WARP: Valued metric for ranking or crap?

[Killakobe81]

Moses death led me to a interesting article regarding Malone's place among the greats. Malone makes the top 10 in WARP ranked by peak season (and the only players to crack a 100) You can read about it here (http://espn.go.com/nba/story/_/id/13655965/nba-moses-malone-truly-unique-talent-according-numbers). I will post my thoughts on the list but would love to hear yours as well.

First, the missing. Kobe, Dirk, Hakeem and Duncan do not make the top 10 by this metric. Kareem also gets penalized because it starts in 78-79 (IIRC) In fact, Moses just "boxed out" Duncan and Hakeem (Kobe was not even close)

Here is the list, my thoughts below:

1. Jordan (the GOAT leads in peak WARP)
2. LeBron (not a shock metrics love his all-around game has most seasons in 1st 2nd highest WARP)
3. David Robinson (I knew metrics loved him bit not THIS much)
4. KG (ditto and could be used in the Dirk vs. KG debate)
5. Magic (not sure I buy a list that has him 5th and behind KG/David but whatever)
6. Stockton (wow, great player but surprised he kills Isiah)
7. Bird (a bit low cant justify him under stockton but OK)
8. Barkley (only he and David are ringless on this list but he was a beast)
9. Shaq (surprised the most dominant ever is this low)
10. Moses

My thoughts ... if you take these at face value:

Lebron if he could ring more has a legit GOAT case. He led this category 7 times so far.
David at his peak was a beast and the numbers suggest better than Tim.
At his best, Magic was slightly better than Bird and Stockton ...

And finally it confirms that MJ at his best he was the GOAT (even over James, Shaq etc.)

Not saying I agree with this list fully but very interesting. Can some of my numbers guys like Amb please share their thoughts? Please leave Kobe out he did not make this list and is irrelevant unless you can bring up facts or another metric on why he (Dirk, Tim or Hakeem) should be on this list.

[Killakobe81]

One last note no offense Kool my brother but this is a a good case for Magic over Kobe as GOAT Laker ...not that I just go by metrics.

[313]

I've heard of it but never really looked into it. What is it based off of?

[Medvedenko]

Here's what I do....I watch the games, and see what impacts the game. A lot of time these stats are just that stats, and there's so much that goes into the game that is beyond # crunching. Just having X player on a team can impact the game with their presence alone, adjusting defenses, match ups and schemes. This will translate in overall tempo and yes adjust the #'s, however not always felt on an individual level. I can see how saber metrics work in baseball and to a degree analytics in football. However, in a game like bball there's so much Eb and flow and intangibles are rarely translated into #'s that impact the game. While a lot do, simple ones like Points, rebounds, assists etc.

There's too much feel in the game that goes beyond the figures.

[Killakobe81]

I've heard of it but never really looked into it. What is it based off of?

Great question you can find a detailed breakdown here (http://www.sonicscentral.com/warp.html)

Here is the first 3 paragraphs from the link ...

WARP stands for Wins Above Replacement Player. The term and concept are borrowed from sabermetrics and, specifically, Baseball Prospectus. Conceptually, the WARP system seeks to evaluate players in the context of a team made up of them and four completely average players. The performance of this team is then compared to that of a team made up of four average players and one replacement-level player. The method draws heavily on the work of Dean Oliver.

What are the benefits of this method?
For one, this rating system is very flexible. Players can be rated on a per-minute basis (using the theoretical "winning percentage" of the team with four average players), in terms of their offense and defense and in terms of their overall value (WARP itself). Using replacement level shows the value of players that can play heavy minutes and avoid injury while continuing to perform above replacement level. Using wins gives a measure of value that is easy to understand and constant over time. Lastly, by eschewing the traditional linear weights method so common in basketball analysis, I believe WARP does a better job of incorporating defensive value.

What are the limitations?
Like all rating systems based on box-score data, WARP cannot account for contributions that are not tracked in the box score, most notably on defense. It does no better than linear weights methods at evaluating players like Bruce Bowen. Also, it requires a number of assumptions - the value of assists, the trade-off between usage and efficiency, and replacement level.

Again some may find it crap but if you stand by win shares etc. this is the way the league is heading ... and if you find fault with this then why is win shares or win.shares per 48 any better?

[Killakobe81]

Here's what I do....I watch the games, and see what impacts the game. A lot of time these stats are just that stats, and there's so much that goes into the game that is beyond # crunching. Just having X player on a team can impact the game with their presence alone, adjusting defenses, match ups and schemes. This will translate in overall tempo and yes adjust the #'s, however not always felt on an individual level. I can see how saber metrics work in baseball and to a degree analytics in football. However, in a game like bball there's so much Eb and flow and intangibles are rarely translated into #'s that impact the game. While a lot do, simple ones like Points, rebounds, assists etc.

