View previous topic :: View next topic |
Author |
Message |
dirka dirka Franchise Player
Joined: 09 Jul 2005 Posts: 14655
|
Posted: Mon Apr 07, 2014 2:31 pm Post subject: |
|
|
Look at that superstar Beard down at 33. I like these stats.
_________________
jbjb wrote: | Echoes from the half empty glass. |
|
|
Back to top |
|
|
ribeye Franchise Player
Joined: 10 Nov 2001 Posts: 12630
|
Posted: Mon Apr 07, 2014 2:41 pm Post subject: |
|
|
Aeneas Hunter wrote: | When I played around with those stats, I noticed something interesting. Go to the chart, click on the column heading for DRPM, and look at the positions of the players at the top of the list. To the extent that this stat is valid, it illustrates something that a lot of us already believed -- even an average big man has a greater defensive impact than just about any perimeter player. |
There is a reason for this, as I actually inquired. Regressions make assumptions. In this case, there is an assumption that a taller man will be a better defender, since, they often are. So, regardless, the +/-, if a 6'2" guy had exactly the same +/- as a 6'10" guy, the 6' 10" guy will end up better. I don't know the weighing of it, so I don't know the extent of the height factor.
Last season, maybe two months or so in, Gasol had a much better DRPM than did Kobe. By 82games data, they played 90some% of the time together, so I couldn't figure why Gasol, who I thought at the time looked lethargic at best, and was run over like he was on the Washington Generals, could be any better than Kobe, when they were on the same unit nearly all the time. The answer, he was taller. _________________ "A metronome keeps time by using a Ringo" |
|
Back to top |
|
|
Omar Little Moderator
Joined: 02 May 2005 Posts: 90306 Location: Formerly Known As 24
|
Posted: Mon Apr 07, 2014 2:46 pm Post subject: |
|
|
I like advanced metrics, but one of the key dangers of aggregating stats (aside from which stats, how they are gathered, and in what proportion) is overaggregation. Simply put, when you try to homogenize things into one be-all-end-all stat, you lose the value of granularity. You also magnify built in bias (PER being an excellent example). I'd rather look at a wide range of more specific aggregations (+/-, PPP, TS%, etc.), and then rank them myself based on particular context. I can use PER against, for example, to get generalized information about man defense. I can compare and cross reference it to other ratings. But I don't just go with PER against as a good standard of defense, nor do I try to come up with one. I look at all of them in context. _________________ “We must always take sides. Neutrality helps the oppressor, never the victim. Silence encourages the tormentor, never the tormented.” ― Elie Wiesel
Last edited by Omar Little on Mon Apr 07, 2014 3:15 pm; edited 1 time in total |
|
Back to top |
|
|
Aeneas Hunter Retired Number
Joined: 12 Jul 2005 Posts: 31763
|
Posted: Mon Apr 07, 2014 3:07 pm Post subject: |
|
|
As I've said many times, different stats are just tools in a toolbox. None of them answer all questions. |
|
Back to top |
|
|
dirka dirka Franchise Player
Joined: 09 Jul 2005 Posts: 14655
|
Posted: Mon Apr 07, 2014 3:11 pm Post subject: |
|
|
Except PP100, that's definitive. _________________
jbjb wrote: | Echoes from the half empty glass. |
|
|
Back to top |
|
|
Omar Little Moderator
Joined: 02 May 2005 Posts: 90306 Location: Formerly Known As 24
|
Posted: Mon Apr 07, 2014 3:18 pm Post subject: |
|
|
PP100/PPP is actually a very good snapshot of how efficient a team or player is, especially if it can be narrowed to specific things, like analyzing a team's efficiency in one set or action vs another. _________________ “We must always take sides. Neutrality helps the oppressor, never the victim. Silence encourages the tormentor, never the tormented.” ― Elie Wiesel |
|
Back to top |
|
|
Aeneas Hunter Retired Number
Joined: 12 Jul 2005 Posts: 31763
|
Posted: Tue Apr 08, 2014 6:50 am Post subject: |
|
|
ribeye wrote: | There is a reason for this, as I actually inquired. Regressions make assumptions. In this case, there is an assumption that a taller man will be a better defender, since, they often are. So, regardless, the +/-, if a 6'2" guy had exactly the same +/- as a 6'10" guy, the 6' 10" guy will end up better. |
I have no idea what you're talking about. How does a regression make an assumption? If you are saying that this stat contains hidden internal assumptions that are designed to help it pass the eyeball test, then it's a junk stat that deserves to get tossed on the garbage heap.
