result would have been similar so do we really want to blame G. Mathews for our recent team performances?   Let have a serious discussion to figure out what is wrong with the numbers if they indeed are favorable to our main starters

 

I guess we are talking about a different topic right now.  Advanced metrics designed to measure player performance are not the same as +/-.  +/- is a highly contextual number that is dependent on a lot of factors outside of player performance.  

For example, Trae had one of his worst games in a while yesterday but had the best +/- of any of our starters at a nice +7.  +/- can't be used as a metric for how well that player performs especially when you are talking about small data sets like a 6 game set or god forbid a single game.  Way too much noise in +/- to just look at how plus or minus it is and conclude that the player is effective or ineffective in those minutes.

This doesn't apply as much to lineup +/- as below.

On 1/9/2024 at 3:02 PM, TheNorthCydeRises said:

The Trae - DJ - Bey - JJ - Capela lineup has been very bad.

You are correct on this. 

The question of why this lineup has struggled is an interesting topic (although it has nothing to do with the validity of advanced metrics).  My personal take is that OO does a ton to help cover for Bey's defensive liabilities.  Bey does not do as well in lineups with Capela.  When JJ is on the floor this is exacerbated because while he is a much better defender than Bey he still has a lot to learn and gets lost on the floor.  Bey and JJ's respective bad and spotty defense paired with CC's lower mobility is a rough pairing and this gets worse when deployed against an opponent's starters versus against a lineup more heavily weighted towards reserves.  As we've talked about, Trae and DJ don't synergize on defense and are collectively too small so you aren't going to make up ground there.

Hunter is missed with Bey becoming the primary wing defender in this lineup when he is not equipped for this role.  Ideally, you don't want to deploy him with that responsibility in any lineup but it isn't like it is some panacea if you put JJ on the wing because JJ has even less experience guarding lead wings than he does opposing 4s and Bey struggles anywhere he plays on defense.

If you are forced by Hunter's lack of availability into running that lineup you really have to hope that Bey can deliver offensively to offset some of the defensive problems but he has been in a deep slump over the last month.  While I think this is more of a cold spell for him and would positively regress over a larger period of time, over the last month (since 12/8) he is shooting:

36.7% FG%

27.4% 3pt% 

That is brutal when he is primarily there to make shots.  

The problem is that the alternative to that lineup if you are going to keep starting JJ and CC is putting in someone who is an even worse fit in the starting lineup like Wesley Matthews, Garrison Mathews, etc.  Until Hunter comes back or a trade happens, I think we will struggle to find a good combination in a starting lineup with JJ and CC.  It would help if Bey would recover his normal shooting on corner 3's.  This year is an extreme outlier so far:

[Peoriabird]

11 hours ago, AHF said:

I guess we are talking about a different topic right now.  Advanced metrics designed to measure player performance are not the same as +/-.  +/- is a highly contextual number that is dependent on a lot of factors outside of player performance.  

For example, Trae had one of his worst games in a while yesterday but had the best +/- of any of our starters at a nice +7.  +/- can't be used as a metric for how well that player performs especially when you are talking about small data sets like a 6 game set or god forbid a single game.  Way too much noise in +/- to just look at how plus or minus it is and conclude that the player is effective or ineffective in those minutes.

This doesn't apply as much to lineup +/- as below.

The question of why this lineup has struggled is an interesting topic (although it has nothing to do with the validity of advanced metrics).  My personal take is that OO does a ton to help cover for Bey's defensive liabilities.  Bey does not do as well in lineups with Capela.  When JJ is on the floor this is exacerbated because while he is a much better defender than Bey he still has a lot to learn and gets lost on the floor.  Bey and JJ's respective bad and spotty defense paired with CC's lower mobility is a rough pairing and this gets worse when deployed against an opponent's starters versus against a lineup more heavily weighted towards reserves.  As we've talked about, Trae and DJ don't synergize on defense and are collectively too small so you aren't going to make up ground there.

 

So if your advance metrics don't account for any of what you just mention then how do they help you determine whether a line up is a winning line up then?..I'm totally confused now.  Obviously you can't just throw a bunch of players together with good advanced metrics and expect that line up to automatically perform well together.  So where is that measure?  That's should be the important metric..Not the ones that don't tell you how well the team plays together or give you some idea what the best combination of players will be.

[TheNorthCydeRises]

The problem is that the alternative to that lineup if you are going to keep starting JJ and CC is putting in someone who is an even worse fit in the starting lineup like Wesley Matthews, Garrison Mathews, etc.  Until Hunter comes back or a trade happens, I think we will struggle to find a good combination in a starting lineup with JJ and CC.  It would help if Bey would recover his normal shooting on corner 3's.  This year is an extreme outlier so far:

 

 

Or just go with a big lineup of Capela - OO - JJ . . . and hope that JJ can keep knocking down 3s at a much higher rate that Bey has done, while putting by far your best defensive frontline in the game.  I still think JJ is more than capable of being a facilitating big in the front court, that could feed the ball to both the guards and the other frontline players.

