PDA

View Full Version : Empty "Advanced stats"?



valade16
10-04-2016, 06:23 PM
It was mentioned in another thread how Rondo's statistical accomplishments in Boston were actually to an extent at the expense of the team's wins and that he often took possessions from Allen and KG.

This got me thinking. What players would you say put up good or very good advanced stats (such as PER, VORP, BPM, WS, etc.) where those stats were to an extent hollow or the player didn't really impact the game to the degree those measures indicated?

If possible do one currently (for reference) and an all-timer.

NOTE: I DON'T want examples of a guy who scored a ton under traditional stats like PPG who wasn't impactful, but the guys where advanced stats say they were very impactful that might otherwise not have been.

nycericanguy
10-04-2016, 06:41 PM
Dmc?

Vee-Rex
10-04-2016, 06:54 PM
Dmc?

He's more of a traditional stat booster.

To OP: I think this is a fun thread. I'll take a look at some players and get back at you.

BoSox47
10-04-2016, 07:08 PM
It was mentioned in another thread how Rondo's statistical accomplishments in Boston were actually to an extent at the expense of the team's wins and that he often took possessions from Allen and KG.

This got me thinking. What players would you say put up good or very good advanced stats (such as PER, VORP, BPM, WS, etc.) where those stats were to an extent hollow or the player didn't really impact the game to the degree those measures indicated?

If possible do one currently (for reference) and an all-timer.

NOTE: I DON'T want examples of a guy who scored a ton under traditional stats like PPG who wasn't impactful, but the guys where advanced stats say they were very impactful that might otherwise not have been.

Early rondo was very impactful on games with his hustle and defense. As soon as KG/Pierce/Ray allen got old he cared less about defense, but for a while he was a good intangibles guy.

ewing
10-04-2016, 07:21 PM
Kevin Martin and KG come to mind.

TheMightyHumph
10-04-2016, 07:23 PM
I refuse to recognize 'advanced' stats

IndyRealist
10-04-2016, 08:09 PM
Kris Humphries, Serge Ibaka, Kevin Love off the top of my head. Damien Lillard, CJ McCollum, DeMarcus Cousins when I looked some stuff up. I could add a ton more if you consider PER "advanced".

Vee-Rex
10-04-2016, 08:22 PM
As with all stats, the problem occurs when people base their arguments solely on one particular stat, or a group of stats. All of TS%, PER, WS/48, BPM/RPM should be used as a tertiary measurement instead of a primary measurement (as they're often used).

WS/48 is terrible in those regards, as an example. The stat's sole purpose is to attempt to show a player's estimated contribution to total wins. The problem is when we compare an elite player (like James Harden) who has less talent surrounding him to a BETTER elite player (like LeBron or Chris Paul) who has more talent surrounding them.

BPM is the same way. RPM is typically regarded as a more sophisticated and accurate version of BPM, but it has its flaws too. Example: All of last year we heard about how Matthew Dellevadova had a higher RPM than Kyrie Irving and how the Cavs were better with him on the floor.

Context: Delly's minutes were almost always exclusive with LeBron's versus the opposing team's bench. LeBron vs. bench is how we built up leads in the 2nd quarter (our best quarter throughout the regular season).

Other same position-same team examples (Cole Aldrich being = to Deandre Jordan in RPM) and (Gorgui Dieng >>>>> Karl Anthony Towns) show that RPM is crap, and should only be acknowledged when it's used as a tertiary form of information WITHIN context.

There's far more examples too but I haven't looked at it in a couple months.

VORP is another advanced stat that is all sunshine-and-rainbows on the surface but can get pretty ugly if you dig deeper. It attempts to place a value of a player based on points per team possessions compared to that of an average team/player. This introduces the flaw of having a guy who is able to put up points due on a team with fewer possessions without accounting for overall impact.

According to VORP - JR Smith > Klay Thompson.

Another flaw is that VORP is based off number totals, which means that a player who misses games will pretty much always have a lower VORP than a COMPARABLE player who plays in all of them.

TS% and PER might be a little bit more rock solid but they too have their flaws. TS% places a specific value on free throws that doesn't account for and-ones and technical shots. Players who are great at finishing with contact but don't get to the free throw line often will have lower TS% than players who never finish with contact but get to the line all the time.

Free throw seekers (like Harden/Lowry) have inflated TS% because their efficiency is heavily based on scoring at the line. Now, this follows the text-book definition of efficiency but that sort of dependency isn't ideal since it relies completely on the whistle (which is why things may change during the playoffs). It's why guys like Deandre Jordan and Dwight Howard have such inflated TS% despite never taking 3-point shots.

Jeremy Lin had a .543 TS% in these last playoffs despite shooting horribly from the field (.41%FG and .21%3pt) but he had an eye-popping .61 FTr (meaning for every shot attempt he shot .61 free throws). Shows just how much free throws skews TS% and we all know free throws depend on the arbitrary whistle of the refs, especially in the playoffs. Some guys have a natural talent for drawing fouls but it's not a skill I'd throw all my weight behind and want to rely on. It goes to show the true nature of the regular season vs. the playoffs and why the playoffs are THE MOST TRUTHFUL representation of a player's skills, smaller sample size included.

So while TS% is a pretty damn accurate reflection on efficiency in a PPP sort of context, it doesn't really tell the whole story about a player's capabilities and why, and it doesn't mean that player A is more efficient than B, especially when the refs are calling it loose. I'd use it as a secondary source of judgment and only primary if used in conjunction with EVERY OTHER efficiency stat.

PER is pretty solid but it isn't perfect. It's a much more complex formula. It's a per-minute representation which can be flawed depending on a team's scheme and tendencies.

Draymond Green for example is someone I believe is a better player than his PER shows, as he pretty much equals out with Kevin Love and that's something I don't think I'd agree with right now.

Green posted a 19.3 PER this past year and Love posted 19.0. Last year, Green posted a 16.4 PER and Love posted 18.8.