Anyway, this isn't APM, though it appears to be derived from it in some respect. I recall looking at APM a year or two ago and concluding that it was garbage. It was a way of saying, "I know that +/- stats suck, but I'll apply some fancy math to make them sound more legitimate." I can't discern enough about the methodology of RPM to draw any hard conclusions one way or the other. |
|
Back to top |
|
|
GoldenThroat Moderator
Joined: 25 Jul 2005 Posts: 37474
|
Posted: Tue Apr 08, 2014 8:37 am Post subject: |
|
|
24 wrote: | PP100/PPP is actually a very good snapshot of how efficient a team or player is, especially if it can be narrowed to specific things, like analyzing a team's efficiency in one set or action vs another. |
Yep. The score of the game (you know, the thing that determines wins and losses) is based on PPP. |
|
Back to top |
|
|
GoldenThroat Moderator
Joined: 25 Jul 2005 Posts: 37474
|
Posted: Tue Apr 08, 2014 8:38 am Post subject: |
|
|
24 wrote: | I like advanced metrics, but one of the key dangers of aggregating stats (aside from which stats, how they are gathered, and in what proportion) is overaggregation. Simply put, when you try to homogenize things into one be-all-end-all stat, you lose the value of granularity. You also magnify built in bias (PER being an excellent example). I'd rather look at a wide range of more specific aggregations (+/-, PPP, TS%, etc.), and then rank them myself based on particular context. I can use PER against, for example, to get generalized information about man defense. I can compare and cross reference it to other ratings. But I don't just go with PER against as a good standard of defense, nor do I try to come up with one. I look at all of them in context. |
Agreed on this as well. |
|
Back to top |
|
|
Omar Little Moderator
Joined: 02 May 2005 Posts: 90306 Location: Formerly Known As 24
|
Posted: Tue Apr 08, 2014 10:03 am Post subject: |
|
|
GoldenThroat wrote: | 24 wrote: | PP100/PPP is actually a very good snapshot of how efficient a team or player is, especially if it can be narrowed to specific things, like analyzing a team's efficiency in one set or action vs another. |
Yep. The score of the game (you know, the thing that determines wins and losses) is based on PPP. |
That said, I'm leery of using it too heavily for judging players overall, because it doesn't factor in making good passes to any real degree, and thus has a bias toward scorers. But it does tell us how efficient as a scorer they are, and if we further divide it up among types of actions (P&R, post up, ISO, spot up, etc), we can see where a player's strengths and weaknesses lie.
For teams, it is a great great tool, especially when combined with opponent PPP stats. _________________ “We must always take sides. Neutrality helps the oppressor, never the victim. Silence encourages the tormentor, never the tormented.” ― Elie Wiesel |
|
Back to top |
|
|
kikanga Retired Number
Joined: 15 Sep 2012 Posts: 29282 Location: La La Land
|
Posted: Wed Apr 09, 2014 1:53 am Post subject: |
|
|
USCandLakers wrote: | Aeneas Hunter wrote: | golaker wrote: | Why do they even bother. Andre Igoudala is 188 spots ahead of Kyrie Irving? Obviously Kyrie has a more significant impact on his teams output when he's on/off the floor, yet this statistic shows him being a d-league player compared to Iggy.
They need to stop with the stats. It's always going to be about the eye-ball test. |
The value of stats is that they can sometimes tell you that your eyeballs are deceiving you. In the case of you and Kyrie Irving, this appears to be the case. |
Uhh yeah, no. It's the other way around. I don't see how it could ever be that way. |
Exactly! Let me put this in the simplest terms possible.
The "eye-ball" test is the source. Advanced statistics are the results.
These advanced stats were created because the old stats (+/-, ORPG, DRPG) weren't detailed or accurate enough to evaluate player worth independently. Even 50 years from now, it is unlikely any stats will ever be able to evaluate more than great basketball minds watching games (the eyeball test)
Let's name some of the flaws in the current gen stats:
Not every bucket is worth the same in a game. Same goes for defense. A key bucket or defensive stop when the other team is on a run matters more than when you are up 5-10, or 10-15, etc. Also, since stats are always averages, matchup variation gets diluted. Jordan hill will put up better stats when he is matched up against another non 7 footer. But the opposite can be said about Kaman. Current gen stats won't tell you that. It will just show you how they averaged out through the year. What about home vs away games. And differing game times (afternoon vs night). Nationally televised vs sydicated games. Player personalities (ie, kobe/dwight). Stats on players playing in and against man to man vs zone. Way more variables exist. And don't forget you have to figure out how to properly weigh each variable at the end. Then you get "the eyeball test" at the level of Phil or Wooden. |
|
Back to top |
|
|
Aeneas Hunter Retired Number
Joined: 12 Jul 2005 Posts: 31763
|
Posted: Wed Apr 09, 2014 5:07 am Post subject: |
|
|
kikanga wrote: | Exactly! Let me put this in the simplest terms possible.