Give Bruno spot minutes at the 5 ( 10 - 16 minutes a game ).  And restrict Bey's minutes to playing the 3 at all times.  And when 12 comes back, Bey's minutes may have to be reduced even more ( or play the guy who is playing best between 12 and Bey ).

 

Edited January 12 by TheNorthCydeRises

[AHF]

13 hours ago, Peoriabird said:

So if your advance metrics don't account for any of what you just mention then how do they help you determine whether a line up is a winning line up then?..I'm totally confused now.  Obviously you can't just throw a bunch of players together with good advanced metrics and expect that line up to automatically perform well together.  So where is that measure?  That's should be the important metric..Not the ones that don't tell you how well the team plays together or give you some idea what the best combination of players will be.

Putting players together that had great advanced metrics is basically how the Dream Team was formed so it could work just fine if you had the option.  In practice that isn't realistic, so nobody should look at any one thing in isolation but you'd do just fine with a lineup of advanced metrics only guys if you could use that.  Here is your rotation with WS/48 leaders by position:

Shai

White

Luka

Giannis

Jokic

 

Haliburton

Mitchell

Kawhi

Lauri

Embiid

 

Those would do pretty darn well in any competition.  You are even getting solid combinations of different skills on both sides of the ball with those particular lineups.

+/- is not an advanced metric that tries to measure individual performance so it can be used but not for that purpose.  As applied to lineups, it gives a simple measure of whether those linesups are outperforming their opposition which can be helpful.  If a lineup is struggling compared to other lineups from the same team, it raises obvious questions about why that is the case.  Is it how that lineup is deployed? Is it the fit of the various players in that lineup?  Beyond +/-, you can obviously look at a wide variety of numbers to anticipate how people might fit together.  For example, if you have two players with super high usage rates in the backcourt together you can anticipate problems since both guys can't have the ball in their hands at the same time.  

I would say that coaches can use advanced metrics to better understand players but no good coach determines their lineups solely by relying on numbers only.  Likewise, no good coach ignores those numbers.  You mesh the a wide variety of data with observations about how players and combinations of players function within the systems and situations on the floor.

But if you are going to use numbers only to do so you can probably do a good job with an offensive lineup without too much trouble but I'm really skeptical that you could do so defensively because defensive metrics just aren't nearly as good as the offensive ones.  You can get from the numbers a pretty good idea of how good of a shooter, passer, rebounder, etc. that a player is but many of the defensive metrics just aren't very good at telling you something important about the individual's defense (example:  a player's defensive rating is more about your team than any individual's performance and so comparing those numbers across teams is almost worthless and even comparing within a team is pretty limited without broader context).

[Peoriabird]

4 hours ago, AHF said:

 I'm really skeptical that you could do so defensively because defensive metrics just aren't nearly as good as the offensive ones.  You can get from the numbers a pretty good idea of how good of a shooter, passer, rebounder, etc. that a player is but many of the defensive metrics just aren't very good at telling you something important about the individual's defense (example:  a player's defensive rating is more about your team than any individual's performance and so comparing those numbers across teams is almost worthless and even comparing within a team is pretty limited without broader context).

Its about time you realized this now go tell you Advanced metrics buddies this so that unnecessary heated debates about a certain Hawk players can finally die.

[Final_quest]

I’ve never seen anyone argue that there are a ton of guys with bad advanced stats every year like Hunter.  Do they exist?  

[Atlantaholic]

 

1 hour ago, Final_quest said:

I’ve never seen anyone argue that there are a ton of guys with bad advanced stats every year like Hunter.  Do they exist?  

Advanced Stats darlings are the ones AHT outlined above.

Advanced Stats duds are the below team. Picking just the starting players with the worst advanced stats at each position: Going to base this off of VORP (value added over replacement level player). Setting minimum minutes at 700.

C - Deandrea Ayton (-0.1)

PF - Jeremy Sochan (-0.7)

PG - Scoot Henderson (-1.2)

SF - Andrew Wiggins (-0.9)   - (For reference Hunter is at -0.1 this year, which is 8th worst for players with over 700 mins.)

SG - Jordan Poole -(1.0)

 

In '23 and '22 the last two seasons Hunter has ranked exactly 2nd worst in the NBA in VORP at his position. The only year he had a positive VORP was his 20 game sophomore season.

