Jester1534 wrote:Idk if there is one but basketball really needs a WAR stat like baseball.
Its simple stat Wins above replacment. How much is that player worth if you were to replace him with league average player at that position. This takes account defense and offense.
This why Byron Buxton War is always postive because no matter how bad he hits. Hes always going to be in top 3 for defensive centerfielders.
So what is difference between starting T.J. McConnell who I looked at some website has him rank 25th out 50 point guards than DLO.
Feel like +/- stat has nothing to do with indiviual player. I do like the per 100 possesion because i believe thats closer to the WAR stat for baseball.
sorry for ramblin ill take my sabermetric baseball crap elsewhere lol
The NBA has win shares. It's an estimate of the number of wins contributed by a player. Here are the Wolves rankings:
#114 - Malik Beasley .8
#127 - Naz Reid .8
#133 - J. Vanderbilt .8
#181 - KAT .6
#188 - Ed Davis .5
#246 - J. McLaughlin .3
#251 - DLO .3
#273 - J. Hernangomez .2
#277 - J. Layman .2
#286 - J. Okogie .2
#292 - Ricky Rubio .2
#312 - J. Culver .1
#332 - J. Nowell .1
#382 - Ashton Hagans 0
#403 - J. McDaniels 0
#470 - Ant Edwards -.8 (worst in the league)
Idk if its similar to win shares in baseball. But the differnce in baseball win shares takes account if your team was good or bad as WAR it doesn't take account if your team sucks or not.
I think you're looking or VORP - Value Over Replacement Player.
Yep this is what im looking for. Pretty much the same thing funny how two sports can come to the same type of stats.
So
Wiggins who people are prasing as two way wiggs is .2 VORP
Russel whos been abysmal is .2 VORP
Someone from the outside I would put hell of a lot more value in this stat than +/-. When it comes to MLB WAR is everything be all stat and no very little disagree with that.
When I was with Texas my goal as scouter was to find late round pitcher with one to two good pitches. Because if we could get there other secondary stuff to league average then they would still be filithy with there one great pitch. Basically saying if we can make this guy league average at everything else and say his slider is + pitch than we can get 8-10 years out of this guy.
Coudnt you make the argument that wolves would be in excat same possesion this year? If they didnt trade for Russel based on the this stat? The only differnece is we would have our draft pick.
I can't speak as intelligently to the MLB, but a few observations:
- There is more flow and moving parts to an NBA game than the MLB. Soccer might be even tougher. The eye test tells me it's just a bit easier to get one's arms around One Statistic To Rule Them All in baseball versus basketball.
- As I have said continuously, gross +/- is a crude statistic with a ton of noise in it over short to medium sample sizes. I agree with that and I don't know anyone here claiming it to be THE best statistic to evaluate a player in any given game or week or month.
- PER (Player Efficiency Rating) might be the most individualized advanced stat out there. As far as I know, it looks almost purely at the counting stats in the box score. The problem is that great defense is almost never recorded in a box score (think close outs, smart positioning, hedging on a screen, deflections that aren't recorded as a block or steal, induced travels, etc.). It's half the game yet there is no good single metric for it so PER totally ignores a player's defense except for rebounding, steals, and blocks. And those aren't necessarily correlated to great defense.
- So the way to get at defense is to use On/Off and +/- data and then try to isolate the player being measured from his team mates. There are stats like D-RPM, D-RAPM, and D-PIPM that try to attempt to do this.
- Bottom line is that when you look at these advanced stats that try to isolate defense and then combine with offensive proficiency, DLO usually comes out as a slight positive most seasons. Wiggins is similar.
thedoper wrote:The problem with any stat based on + - is that players who play with great players also consistently do well in it. See all of the Bucks that were in the top 20 last year. That in itself means there's likely at least a 20% margin of error in how it ranks players relative to eachother (likely a lot higher). In any other statistical analysis outside of basketball that wouldnt be called noise, that would get you laughed out of the room.
If in a whole year if all of the players who played with Giannis are inflated it shouldnt be used as a primary individual stat and at best is noisy when applied that way. It should never be used in my opinion to negate a players clearly positive raw numbers. Its also highly situational. I think its clear that a good (not great) starter on a really bad team will consistently have a bad +-. DLo has only played on one team that was even good.