The "eye-ball" test is the source. Advanced statistics are the results. |
You're going to need to put it more simply than that, because that doesn't make any sense. |
|
Back to top |
|
|
kikanga Retired Number
Joined: 15 Sep 2012 Posts: 29282 Location: La La Land
|
Posted: Wed Apr 09, 2014 10:01 am Post subject: |
|
|
Aeneas Hunter wrote: | kikanga wrote: | Exactly! Let me put this in the simplest terms possible.
The "eye-ball" test is the source. Advanced statistics are the results. |
You're going to need to put it more simply than that, because that doesn't make any sense. |
Lol. You're funny |
|
Back to top |
|
|
Aeneas Hunter Retired Number
Joined: 12 Jul 2005 Posts: 31763
|
Posted: Wed Apr 09, 2014 10:43 am Post subject: |
|
|
I was being dead serious. I don't see how advanced statistics are the result of the eyeball test. For that matter, I don't see how the "eyeball test" of coaches like PJ and Wooden has much of anything to do with what we're discussing. But whatever rocks your boat. |
|
Back to top |
|
|
kikanga Retired Number
Joined: 15 Sep 2012 Posts: 29282 Location: La La Land
|
Posted: Wed Apr 09, 2014 12:03 pm Post subject: |
|
|
Aeneas Hunter wrote: | I was being dead serious. I don't see how advanced statistics are the result of the eyeball test. For that matter, I don't see how the "eyeball test" of coaches like PJ and Wooden has much of anything to do with what we're discussing. But whatever rocks your boat. |
Still laughing |
|
Back to top |
|
|
Omar Little Moderator
Joined: 02 May 2005 Posts: 90306 Location: Formerly Known As 24
|
Posted: Wed Apr 09, 2014 1:34 pm Post subject: |
|
|
kikanga wrote: | Aeneas Hunter wrote: | I was being dead serious. I don't see how advanced statistics are the result of the eyeball test. For that matter, I don't see how the "eyeball test" of coaches like PJ and Wooden has much of anything to do with what we're discussing. But whatever rocks your boat. |
Still laughing |
And still wrong. _________________ “We must always take sides. Neutrality helps the oppressor, never the victim. Silence encourages the tormentor, never the tormented.” ― Elie Wiesel |
|
Back to top |
|
|
Theseus Franchise Player
Joined: 15 Dec 2007 Posts: 14166
|
Posted: Wed Apr 09, 2014 1:54 pm Post subject: |
|
|
I can't wait for 100% accurate i mean it this time true shooting to come out, then we can have further insight into the greatness of patrick beverly and nick collison |
|
Back to top |
|
|
The_Dynasty24 Star Player
Joined: 12 Jun 2013 Posts: 2840
|
Posted: Wed Apr 09, 2014 3:40 pm Post subject: |
|
|
I think people feel the need to bash advanced stats because they don't necessarily agree with what that person thinks they are seeing in the game. People act like stats such as this are arguing Nick Collison is an elite player in the game, when in actuality its only ONE factor that can be used to evaluate players.
The "eye-ball" test can let you form an opinion on a player, and advanced stats should be a great tool to augment the "data" your eye-ball test has provided. |
|
Back to top |
|
|
Aeneas Hunter Retired Number
Joined: 12 Jul 2005 Posts: 31763
|
Posted: Wed Apr 09, 2014 4:52 pm Post subject: |
|
|
The_Dynasty24 wrote: | I think people feel the need to bash advanced stats because they don't necessarily agree with what that person thinks they are seeing in the game. People act like stats such as this are arguing Nick Collison is an elite player in the game, when in actuality its only ONE factor that can be used to evaluate players.
The "eye-ball" test can let you form an opinion on a player, and advanced stats should be a great tool to augment the "data" your eye-ball test has provided. |
Sure. One of the tricks with +/- stats in general is that you must be careful to make sure that you are looking at comparable players. Collison plays about 16 minutes per game and does not start. The +/- stat tells you that he is having a terrific impact in that role. It does not tell you that he would be a top 10 player as a starter playing 35 minutes. It also does not tell you that his skills are in the top 10. |
|
Back to top |
|
|
ribeye Franchise Player
Joined: 10 Nov 2001 Posts: 12630
|
Posted: Thu Apr 10, 2014 11:49 am Post subject: |
|
|
The +/- or on/off stats are useful but have two important limitations. One is that basic +/- factors all five players collectively, so that when we compare (82games) a guy like Ricky Rubio, with a +15.3 to Kyrie Irving who is a -.3, we scratch our heads. It is helpful to tell how well a team's starters do in comparison to the bench, but the value beyond is limited.