At the end of the day, stats without context are crap. It goes that way with advanced stats as well as traditional stats.

The best way, IMO, to measure players is to take a combination of all stats, traditional and advanced, apply some logic (unfortunately we can't put a numerical value on this), recognize and acknowledge traits/skills/strengths/weaknesses that aren't statistically accounted for (things like clutch and certain defensive stats are poorly represented), then logically debate.

Anyone trying to prove a point and start dropping cherry-picked advanced stats is simply not worth arguing with.

valade16
10-04-2016, 08:38 PM
As with all stats, the problem occurs when people base their arguments solely on one particular stat, or a group of stats. All of TS%, PER, WS/48, BPM/RPM should be used as a tertiary measurement instead of a primary measurement (as they're often used).

WS/48 is terrible in those regards, as an example. The stat's sole purpose is to attempt to show a player's estimated contribution to total wins. The problem is when we compare an elite player (like James Harden) who has less talent surrounding him to a BETTER elite player (like LeBron or Chris Paul) who has more talent surrounding them.

BPM is the same way. RPM is typically regarded as a more sophisticated and accurate version of BPM, but it has its flaws too. Example: All of last year we heard about how Matthew Dellevadova had a higher RPM than Kyrie Irving and how the Cavs were better with him on the floor.

Context: Delly's minutes were almost always exclusive with LeBron's versus the opposing team's bench. LeBron vs. bench is how we built up leads in the 2nd quarter (our best quarter throughout the regular season).

Other same position-same team examples (Cole Aldrich being = to Deandre Jordan in RPM) and (Gorgui Dieng >>>>> Karl Anthony Towns) show that RPM is crap, and should only be acknowledged when it's used as a tertiary form of information WITHIN context.

There's far more examples too but I haven't looked at it in a couple months.

VORP is another advanced stat that is all sunshine-and-rainbows on the surface but can get pretty ugly if you dig deeper. It attempts to place a value of a player based on points per team possessions compared to that of an average team/player. This introduces the flaw of having a guy who is able to put up points due on a team with fewer possessions without accounting for overall impact.

According to VORP - JR Smith > Klay Thompson.

Another flaw is that VORP is based off number totals, which means that a player who misses games will pretty much always have a lower VORP than a COMPARABLE player who plays in all of them.

TS% and PER might be a little bit more rock solid but they too have their flaws. TS% places a specific value on free throws that doesn't account for and-ones and technical shots. Players who are great at finishing with contact but don't get to the free throw line often will have lower TS% than players who never finish with contact but get to the line all the time.

Free throw seekers (like Harden/Lowry) have inflated TS% because their efficiency is heavily based on scoring at the line. Now, this follows the text-book definition of efficiency but that sort of dependency isn't ideal since it relies completely on the whistle (which is why things may change during the playoffs). It's why guys like Deandre Jordan and Dwight Howard have such inflated TS% despite never taking 3-point shots.

Jeremy Lin had a .543 TS% in these last playoffs despite shooting horribly from the field (.41%FG and .21%3pt) but he had an eye-popping .61 FTr (meaning for every shot attempt he shot .61 free throws). Shows just how much free throws skews TS% and we all know free throws depend on the arbitrary whistle of the refs, especially in the playoffs. Some guys have a natural talent for drawing fouls but it's not a skill I'd throw all my weight behind and want to rely on. It goes to show the true nature of the regular season vs. the playoffs and why the playoffs are THE MOST TRUTHFUL representation of a player's skills, smaller sample size included.

So while TS% is a pretty damn accurate reflection on efficiency in a PPP sort of context, it doesn't really tell the whole story about a player's capabilities and why, and it doesn't mean that player A is more efficient than B, especially when the refs are calling it loose. I'd use it as a secondary source of judgment and only primary if used in conjunction with EVERY OTHER efficiency stat.

PER is pretty solid but it isn't perfect. It's a much more complex formula. It's a per-minute representation which can be flawed depending on a team's scheme and tendencies.

Draymond Green for example is someone I believe is a better player than his PER shows, as he pretty much equals out with Kevin Love and that's something I don't think I'd agree with right now.

Green posted a 19.3 PER this past year and Love posted 19.0. Last year, Green posted a 16.4 PER and Love posted 18.8.

At the end of the day, stats without context are crap. It goes that way with advanced stats as well as traditional stats.

The best way, IMO, to measure players is to take a combination of all stats, traditional and advanced, apply some logic (unfortunately we can't put a numerical value on this), recognize and acknowledge traits/skills/strengths/weaknesses that aren't statistically accounted for (things like clutch and certain defensive stats are poorly represented), then logically debate.

Anyone trying to prove a point and start dropping cherry-picked advanced stats is simply not worth arguing with.

Excellent post and I 100% agree with this. So given everything you said, which guy's advanced stats are inflated compared to his actual worth when you apply context?

IndyRealist
10-04-2016, 08:55 PM
TS% attempts to account for and-1 and technical fouls by using a .44 multiplier (I think it's 12% of free throws are not the result of a shooting foul, so 88%/2). Otherwise true.

PER is f'king horrid. If you shoot above 30.4% on 2s and 21.4% on 3s, you will raise your PER by taking more shots. PER says you are a better player by taking more shots at 31% FG. ( SOURCE (http://wagesofwins.com/2006/11/17/a-comment-on-the-player-efficiency-rating/) ) That's why players like Jamal Crawford have always looked decent in PER, despite shooting average to below average percentages.

valade16
10-04-2016, 08:57 PM
TS% attempts to account for and-1 and technical fouls by using a .44 multiplier (I think it's 12% of free throws are not the result of a shooting foul, so 88%/2). Otherwise true.