+ - and all the stats that use that metric as a foundation are best used as a measure of a players adaptability and complimentary characteristics within a particular lineup, not their contribution to winning or to the success of that lineup. I do think it has a place in this discussion. For instance DLo seems to be a tough player to find the right complimentary pieces for in a lineup. I think that statement would fit the statistical narrative that +- paints.
I dont think its a huge revelation that great players should have a higher +-, they do. But I dont believe you can in any way use +- to determine a player's greatness.
Doper, it's not just net rating that I look at (which is basically the difference in points per 100 possessions when the player is on the floor and the metric you are referring to above). It's also net On/Off rating, which shows the difference between what happens to team performance when the player is on vs. off the floor. So you are only looking at one aspect of +/-.
The other thing you need to look at is how does the player do versus other starters on the same team in the same metric? In other words, now you are comparing apples to apples.
So if I look at the last two full seasons for Giannis, yes, the guys that start with him get inflated +/- and net on/off ratings, but guess whose is better than the other starters? Giannis. Huh, I wonder why that is? May be because he's the best player on the team.
The statistical analysts you cite above also take relativity as a sound scientific concept - almost like grading on a curve.
So to summarize:
1. +/- and net On/Off should be looked at over long time periods/large sample sizes, because yes, it is noisier that purely individual stats like PER.
2. +/- should be looked at relative to other players on the same team in a similar situation (starter vs. bench).
3. Yes, it should be combined with individual performance metrics to form an overall view of a player.
Based on those three factors, I consider DLO to be a neutral at best player in terms of team impact that needs a really specific set of skill-sets around him to be a net positive in that area. I'd be fine with that if he were paid $10-15M per year vs. on a max deal.
Thanks for sharing your perspectives. I dont see how relativity, time, and a social construct of grading on a curve clears up bad data and am not sure you presented exactly what you may have meant as the mathematical reason for your inclusion of these terms? I do like the rules you are setting out for using this data though. It seems to follow a logical and guarded approach to using extremely noisy data for decision making and analysis.
I think you are one of the less reductionist people who like using this data in your positions and I find it really interesting and insightful. I also think that Basketball forcing people to look at how complex math is used in analyzing data is also really cool. That being said I do think there are a lot of people (not you from my experience) who start with an advanced stat to make an analysis of certain player, which is counterintuitive, and quite frankly not a representation of proper statistical analysis as I understand it.
thedoper wrote:The problem with any stat based on + - is that players who play with great players also consistently do well in it. See all of the Bucks that were in the top 20 last year. That in itself means there's likely at least a 20% margin of error in how it ranks players relative to eachother (likely a lot higher). In any other statistical analysis outside of basketball that wouldnt be called noise, that would get you laughed out of the room.
If in a whole year if all of the players who played with Giannis are inflated it shouldnt be used as a primary individual stat and at best is noisy when applied that way. It should never be used in my opinion to negate a players clearly positive raw numbers. Its also highly situational. I think its clear that a good (not great) starter on a really bad team will consistently have a bad +-. DLo has only played on one team that was even good.
+ - and all the stats that use that metric as a foundation are best used as a measure of a players adaptability and complimentary characteristics within a particular lineup, not their contribution to winning or to the success of that lineup. I do think it has a place in this discussion. For instance DLo seems to be a tough player to find the right complimentary pieces for in a lineup. I think that statement would fit the statistical narrative that +- paints.
I dont think its a huge revelation that great players should have a higher +-, they do. But I dont believe you can in any way use +- to determine a player's greatness.
Doper, it's not just net rating that I look at (which is basically the difference in points per 100 possessions when the player is on the floor and the metric you are referring to above). It's also net On/Off rating, which shows the difference between what happens to team performance when the player is on vs. off the floor. So you are only looking at one aspect of +/-.
The other thing you need to look at is how does the player do versus other starters on the same team in the same metric? In other words, now you are comparing apples to apples.
So if I look at the last two full seasons for Giannis, yes, the guys that start with him get inflated +/- and net on/off ratings, but guess whose is better than the other starters? Giannis. Huh, I wonder why that is? May be because he's the best player on the team.