I don't think there is a soul alive who thinks Rubio is that much better, and, probably not better at all, according to most, than Irving. But Rubio's unit is far better because of teammates. Irving doesn't have Kevin Love.
The other problem has to do with advanced +/-. Here, data is important and when there are so many combinations of players, it takes a whole season to get an idea, but really, several to have enough to get beyond such a limited sample size, to see how well any player does.
If Kobe and Pau play together the vast amount of time, they will have limited minutes with other combinations where they play apart, so it takes years to assemble enough data, to get an idea of a singles player's unique contribution to +/- _________________ "A metronome keeps time by using a Ringo" |
|
Back to top |
|
|
Aeneas Hunter Retired Number
Joined: 12 Jul 2005 Posts: 31763
|
Posted: Thu Apr 10, 2014 11:53 am Post subject: |
|
|
ribeye wrote: | I don't think there is a soul alive who thinks Rubio is that much better, and, probably not better at all, according to most, than Irving. |
Actually, I suspect that there are a fair number of people who think just that. |
|
Back to top |
|
|
ribeye Franchise Player
Joined: 10 Nov 2001 Posts: 12630
|
Posted: Thu Apr 10, 2014 12:53 pm Post subject: |
|
|
Aeneas Hunter wrote: | ribeye wrote: | I don't think there is a soul alive who thinks Rubio is that much better, and, probably not better at all, according to most, than Irving. |
Actually, I suspect that there are a fair number of people who think just that. |
Can't a fair number be that which isn't most? Do you think that most feel Rubio is the better player? To be honest, it seems close, though I haven't watched either enough to have a definitive position either way. Regardless, from what I hear, I get a sense that there are many who still have extremely high regard for Irving, enough to favor him. _________________ "A metronome keeps time by using a Ringo" |
|
Back to top |
|
|
Chronicle Retired Number
Joined: 21 Jul 2012 Posts: 31935 Location: Manhattan
|
Posted: Thu Apr 10, 2014 12:56 pm Post subject: |
|
|
They both stink _________________ Kobe |
|
Back to top |
|
|
Aeneas Hunter Retired Number
Joined: 12 Jul 2005 Posts: 31763
|
Posted: Thu Apr 10, 2014 1:15 pm Post subject: |
|
|
ribeye wrote: | Aeneas Hunter wrote: | ribeye wrote: | I don't think there is a soul alive who thinks Rubio is that much better, and, probably not better at all, according to most, than Irving. |
Actually, I suspect that there are a fair number of people who think just that. |
Can't a fair number be that which isn't most? Do you think that most feel Rubio is the better player? To be honest, it seems close, though I haven't watched either enough to have a definitive position either way. Regardless, from what I hear, I get a sense that there are many who still have extremely high regard for Irving, enough to favor him. | .
I was referring to the "soul alive" part. |
|
Back to top |
|
|
GoldenThroat Moderator
Joined: 25 Jul 2005 Posts: 37474
|
Posted: Fri Apr 11, 2014 8:36 am Post subject: |
|
|
ribeye wrote: | The +/- or on/off stats are useful but have two important limitations. One is that basic +/- factors all five players collectively, so that when we compare (82games) a guy like Ricky Rubio, with a +15.3 to Kyrie Irving who is a -.3, we scratch our heads. It is helpful to tell how well a team's starters do in comparison to the bench, but the value beyond is limited.
I don't think there is a soul alive who thinks Rubio is that much better, and, probably not better at all, according to most, than Irving. But Rubio's unit is far better because of teammates. Irving doesn't have Kevin Love.
The other problem has to do with advanced +/-. Here, data is important and when there are so many combinations of players, it takes a whole season to get an idea, but really, several to have enough to get beyond such a limited sample size, to see how well any player does.
If Kobe and Pau play together the vast amount of time, they will have limited minutes with other combinations where they play apart, so it takes years to assemble enough data, to get an idea of a singles player's unique contribution to +/- |
The intention of this statistic is to isolate individual +/- rather than it just being a result of a 5 man unit. We don't know what their methodology is, but their stated intent is to address these limitations. |
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
|