PER is f'king horrid. If you shoot above 30.4% on 2s and 21.4% on 3s, you will raise your PER by taking more shots. PER says you are a better player by taking more shots at 31% FG. ( SOURCE (http://wagesofwins.com/2006/11/17/a-comment-on-the-player-efficiency-rating/) ) That's why players like Jamal Crawford have always looked decent in PER, despite shooting average to below average percentages.

Given that, I don't understand why they don't simply adjust the formula to raise the FG% at each distance you need to make a positive impact. Or is it not that simple?

IndyRealist
10-04-2016, 09:08 PM
Given that, I don't understand why they don't simply adjust the formula to raise the FG% at each distance you need to make a positive impact. Or is it not that simple?

At the time, NBA metrics were in their infancy and PER was about as good as it got. When actual academics got involved and dissected it, they tore it apart. Hollinger's ENTIRE career and reputation focused solely around PER, so he defended it. Changing PER would be admitting he was completely wrong. There's a reason he abandoned PER to ESPN when he went to the Grizzlies. And one of the first moves he made was the very unPER-like decision to trade Rudy Gay. ESPN is too invested in people believing PER to go and change the formula now.

warfelg
10-05-2016, 07:07 AM
You can take anyone who does some decent things but plays in a limited (5-15 minutes) role and they go in this list to be honest.

It's one of the worst arguments I've seen. A low usage, low minutes player has good advanced stats, someone argues for them to play more, they get more time and advanced stats drop with a higher usage and higher minute total.

Scoots
10-05-2016, 08:00 AM
PER is broken and everybody arguing the value of an all-around player vs a chucker knows it.

TS% is broken and the man who created it, Morey, admitted that into a microphone a couple years ago at the Sloan Conference.

What we need and don't have access to yet is some sort of situational plus minus that takes opponent and teammates out of it and to play that against the particular skills/situations most needed by any given team in their needed situations. Of course this is what the teams are trying to do in evaluating players and they have whole staffs of people trying to do just that.

warfelg
10-05-2016, 08:58 AM
PER is broken and everybody arguing the value of an all-around player vs a chucker knows it.

TS% is broken and the man who created it, Morey, admitted that into a microphone a couple years ago at the Sloan Conference.

What we need and don't have access to yet is some sort of situational plus minus that takes opponent and teammates out of it and to play that against the particular skills/situations most needed by any given team in their needed situations. Of course this is what the teams are trying to do in evaluating players and they have whole staffs of people trying to do just that.

I feel like box plus-minus comes close to what you are looking for. Takes the players affect on only his teams performance.

Reb% comes close because it takes percent of available rebounds.

As SportVu becomes more understood and implement the information I think NBAs advanced stats will get better.

The other issue I have:
Advanced stats in baseball is great, because despite being 9-on-9, it's it very much a 1-on-1 sport. Basketball and football are not, and even with advanced stats, it takes a lot of watching to still understand.

Hawkeye15
10-05-2016, 09:25 AM
As with all stats, the problem occurs when people base their arguments solely on one particular stat, or a group of stats. All of TS%, PER, WS/48, BPM/RPM should be used as a tertiary measurement instead of a primary measurement (as they're often used).

WS/48 is terrible in those regards, as an example. The stat's sole purpose is to attempt to show a player's estimated contribution to total wins. The problem is when we compare an elite player (like James Harden) who has less talent surrounding him to a BETTER elite player (like LeBron or Chris Paul) who has more talent surrounding them.

BPM is the same way. RPM is typically regarded as a more sophisticated and accurate version of BPM, but it has its flaws too. Example: All of last year we heard about how Matthew Dellevadova had a higher RPM than Kyrie Irving and how the Cavs were better with him on the floor.

Context: Delly's minutes were almost always exclusive with LeBron's versus the opposing team's bench. LeBron vs. bench is how we built up leads in the 2nd quarter (our best quarter throughout the regular season).

Other same position-same team examples (Cole Aldrich being = to Deandre Jordan in RPM) and (Gorgui Dieng >>>>> Karl Anthony Towns) show that RPM is crap, and should only be acknowledged when it's used as a tertiary form of information WITHIN context.

There's far more examples too but I haven't looked at it in a couple months.

VORP is another advanced stat that is all sunshine-and-rainbows on the surface but can get pretty ugly if you dig deeper. It attempts to place a value of a player based on points per team possessions compared to that of an average team/player. This introduces the flaw of having a guy who is able to put up points due on a team with fewer possessions without accounting for overall impact.

According to VORP - JR Smith > Klay Thompson.

Another flaw is that VORP is based off number totals, which means that a player who misses games will pretty much always have a lower VORP than a COMPARABLE player who plays in all of them.

TS% and PER might be a little bit more rock solid but they too have their flaws. TS% places a specific value on free throws that doesn't account for and-ones and technical shots. Players who are great at finishing with contact but don't get to the free throw line often will have lower TS% than players who never finish with contact but get to the line all the time.

Free throw seekers (like Harden/Lowry) have inflated TS% because their efficiency is heavily based on scoring at the line. Now, this follows the text-book definition of efficiency but that sort of dependency isn't ideal since it relies completely on the whistle (which is why things may change during the playoffs). It's why guys like Deandre Jordan and Dwight Howard have such inflated TS% despite never taking 3-point shots.

Jeremy Lin had a .543 TS% in these last playoffs despite shooting horribly from the field (.41%FG and .21%3pt) but he had an eye-popping .61 FTr (meaning for every shot attempt he shot .61 free throws). Shows just how much free throws skews TS% and we all know free throws depend on the arbitrary whistle of the refs, especially in the playoffs. Some guys have a natural talent for drawing fouls but it's not a skill I'd throw all my weight behind and want to rely on. It goes to show the true nature of the regular season vs. the playoffs and why the playoffs are THE MOST TRUTHFUL representation of a player's skills, smaller sample size included.