The statistical analysts you cite above also take relativity as a sound scientific concept - almost like grading on a curve.
So to summarize:
1. +/- and net On/Off should be looked at over long time periods/large sample sizes, because yes, it is noisier that purely individual stats like PER.
2. +/- should be looked at relative to other players on the same team in a similar situation (starter vs. bench).
3. Yes, it should be combined with individual performance metrics to form an overall view of a player.
Based on those three factors, I consider DLO to be a neutral at best player in terms of team impact that needs a really specific set of skill-sets around him to be a net positive in that area. I'd be fine with that if he were paid $10-15M per year vs. on a max deal.
Thanks for sharing your perspectives. I dont see how relativity, time, and a social construct of grading on a curve clears up bad data and am not sure you presented exactly what you may have meant as the mathematical reason for your inclusion of these terms? I do like the rules you are setting out for using this data though. It seems to follow a logical and guarded approach to using extremely noisy data for decision making and analysis.
I think you are one of the less reductionist people who like using this data in your positions and I find it really interesting and insightful. I also think that Basketball forcing people to look at how complex math is used in analyzing data is also really cool. That being said I do think there are a lot of people (not you from my experience) who start with an advanced stat to make an analysis of certain player, which is counterintuitive, and quite frankly not a representation of proper statistical analysis as I understand it.[/quo
This is a very credible statement and one which I have discussed in the past. Advanced stats or any stats for that matter can't be your argument. They can be used to support your argument but they shouldn't BE the argument.
I don't think I'm any smarter than the average guy on this board, but I do have a background in statistics. Statistics are best used as part of an overall comprehensive analysis. I get a little put off when someone presents a metric and then draws conclusions from it. It really isn't a sound method of reasoning.
thedoper wrote:The problem with any stat based on + - is that players who play with great players also consistently do well in it. See all of the Bucks that were in the top 20 last year. That in itself means there's likely at least a 20% margin of error in how it ranks players relative to eachother (likely a lot higher). In any other statistical analysis outside of basketball that wouldnt be called noise, that would get you laughed out of the room.
If in a whole year if all of the players who played with Giannis are inflated it shouldnt be used as a primary individual stat and at best is noisy when applied that way. It should never be used in my opinion to negate a players clearly positive raw numbers. Its also highly situational. I think its clear that a good (not great) starter on a really bad team will consistently have a bad +-. DLo has only played on one team that was even good.
+ - and all the stats that use that metric as a foundation are best used as a measure of a players adaptability and complimentary characteristics within a particular lineup, not their contribution to winning or to the success of that lineup. I do think it has a place in this discussion. For instance DLo seems to be a tough player to find the right complimentary pieces for in a lineup. I think that statement would fit the statistical narrative that +- paints.
I dont think its a huge revelation that great players should have a higher +-, they do. But I dont believe you can in any way use +- to determine a player's greatness.
Doper, it's not just net rating that I look at (which is basically the difference in points per 100 possessions when the player is on the floor and the metric you are referring to above). It's also net On/Off rating, which shows the difference between what happens to team performance when the player is on vs. off the floor. So you are only looking at one aspect of +/-.
The other thing you need to look at is how does the player do versus other starters on the same team in the same metric? In other words, now you are comparing apples to apples.
So if I look at the last two full seasons for Giannis, yes, the guys that start with him get inflated +/- and net on/off ratings, but guess whose is better than the other starters? Giannis. Huh, I wonder why that is? May be because he's the best player on the team.
The statistical analysts you cite above also take relativity as a sound scientific concept - almost like grading on a curve.
So to summarize:
1. +/- and net On/Off should be looked at over long time periods/large sample sizes, because yes, it is noisier that purely individual stats like PER.
2. +/- should be looked at relative to other players on the same team in a similar situation (starter vs. bench).
3. Yes, it should be combined with individual performance metrics to form an overall view of a player.
Based on those three factors, I consider DLO to be a neutral at best player in terms of team impact that needs a really specific set of skill-sets around him to be a net positive in that area. I'd be fine with that if he were paid $10-15M per year vs. on a max deal.