There's too much feel in the game that goes beyond the figures.

Great points dont disagree. But although we may quibble with the order (KG/David are so high) or a few missing guys no one can argue those 10 are all great top 25 players. These numbers place them 10th but some of the things you mention could place other players above these 10. but I dint make this list. In fact, Im accused of always defending Kobe and he is nowhere near the top 10 on this list. at least not top 12 ...

[Raven]

the stat can't be used to categorize how good a player is, it only speaks about what it's name says. Anything else imo is abusing the stat. It speaks more about how teams were created with those players than about which one was better.

[spurraider21]

One last note no offense Kool my brother but this is a a good case for Magic over Kobe as GOAT Laker ...not that I just go by metrics.
:lmao the fact that you need to have this comment just shows how much the rapist has laker nation by the balls

[Killakobe81]

the stat can't be used to categorize how good a player is, it only speaks about what it's name says. Anything else imo is abusing the stat. It speaks more about how teams were created with those players than about which one was better.

OK well how is that different than most advanced metrics? Are you against win shares or win shares per 48? How is it abusing? Baseball has been using it for a while ...

[spurraider21]

Here's what I do....I watch the games, and see what impacts the game. A lot of time these stats are just that stats, and there's so much that goes into the game that is beyond # crunching. Just having X player on a team can impact the game with their presence alone, adjusting defenses, match ups and schemes. This will translate in overall tempo and yes adjust the #'s, however not always felt on an individual level. I can see how saber metrics work in baseball and to a degree analytics in football. However, in a game like bball there's so much Eb and flow and intangibles are rarely translated into #'s that impact the game. While a lot do, simple ones like Points, rebounds, assists etc.

There's too much feel in the game that goes beyond the figures.
aka "killer instinct"

[Killakobe81]

:lmao the fact that you need to have this comment just shows how much the rapist has laker nation by the balls

Huh? Me and Kool disagree on who the GOAT Laker is but really tough to choose anyone but those two. Metrics favor Magic, raw stats Kobe. Kareem and Shaq were great but it's a two man race ...Magic wins for me.

[Killakobe81]

aka "killer instinct"

But intangibles are another reason why I would Take Magic over Kobe cant ignore them completely but it shouldnt erase metrics completely either. Players should be ranked with a holistic approach if not you can argue KG and David meant more to their teams and the league than Tim or Hakeem.

[spurraider21]

But intangibles are another reason why I would Take Magic cant ignore them completely but it shouldnt erase metrics completely either. Players should be ranked with a holistic approach if not you can argue KG and David meant more to their teams and the league than Tim or Hakeem.
VORP/WARP is just one advanced stat. duncan trumps KG in others like win shares, or just about any metric when you look at their playoff numbers

[ambchang]

WARP has a few fatal flaws:

1) It's based heavily on traditional stats. It's advanced like PER is advanced, which is to say it isn't advanced at all. Stat padders get rewarded in WARP as much as in traditional stats.
2) It assigns arbitrary value to traditional stats. Why is there a multiplier of 0.75 for assists?
3) It depends heavily on the quality of your teammates. If you have horrible teammates to replace you, you get rewarded for having a higher team rating improvement with you on the court vs. with you off it. It's a lot different having Howard Eisley replacing you compared to having Kevin Johnson subbing for you.
4) It is horrible in evaluating defense. Absolutely horrible. Again, a strong defensive team like the 04 Pistons will have the indivdiuals penalized. It's just not very developed.
5) It's era dependent. A replacement player in 1976 (dark age) is different from a replacement player in 1987. A replacement shooting guard today is a lot weaker than a replacement SG in the 80s, a replacement SF today is a lot stronger than a replacement SF in the 90s or 80s. A replacement C today is a lot weaker than a replacement C in the 90s. While WARP in basketball doesn't really take positions into account, rebounds and blks or other positionally-biased metrics do come into account. To take it one step further, a C in the 80s is expected to rebound and block shots, a C in the 90s is expected to be the defensive lynch pin, a C today is expected to draw the defense out to the perimeter to open up the lane and open up the offense. The roles are totally different, and to compare players across era doesn't work with WARP.

Overall, it is directionally correct. Good players rank high, bad players rank low. But it most definitely NOT a comprehensive metric than can just be used to rank players. You take a whole bunch of factors into account, and when a player is ranked consistently high on a whole bunch of metrics, it does signify that said player was a great player. Conversely, if a player consistently ranks low across a large number of metrics, you have to say that the player really isn't that great at all.