So while TS% is a pretty damn accurate reflection on efficiency in a PPP sort of context, it doesn't really tell the whole story about a player's capabilities and why, and it doesn't mean that player A is more efficient than B, especially when the refs are calling it loose. I'd use it as a secondary source of judgment and only primary if used in conjunction with EVERY OTHER efficiency stat.

PER is pretty solid but it isn't perfect. It's a much more complex formula. It's a per-minute representation which can be flawed depending on a team's scheme and tendencies.

Draymond Green for example is someone I believe is a better player than his PER shows, as he pretty much equals out with Kevin Love and that's something I don't think I'd agree with right now.

Green posted a 19.3 PER this past year and Love posted 19.0. Last year, Green posted a 16.4 PER and Love posted 18.8.

At the end of the day, stats without context are crap. It goes that way with advanced stats as well as traditional stats.

The best way, IMO, to measure players is to take a combination of all stats, traditional and advanced, apply some logic (unfortunately we can't put a numerical value on this), recognize and acknowledge traits/skills/strengths/weaknesses that aren't statistically accounted for (things like clutch and certain defensive stats are poorly represented), then logically debate.

Anyone trying to prove a point and start dropping cherry-picked advanced stats is simply not worth arguing with.

I appreciate the detailed breakdown, but at the end of the day, the bold is so true.

Vee-Rex
10-05-2016, 10:45 AM
Excellent post and I 100% agree with this. So given everything you said, which guy's advanced stats are inflated compared to his actual worth when you apply context?

I don't think anyone has most/all of their advanced stats inflated, but certain guys definitely have a higher TS% than their actual worth or a higher VORP or WS/48 for sure. I'd have to look through a lot of players (I'll do it with more time).

Chronz
10-05-2016, 10:56 AM
Alot of what you're asking for is highly subjective/intangible, its different than the eye test in that the player can have many great games where he is undoubtedly worth his role, this is your gut test, where you just believe something nefarious is at play behind the scenes so you just have to believe it until proven otherwise. I was never proven otherwise with Rondo but I will say this, those old legs needed him to carry the load for 82 games (even though he tanked his effort when people weren't watching). Its what allowed them to up their game come playoffs and exceed their relatively mediocre RS record. That said, him being the work horse doesn't make him their best or more productive player. He wasn't that much a difference maker to exonerate his ballhoggery in the grand sense of comparing players across positions/eras.

All that said, you can find little hints of statistical influence of when players might be overrated vs their pretty stats. One of my methods was comparing +/- rates across the popular weighted barometers, saw a graph on it not too long ago.

An example of a guy that stands out today is Giannis, decently high numbers (high enough to get him voted on PSD's top-30) but low team influence. Early season Whiteside when he was first benched was like that but he finished the year strong to get his defensive marks up. Im curious to see how some of those numbers hold up this season. I want to see how Pau's defection influences his former/new team too, theres some evidence that he was collecting some meaningless rebounds and not really helping the team on that end.

-----------------------------------------------------------------------------

PER has its uses, you just have to keep in mind its on the extreme end of the usage/efficiency spectrum. You might not believe in that but others are free to do so. And the scoring threshold isn't as low when standardized to the league damnit.

Win Shares are a team stat yes, but the side we care about the most is the offensive side anyways (if we compare it to PER we may as well focus on offense like PER) and the team influence on that end is practically non-existant. If you put up the same stats, you will end up with the same OWS regardless of team level and the reason this works out is because there is some correlation to winning going on here. Its really hard to lose games if you have very efficient players. Defensively is where they attempt to make up the difference and if you feel the player was credited appropriately then feel free to add that to your overall point/summation but many times you're just not buying it.

In this situation, look at a players DWS and compare to guys with similar marks, where do they rank in the various defensive +/- techniques and does he grade poorly there? You just have valid statistical reasoning that hes being over credited, maybe hes hogging rebounds but the team is actually rebounding better without him. Maybe hes hunting for blocks but sometimes maybe the team is forcing more turnovers without you and you have very little to do with that, who knows, there is no exact science here but thats what makes it interesting to me.

IndyRealist
10-05-2016, 12:13 PM
"Usage/efficiency spectrum" implies an inverse correlation between the two, and that's never been proven that I know of.

Some players are more efficient with less shots, but some are more efficient when they get more shots, since they can "get into a rhythm". That very logic comes up every time a primary option comes back from injury and plays off the bench.

Hollinger's argument about normalizing PER is a straw man. He effectively says that an average player shooting an average % at an average rate is average in PER. Well, no sh**. The point is that a below average shooter will raise his PER by shooting more, even though he shoots poorly.

You can read the whole argument here: http://wagesofwins.com/2006/11/24/john-hollinger-responds/

mngopher35
10-05-2016, 03:22 PM
Some good answers on how it really depends on what stats you are using and you can never rely on just one.

So to answer this question knowing that individual stats might overrate a player I think players like Klove (on Wolves)/Cousins are solid answers for who might be somewhat overrated due to their stats but it isn't really empty or anything. I think Amare had his stats inflated next to nash as well generally speaking (mostly due to efficiency/team success probably). My guess is these are the type of players this thread is asking for when it says empty advanced stats though, ones who just don't seem to match up to their numbers in certain scenarios?

I am not so sure about players with empty advanced stats overall, I think that would be pretty tough to raise stats across the board unless you truly were having a big impact. Many times I think people that get overrated due to a stat like PER and volume see a lack of similar success in BPM which seems to be more efficiency/rate based for example.

IndyRealist
10-05-2016, 07:50 PM
Sometimes PSD wants to make me pull my hair out, but today I got into a facebook convo about basketball and HOLY SH** the average fan doesn't know squat about basketball. It makes me incredibly sad that we're lumped in with them.

Scoots
10-06-2016, 09:45 AM
Sometimes PSD wants to make me pull my hair out, but today I got into a facebook convo about basketball and HOLY SH** the average fan doesn't know squat about basketball. It makes me incredibly sad that we're lumped in with them.