 

 

 

[AHF]

2 hours ago, Peoriabird said:

Its about time you realized this now go tell you Advanced metrics buddies this so that unnecessary heated debates about a certain Hawk players can finally die.

I have no idea what you are talking about.  The limitations of defensive metrics are well known and have been discussed around here numerous times.  But the limitations of things like DRTG don't make advanced metrics any less meaningful or just wash away Hunter's subpar numbers.  What they mean to me is that you need to mentally adjust Hunter up since his best defensive attributes aren't fully captured by defensive focused metrics .  I think you need to look harder at things like how he impacts lineup performance when he plays and take his numbers with that context in mind.  

There are plenty of good defenders who don't have any problems putting up amazing metrics.  Look at Mikal Bridges who is another wing but who is both a better defender and a standout for his overall production and efficiency.

This gap is not because of the limits of defensive metrics.

[Peoriabird]

2 hours ago, AHF said:

I have no idea what you are talking about.  The limitations of defensive metrics are well known and have been discussed around here numerous times.  But the limitations of things like DRTG don't make advanced metrics any less meaningful or just wash away Hunter's subpar numbers.  What they mean to me is that you need to mentally adjust Hunter up since his best defensive attributes aren't fully captured by defensive focused metrics .  I think you need to look harder at things like how he impacts lineup performance when he plays and take his numbers with that context in mind.  

There are plenty of good defenders who don't have any problems putting up amazing metrics.  Look at Mikal Bridges who is another wing but who is both a better defender and a standout for his overall production and efficiency.

This gap is not because of the limits of defensive metrics.

You literally just proved my point!!!  Bridges with all of his great metrics are not converting to Nets wins.  The Nets are 16-22.  I know you are going to find blame with the rest of the players to protect your precious metrics.  But I'm willing to listen

[Peoriabird]

Can't wait to here how the Advanced Metrics boys explain the blow out loss against the Wizards at home.

[AHF]

The idea that Mikal Bridges career to date hasn’t taken a **** on Hunters is crazy.  He is a much, much better player.   Do you really think that Hunter has the better career than Bridges?!?  Even Supes will tell you that this is a no-brainer so rather than proving your point that advanced metrics are meaningless they are actually reflecting a reality that one player is significantly better than the other.

As for the second post, the idea that advanced metrics wouldn’t reflect how craptastic the team played tonight is laughable.  For example, nearly every player’s TS%, WS, PER, etc if generated solely for this game would be much worse than the season numbers.  If the Hawks players played like that all season their metrics would all be abysmal.  Metrics reflect the performance on the floor - conversely, the metrics would be amazing if they all played their best games all season.  Instead the metrics are between those extremes because typically they play better than tonight but not as good as their best games.  The metrics are reflections of the full season to date (at least as typically used).  Metrics are reflections of performance and aren’t really “predictive” except to the extent that you assume players will maintain their prior production.  (Which is to say variability in a given game is huge and that it is expected that players go through ups and downs.  This is why a meaningful sample size is so important.)

I’m not discussing this topic on this thread anymore until you do some research on the basics because what you say you expect the metrics to show and what you state that they are actually showing do not match up with what they actually do and say.

[Peoriabird]

21 hours ago, AHF said:

The idea that Mikal Bridges career to date hasn’t taken a **** on Hunters is crazy.  He is a much, much better player.   Do you really think that Hunter has the better career than Bridges?!?  Even Supes will tell you that this is a no-brainer so rather than proving your point that advanced metrics are meaningless they are actually reflecting a reality that one player is significantly better than the other.

 

You keep talking about comparisons...I keep talking about wins... something you go out of your way not to talk about.  Your metrics don't convert to wins periods.  Who cares who think what player is better...Where are his wins????

[AHF]

2 hours ago, Peoriabird said:

You keep talking about comparisons...I keep talking about wins something you go out of your way not to talk about.  You metrics don't convert to wins periods.  Who cares who think what player is better...Where are his wins????

This is like complaining that a bad player on a losing team has an outstanding ft%.  Obviously team performance correlates with team metrics.   Individual players drive their own metrics and you can have great players on bad teams and bad players on great teams.  
 

Not shockingly, nobody on the Hawks is having a standout year by their metrics except for JJ.  Most everyone else is down relative to prior strong seasons.  So surprise the team is also having a down year.  Go figure.

All the numbers you’ve brought up validate the usefulness of advanced metrics.  And of course one way of validating their usefulness is whether better players have better metrics.  But NBA teams validated this stuff decades ago and have leaned harder and harder into the numbers over time so why are we even talking about it as if that is an open question?  The numbers aren’t the be all end all.  But no team worth its salt doesn’t pay close attention to them.