Thanks for sharing your perspectives. I dont see how relativity, time, and a social construct of grading on a curve clears up bad data and am not sure you presented exactly what you may have meant as the mathematical reason for your inclusion of these terms? I do like the rules you are setting out for using this data though. It seems to follow a logical and guarded approach to using extremely noisy data for decision making and analysis.
I think you are one of the less reductionist people who like using this data in your positions and I find it really interesting and insightful. I also think that Basketball forcing people to look at how complex math is used in analyzing data is also really cool. That being said I do think there are a lot of people (not you from my experience) who start with an advanced stat to make an analysis of certain player, which is counterintuitive, and quite frankly not a representation of proper statistical analysis as I understand it.
I find it interesting and someone like DLO especially interesting because, similar to Zach LaVine, there is a disconnect between his individual advanced metrics (like PER and BPM) which seems to really firmly put DLO in the "above average
thedoper wrote:
Q12543 wrote:
thedoper wrote:The problem with any stat based on + - is that players who play with great players also consistently do well in it. See all of the Bucks that were in the top 20 last year. That in itself means there's likely at least a 20% margin of error in how it ranks players relative to eachother (likely a lot higher). In any other statistical analysis outside of basketball that wouldnt be called noise, that would get you laughed out of the room.
If in a whole year if all of the players who played with Giannis are inflated it shouldnt be used as a primary individual stat and at best is noisy when applied that way. It should never be used in my opinion to negate a players clearly positive raw numbers. Its also highly situational. I think its clear that a good (not great) starter on a really bad team will consistently have a bad +-. DLo has only played on one team that was even good.
+ - and all the stats that use that metric as a foundation are best used as a measure of a players adaptability and complimentary characteristics within a particular lineup, not their contribution to winning or to the success of that lineup. I do think it has a place in this discussion. For instance DLo seems to be a tough player to find the right complimentary pieces for in a lineup. I think that statement would fit the statistical narrative that +- paints.
I dont think its a huge revelation that great players should have a higher +-, they do. But I dont believe you can in any way use +- to determine a player's greatness.
Doper, it's not just net rating that I look at (which is basically the difference in points per 100 possessions when the player is on the floor and the metric you are referring to above). It's also net On/Off rating, which shows the difference between what happens to team performance when the player is on vs. off the floor. So you are only looking at one aspect of +/-.
The other thing you need to look at is how does the player do versus other starters on the same team in the same metric? In other words, now you are comparing apples to apples.
So if I look at the last two full seasons for Giannis, yes, the guys that start with him get inflated +/- and net on/off ratings, but guess whose is better than the other starters? Giannis. Huh, I wonder why that is? May be because he's the best player on the team.
The statistical analysts you cite above also take relativity as a sound scientific concept - almost like grading on a curve.
So to summarize:
1. +/- and net On/Off should be looked at over long time periods/large sample sizes, because yes, it is noisier that purely individual stats like PER.
2. +/- should be looked at relative to other players on the same team in a similar situation (starter vs. bench).
3. Yes, it should be combined with individual performance metrics to form an overall view of a player.
Based on those three factors, I consider DLO to be a neutral at best player in terms of team impact that needs a really specific set of skill-sets around him to be a net positive in that area. I'd be fine with that if he were paid $10-15M per year vs. on a max deal.
Thanks for sharing your perspectives. I dont see how relativity, time, and a social construct of grading on a curve clears up bad data and am not sure you presented exactly what you may have meant as the mathematical reason for your inclusion of these terms? I do like the rules you are setting out for using this data though. It seems to follow a logical and guarded approach to using extremely noisy data for decision making and analysis.
I think you are one of the less reductionist people who like using this data in your positions and I find it really interesting and insightful. I also think that Basketball forcing people to look at how complex math is used in analyzing data is also really cool. That being said I do think there are a lot of people (not you from my experience) who start with an advanced stat to make an analysis of certain player, which is counterintuitive, and quite frankly not a representation of proper statistical analysis as I understand it.
I find the DLO's of the NBA somewhat fascinating because they have the dual statistical narratives of highly skilled borderline all-star and lineup wrecker. Which do you believe more?
Then you have the Rubios of the NBA which, with the exception of this season, has a similar disconnect in narratives as DLO, but they are flipped. Mediocre individual player vs. lineup enhancer.