I watch games with my 90 year old Mom and explain sets to her. She loves watching the games but she doesn't really care about the technical aspects. I think most of the "fans" out there just like watching the game and don't care about the technical distinctions between a pick and roll and a wing screen. No need to be sad about being more passionate about the technical aspects of a game.

Hawkeye15
10-06-2016, 09:53 AM
Sometimes PSD wants to make me pull my hair out, but today I got into a facebook convo about basketball and HOLY SH** the average fan doesn't know squat about basketball. It makes me incredibly sad that we're lumped in with them.

that is why most of us came here. I am lucky, my brother studies the game as much as I do, played college ball, and my best friend is pretty into basketball (and played as a kid), so I at least have 2 people that I can talk ball with. I also work with a dude who grew up playing on the streets of NYC, and played college ball (dude is 6'8", and jacked), and he is pretty knowledgeable about that game. My dad even knows a ton, he watches a lot of college (used to watch NBA before that tattoo's, etc), and played college ball.

Most people, don't have the desire to really study basketball like we do. Nothing wrong with that. It's the sport we chose to follow with a passion, and take it up 10 notches as a fan. The sport (all sports) need the fairweather, causal fan as well, that don't really understand everything they watch. hell the NFL has a zillion of these fans...

Chronz
10-06-2016, 02:54 PM
"Usage/efficiency spectrum" implies an inverse correlation between the two, and that's never been proven that I know of.

Some players are more efficient with less shots, but some are more efficient when they get more shots, since they can "get into a rhythm". That very logic comes up every time a primary option comes back from injury and plays off the bench.

Hollinger's argument about normalizing PER is a straw man. He effectively says that an average player shooting an average % at an average rate is average in PER. Well, no sh**. The point is that a below average shooter will raise his PER by shooting more, even though he shoots poorly.

You can read the whole argument here: http://wagesofwins.com/2006/11/24/john-hollinger-responds/
Well is anything ever really proven, I believe in skill curves and you dont, that hasn't changed since this happened: http://forums.prosportsdaily.com/showthread.php?740878-The-Moneyball-of-Basketball&p=23020210#post23020210

I've never seen a truly exhaustive study on it because so much goes on besides just a change in usage, I can find arguments from both sides but feel free to point me to one thats not WoW centered unless absolutely necessary.

I do know some of the analysis used to discover any sort of relationship is usually kept close to the chest. Dean O was the first guy to deliver the message to my head but IIRC he couldn't release everything because its what got him into the NBA. Hes posted graphs showing his results over the years to hint at a correlation but its on a game to game basis and I feel like that can be prone to randomness or quality of opponent.

I try to isolate exact incidents/scenarios where less variables may come into play. Its why I showed you the Artest/Ariza example. Another would be to look at Pau's swap from Memphis to LA, the decrease in usage and increase in overall offensive efficiency lines up with the 1.5:1 ratio I remember reading about.

We've never seen eye to eye on shot creation and its implied difficulty but I would agree some are better/less prepared for more/less shots. Hard to find a correlation when it works out that way but do you really expect stagnation? Like when Harden went from super efficient sub for OKC to consistent go-to guy in Houston, were you expecting a dropoff or not?

Chronz
10-06-2016, 03:11 PM
Some good answers on how it really depends on what stats you are using and you can never rely on just one.

So to answer this question knowing that individual stats might overrate a player I think players like Klove (on Wolves)/Cousins are solid answers for who might be somewhat overrated due to their stats but it isn't really empty or anything. I think Amare had his stats inflated next to nash as well generally speaking (mostly due to efficiency/team success probably). My guess is these are the type of players this thread is asking for when it says empty advanced stats though, ones who just don't seem to match up to their numbers in certain scenarios?

I am not so sure about players with empty advanced stats overall, I think that would be pretty tough to raise stats across the board unless you truly were having a big impact. Many times I think people that get overrated due to a stat like PER and volume see a lack of similar success in BPM which seems to be more efficiency/rate based for example.

Ways to inflate your own stats. Chase blocks, only pass when you feel it can lead to an assist (even if its a low% assist), chase offensive rebounds in place of getting back defensively/chase defensive rebounds instead of helping out your teammates defensively (though these can hurt your Defensive metrics) all of these can radically alter the way you are perceived and analyzed and to me, Rondo was lucky to be on such a disciplined defensive team as he was the only guy who could crash the boards while his teammates ran back (though Doc did say hes the only guy trusted to do so because he can run back quickly but could over do it), he often only passed if it led to his assist though in his defense, he would sometimes refuse to take a shot which would have heightened his value more than an assist would.

In most cases, you can find hollow stats in +/- based analysis, if the team doesn't really need it from you or if it prevents the team from running other stuff. Like Chris Webber had a decent PER but he was hurting his team more than helping it with his high usage. It was not surprising that the Kings had so much success in his absence or that it didn't miss a beat in games he missed.

I remember reading a fantasy sports mag where they wondered if he was truly worthy of the tag "Franchise Player" despite his 20-10-5's.


Curious about your thoughts on usage vs efficiency, always felt you agreed in some sort of skill curve, its why you talked about how we might overrate efficiency the other day, pretty sure you brought up AI in the past too. Like he was more efficient in Denver and I think we agree its in some part due to the decrease in usage which allowed him to pick and choose his spots better.

Chronz
10-06-2016, 03:24 PM
Look at Rondo vs George Hill. Doesn't sound like much of a comparison but I have reason to believe the Celtics would have been better off with a PG like him rather than someone who stunted the touches of his teammates.


Not sure if you guys have read my thoughts on the different aspects of passing/creating for others but here it is: http://forums.prosportsdaily.com/showthread.php?744442-What-is-the-most-important-aspect-of-passing


George Hill doesn't dominate the ball, he is a classic distributor and Rondo is a ball hogging "pure" PG aka a playmaker/facilitator. The one year George Hill was forced into a primary role was the year PG was injured and much of their offense was gone, that half a season we saw Hill up his advanced stats across the board to even higher levels than prime Rondo (higher usage with no loss in efficiency as well, Indy). I was one of the few who wasn't laughing when someone said he and Kyrie were at one point comparable. Its because people think because they dont see you doing something or producing boxscore numbers that you're not really out there doing anything special. They dont see how valuable it is to have someone capable of running the offense but also being an outlet option for others.

Gimme George Hill over Rondo any day and you will find it easier to build a TEAM.

IndyRealist
10-06-2016, 03:53 PM
Well is anything ever really proven, I believe in skill curves and you dont, that hasn't changed since this happened: http://forums.prosportsdaily.com/showthread.php?740878-The-Moneyball-of-Basketball&p=23020210#post23020210

I've never seen a truly exhaustive study on it because so much goes on besides just a change in usage, I can find arguments from both sides but feel free to point me to one thats not WoW centered unless absolutely necessary.

I do know some of the analysis used to discover any sort of relationship is usually kept close to the chest. Dean O was the first guy to deliver the message to my head but IIRC he couldn't release everything because its what got him into the NBA. Hes posted graphs showing his results over the years to hint at a correlation but its on a game to game basis and I feel like that can be prone to randomness or quality of opponent.

I try to isolate exact incidents/scenarios where less variables may come into play. Its why I showed you the Artest/Ariza example. Another would be to look at Pau's swap from Memphis to LA, the decrease in usage and increase in overall offensive efficiency lines up with the 1.5:1 ratio I remember reading about.

We've never seen eye to eye on shot creation and its implied difficulty but I would agree some are better/less prepared for more/less shots. Hard to find a correlation when it works out that way but do you really expect stagnation? Like when Harden went from super efficient sub for OKC to consistent go-to guy in Houston, were you expecting a dropoff or not?

Usage curves are really not a trade secret, it just takes someone with access to the data to crunch the numbers. While we can all look it up on BBR, we don't have the formatted numbers to just insert it into a spreadsheet for 30 seasons worth of players.

That's the real problem with analytics at this point. Everything is proprietary and no one shares their method. Consequently there's a lot of bad analysis out there and no way to fact check it. The people using it and the people in charge don't understand it enough to say "this is bunk". All you can do is eye test it and say "it has Matter Bonner HOW high?!" which is not a real measure of anything. If you don't know WHY it's wrong you don't know that it is wrong.

mngopher35
10-06-2016, 03:56 PM
Ways to inflate your own stats. Chase blocks, only pass when you feel it can lead to an assist (even if its a low% assist), chase offensive rebounds in place of getting back defensively/chase defensive rebounds instead of helping out your teammates defensively (though these can hurt your Defensive metrics) all of these can radically alter the way you are perceived and analyzed and to me, Rondo was lucky to be on such a disciplined defensive team as he was the only guy who could crash the boards while his teammates ran back (though Doc did say hes the only guy trusted to do so because he can run back quickly but could over do it), he often only passed if it led to his assist though in his defense, he would sometimes refuse to take a shot which would have heightened his value more than an assist would.

In most cases, you can find hollow stats in +/- based analysis, if the team doesn't really need it from you or if it prevents the team from running other stuff. Like Chris Webber had a decent PER but he was hurting his team more than helping it with his high usage. It was not surprising that the Kings had so much success in his absence or that it didn't miss a beat in games he missed.

I remember reading a fantasy sports mag where they wondered if he was truly worthy of the tag "Franchise Player" despite his 20-10-5's.

I agree with a lot of what you say and the +/- and RPM type of stats are what I mean in how it is tough to raise advanced stats across the board. If you are chasing specific box score stats it can show up lesser in other analysis as you pointed out. I do like the specific examples you give as not contesting to go for the rebound, chasing blocks and playing for the assist is something I think definitely happens with some players.

I actually haven't looked a ton into Webber to be honest but I will take a closer look, I think he has always been a good defender but could see his offense maybe not matching up to the stats like PER? Will try looking into that sometime.



Curious about your thoughts on usage vs efficiency, always felt you agreed in some sort of skill curve, its why you talked about how we might overrate efficiency the other day, pretty sure you brought up AI in the past too. Like he was more efficient in Denver and I think we agree its in some part due to the decrease in usage which allowed him to pick and choose his spots better.

Yes I have brought it up multiple times depending on the situations but it is honestly more of an eye test thing for me than statistical analysis as I don't have specifics to back it up. You seem to get where I am coming from on it though.

Basically I think when a player has to raise his usage or carry a bigger load than he should it can really affect the efficiency numbers (especially come playoff time where team defenses focus in more on weaknesses). Being the main playmaker/creator for others to me is a factor in an individuals own efficiency numbers and having someone else or multiple others to do that for you (or take part of the load) is big. It is also a big reason why I always separate the main threats from secondary threats when comparing individual players and their numbers, being the one with all the attention on you defensively is big.

Webber would probably be an example of someone I wouldn't be as lenient towards for his efficiency because he was on some pretty well rounded teams offensively. Not sure about his usage but this is why I say it is more of an eye test thing for me where if I notice someone is carrying a major load/defensive attention due to the lack of help I try to take it into account when looking at the numbers and a drop in efficiency is more excusable to me. If I think someone is doing too much and it isn't because of a lack of help then I won't be as forgiving in my rankings (also applies to AI imo actually, he is a tricky one where it is a mix of lack of help and his own issues to me).

mngopher35
10-06-2016, 04:00 PM
Look at Rondo vs George Hill. Doesn't sound like much of a comparison but I have reason to believe the Celtics would have been better off with a PG like him rather than someone who stunted the touches of his teammates.


Not sure if you guys have read my thoughts on the different aspects of passing/creating for others but here it is: http://forums.prosportsdaily.com/showthread.php?744442-What-is-the-most-important-aspect-of-passing


George Hill doesn't dominate the ball, he is a classic distributor and Rondo is a ball hogging "pure" PG aka a playmaker/facilitator. The one year George Hill was forced into a primary role was the year PG was injured and much of their offense was gone, that half a season we saw Hill up his advanced stats across the board to even higher levels than prime Rondo (higher usage with no loss in efficiency as well, Indy). I was one of the few who wasn't laughing when someone said he and Kyrie were at one point comparable. Its because people think because they dont see you doing something or producing boxscore numbers that you're not really out there doing anything special. They dont see how valuable it is to have someone capable of running the offense but also being an outlet option for others.

Gimme George Hill over Rondo any day and you will find it easier to build a TEAM.

I would agree with that too, I actually think Hill is underrated somewhat for this reason. Getting him was a nice boost for Utah imo.

IndyRealist
10-06-2016, 05:21 PM
Well is anything ever really proven, I believe in skill curves and you dont, that hasn't changed since this happened: http://forums.prosportsdaily.com/showthread.php?740878-The-Moneyball-of-Basketball&p=23020210#post23020210

I've never seen a truly exhaustive study on it because so much goes on besides just a change in usage, I can find arguments from both sides but feel free to point me to one thats not WoW centered unless absolutely necessary.

I do know some of the analysis used to discover any sort of relationship is usually kept close to the chest. Dean O was the first guy to deliver the message to my head but IIRC he couldn't release everything because its what got him into the NBA. Hes posted graphs showing his results over the years to hint at a correlation but its on a game to game basis and I feel like that can be prone to randomness or quality of opponent.

I try to isolate exact incidents/scenarios where less variables may come into play. Its why I showed you the Artest/Ariza example. Another would be to look at Pau's swap from Memphis to LA, the decrease in usage and increase in overall offensive efficiency lines up with the 1.5:1 ratio I remember reading about.

We've never seen eye to eye on shot creation and its implied difficulty but I would agree some are better/less prepared for more/less shots. Hard to find a correlation when it works out that way but do you really expect stagnation? Like when Harden went from super efficient sub for OKC to consistent go-to guy in Houston, were you expecting a dropoff or not?

Dean Oliver, p 239 of Basketball on Paper:

Skill curves are neat to look at, but you probably don’t want to know the details about how to make them. Generally, they come from looking at box scores and general trends that players show when using a lot of possessions. How much better do players seem to get if they use fewer possessions? If they seem to get worse when they use fewer possessions, I say, ‘That’s not right.’ It’s just not a sustainable trend. Effectively, I force the curves to be declining. The details of how I do that would scare someone who doesn’t know formal statistics. They would scare someone who did know formal statistics for different reasons

He very clearly states he ignored data that didn't agree with his predetermined answer.

Chronz
10-06-2016, 05:53 PM
F all that have you guys heard the news? Just came back from booz shoppin and got word that rob is NOT the father. Blacc chyna wronged him

europagnpilgrim
10-07-2016, 02:55 PM
F all that have you guys heard the news? Just came back from booz shoppin and got word that rob is NOT the father. Blacc chyna wronged him

I haven't until now and we all know the ''show'' must go on and the mother still collects her 10-15 manager pct off her kids as usual

they should do a episode on Maury one of these days

Jeffy25
10-07-2016, 05:15 PM
I feel like a lot of people feel like advanced statistics are some mystical boogie-man out there to tear down the legends of players like AI and Kobe or something.

Statistics do one thing and one thing only. They answer a question.

How many points per game is this player averaging over this many games?
How efficient is this player when he is on the court?
How many points does this players team net allow or go ahead when this player is on the court?

These are numbers that are answering those questions. They are not saying who is more valuable or who is the better athlete or basketball player.

These numbers can help you paint a picture to help answer those questions, but they are imperfect.

Because the questions we are asking (best player ever, best player this year, best player all-time ranking, etc) are imperfect questions that require a plethora of variables and information that can't be isolated to specific answers because some stats so they were elite at some aspect of the game.

Basketball is a sport where 10 players are moving in combination of one another on the court. Different pressurized moments exist affecting win probability and effort is pushed at varying times as well.

But those gut feelings fans have when watching players are incredibly fallible and nobody should be answering their top 'whatever' players based on what they have seen or think they remember. There is absolutely zero chance your eyes or memories can recall anything in basketball history or current season well enough to assess anything even remotely close to accurate like that. You also don't have a memory or vision of the game that understands efficiency and talent to even begin to ***** if AI was really creating shots that nobody else could. How often did he run into defensive clouded zones on the court that someone like Ray Allen may have avoided? That's simplistic information that you can't measure from your couch, heck not even a professionally trained scout may be able to answer that. On top of that, you don't have the ability to watch ever game, with perfect memory, and perfectly assess all players equally across time. Just admit it, you are a fan of that player.


Advanced stats aren't out there to get you or your favorite player. They answer questions. That's all they do. And some of the questions we are asking can't be answered with advanced statistics yet. But having motion tracked cameras on the court will help us assess and understand the game better and better over time and new statistics will come out with more information than we have now.

FlashBolt
10-07-2016, 09:59 PM
As with all stats, the problem occurs when people base their arguments solely on one particular stat, or a group of stats. All of TS%, PER, WS/48, BPM/RPM should be used as a tertiary measurement instead of a primary measurement (as they're often used).

WS/48 is terrible in those regards, as an example. The stat's sole purpose is to attempt to show a player's estimated contribution to total wins. The problem is when we compare an elite player (like James Harden) who has less talent surrounding him to a BETTER elite player (like LeBron or Chris Paul) who has more talent surrounding them.

BPM is the same way. RPM is typically regarded as a more sophisticated and accurate version of BPM, but it has its flaws too. Example: All of last year we heard about how Matthew Dellevadova had a higher RPM than Kyrie Irving and how the Cavs were better with him on the floor.

Context: Delly's minutes were almost always exclusive with LeBron's versus the opposing team's bench. LeBron vs. bench is how we built up leads in the 2nd quarter (our best quarter throughout the regular season).

Other same position-same team examples (Cole Aldrich being = to Deandre Jordan in RPM) and (Gorgui Dieng >>>>> Karl Anthony Towns) show that RPM is crap, and should only be acknowledged when it's used as a tertiary form of information WITHIN context.

There's far more examples too but I haven't looked at it in a couple months.

VORP is another advanced stat that is all sunshine-and-rainbows on the surface but can get pretty ugly if you dig deeper. It attempts to place a value of a player based on points per team possessions compared to that of an average team/player. This introduces the flaw of having a guy who is able to put up points due on a team with fewer possessions without accounting for overall impact.

According to VORP - JR Smith > Klay Thompson.

Another flaw is that VORP is based off number totals, which means that a player who misses games will pretty much always have a lower VORP than a COMPARABLE player who plays in all of them.

TS% and PER might be a little bit more rock solid but they too have their flaws. TS% places a specific value on free throws that doesn't account for and-ones and technical shots. Players who are great at finishing with contact but don't get to the free throw line often will have lower TS% than players who never finish with contact but get to the line all the time.

Free throw seekers (like Harden/Lowry) have inflated TS% because their efficiency is heavily based on scoring at the line. Now, this follows the text-book definition of efficiency but that sort of dependency isn't ideal since it relies completely on the whistle (which is why things may change during the playoffs). It's why guys like Deandre Jordan and Dwight Howard have such inflated TS% despite never taking 3-point shots.

Jeremy Lin had a .543 TS% in these last playoffs despite shooting horribly from the field (.41%FG and .21%3pt) but he had an eye-popping .61 FTr (meaning for every shot attempt he shot .61 free throws). Shows just how much free throws skews TS% and we all know free throws depend on the arbitrary whistle of the refs, especially in the playoffs. Some guys have a natural talent for drawing fouls but it's not a skill I'd throw all my weight behind and want to rely on. It goes to show the true nature of the regular season vs. the playoffs and why the playoffs are THE MOST TRUTHFUL representation of a player's skills, smaller sample size included.

So while TS% is a pretty damn accurate reflection on efficiency in a PPP sort of context, it doesn't really tell the whole story about a player's capabilities and why, and it doesn't mean that player A is more efficient than B, especially when the refs are calling it loose. I'd use it as a secondary source of judgment and only primary if used in conjunction with EVERY OTHER efficiency stat.

PER is pretty solid but it isn't perfect. It's a much more complex formula. It's a per-minute representation which can be flawed depending on a team's scheme and tendencies.

Draymond Green for example is someone I believe is a better player than his PER shows, as he pretty much equals out with Kevin Love and that's something I don't think I'd agree with right now.

Green posted a 19.3 PER this past year and Love posted 19.0. Last year, Green posted a 16.4 PER and Love posted 18.8.

At the end of the day, stats without context are crap. It goes that way with advanced stats as well as traditional stats.

The best way, IMO, to measure players is to take a combination of all stats, traditional and advanced, apply some logic (unfortunately we can't put a numerical value on this), recognize and acknowledge traits/skills/strengths/weaknesses that aren't statistically accounted for (things like clutch and certain defensive stats are poorly represented), then logically debate.

Anyone trying to prove a point and start dropping cherry-picked advanced stats is simply not worth arguing with.

Which is why Kyrie Irving > Lowry but according to Hawkeye, Kyrie is on the level of Rubio.. Oh boy...

IndyRealist
10-08-2016, 04:12 PM
I feel like a lot of people feel like advanced statistics are some mystical boogie-man out there to tear down the legends of players like AI and Kobe or something.

Statistics do one thing and one thing only. They answer a question.

How many points per game is this player averaging over this many games?
How efficient is this player when he is on the court?
How many points does this players team net allow or go ahead when this player is on the court?

These are numbers that are answering those questions. They are not saying who is more valuable or who is the better athlete or basketball player.

These numbers can help you paint a picture to help answer those questions, but they are imperfect.

Because the questions we are asking (best player ever, best player this year, best player all-time ranking, etc) are imperfect questions that require a plethora of variables and information that can't be isolated to specific answers because some stats so they were elite at some aspect of the game.

Basketball is a sport where 10 players are moving in combination of one another on the court. Different pressurized moments exist affecting win probability and effort is pushed at varying times as well.

But those gut feelings fans have when watching players are incredibly fallible and nobody should be answering their top 'whatever' players based on what they have seen or think they remember. There is absolutely zero chance your eyes or memories can recall anything in basketball history or current season well enough to assess anything even remotely close to accurate like that. You also don't have a memory or vision of the game that understands efficiency and talent to even begin to ***** if AI was really creating shots that nobody else could. How often did he run into defensive clouded zones on the court that someone like Ray Allen may have avoided? That's simplistic information that you can't measure from your couch, heck not even a professionally trained scout may be able to answer that. On top of that, you don't have the ability to watch ever game, with perfect memory, and perfectly assess all players equally across time. Just admit it, you are a fan of that player.


Advanced stats aren't out there to get you or your favorite player. They answer questions. That's all they do. And some of the questions we are asking can't be answered with advanced statistics yet. But having motion tracked cameras on the court will help us assess and understand the game better and better over time and new statistics will come out with more information than we have now.

qft.

europagnpilgrim
10-10-2016, 01:38 AM
Which is why Kyrie Irving > Lowry but according to Hawkeye, Kyrie is on the level of Rubio.. Oh boy...

Which he is totally out his bball mind

its a reason why Rubio was stashed away overseas after being drafted way way way too high and Kyrie was a top pick after injuring his toe/foot in college, because he was just always better

Lowry is more comparable but he doesn't have the takeover skills/ability of Kyrie but he is a better tough nose defender, both are better than Rubio by a nice margin