[ambchang]

Directly to the rankings, I am not surprised Robinson was ranked that high. He was a monster statistically, too bad he just had a horrible team built around him. Switch him and pedophile and I can guarantee the Jazz would have won at least 1 championship, even going through the Jordan Bulls at least once or twice.

[Killakobe81]

WARP has a few fatal flaws:

1) It's based heavily on traditional stats. It's advanced like PER is advanced, which is to say it isn't advanced at all. Stat padders get rewarded in WARP as much as in traditional stats.
2) It assigns arbitrary value to traditional stats. Why is there a multiplier of 0.75 for assists?
3) It depends heavily on the quality of your teammates. If you have horrible teammates to replace you, you get rewarded for having a higher team rating improvement with you on the court vs. with you off it. It's a lot different having Howard Eisley replacing you compared to having Kevin Johnson subbing for you.
4) It is horrible in evaluating defense. Absolutely horrible. Again, a strong defensive team like the 04 Pistons will have the indivdiuals penalized. It's just not very developed.
5) It's era dependent. A replacement player in 1976 (dark age) is different from a replacement player in 1987. A replacement shooting guard today is a lot weaker than a replacement SG in the 80s, a replacement SF today is a lot stronger than a replacement SF in the 90s or 80s. A replacement C today is a lot weaker than a replacement C in the 90s. While WARP in basketball doesn't really take positions into account, rebounds and blks or other positionally-biased metrics do come into account. To take it one step further, a C in the 80s is expected to rebound and block shots, a C in the 90s is expected to be the defensive lynch pin, a C today is expected to draw the defense out to the perimeter to open up the lane and open up the offense. The roles are totally different, and to compare players across era doesn't work with WARP.

Overall, it is directionally correct. Good players rank high, bad players rank low. But it most definitely NOT a comprehensive metric than can just be used to rank players. You take a whole bunch of factors into account, and when a player is ranked consistently high on a whole bunch of metrics, it does signify that said player was a great player. Conversely, if a player consistently ranks low across a large number of metrics, you have to say that the player really isn't that great at all.

Some nice thoughts. Some counters:

Some defensive players like KG and David ranked very high.
Jordan and Lebron are not defenders on that level but also ranked very high actually the highest.
Of course stat padders will get rewarded, and with assist being assigned such a high value outside of Jordan the entire top 5-7 is made up of guys who are pass first guys. But I never really considered David or KG stat padders but more stat box "stuffers" even Magic was that to a lessor degree.
comparing guys across any eras is difficult for any metric unless you adjust for pace. Still leading multiple years in this stat like Jordan and Lebron just confirms what we already knew they are both truly great.

[Raven]

OK well how is that different than most advanced metrics? Are you against win shares or win shares per 48? How is it abusing? Baseball has been using it for a while ...

I think it's self explanatory, really.. and yeah, win shares suck too. offensive, defensive and net rating are waaaaay better imo.

[Killakobe81]

I think it's self explanatory, really.. and yeah, win shares suck too. offensive, defensive and net rating are waaaaay better imo.

Fair enough. But you can see how if I love David for example I can pick this metric to support my argument and someone can pick another to push him down? Truth is no real great one way to rank players, we need multiple ways to do it effectively.

[Raven]

Fair enough. But you can see how if I love David for example I can pick this metric to support my argument and someone can pick another to push him down? Truth is no real great one way to rank players, we need multiple ways to do it effectively.

well, since it's a team game there can never be an absolute truth on a single player. You can get fairly close though, but I doubt saying how better x was than his bench warmer, is the way, since it's good management to keep your best players fresh.

[Killakobe81]

well, since it's a team game there can never be an absolute truth on a single player. You can get fairly close though, but I doubt saying how better x was than his bench warmer, is the way, since it's good management to keep your best players fresh.

Yes, it does seem convoluted but I think the science behind the metric is to value how much better (or worse) off you would be by letting player X go for Y. Which has value (as a GM) on how you decide to spend millions of dollars. and if that is how teams are making decisions ... I get it ... even if I don't agree. Used here I think it is meant to signify the gap between Lebron/MJ and the average player at that position.
Jordan having the largest spread in that measure is no surprise. Drexler was arguably the #2 SG of a good chunk of Mj's career but most would agree he was a great HOF player but nowhere near MJ ... but before today havent given this much thought. It's why I asked guys like Amb who actually live for this shit ...after PER I just feel this shit has gone to far ...BUT I still find rankings like these and clutch rankings interesting since they differ from what we "expect".

[Raven]

Yes, it does seem convoluted but I think the science behind the metric is to value how much better (or worse) off you would be by letting player X go for Y. Which has value (as a GM) on how you decide to spend millions of dollars. and if that is how teams are making decisions ... I get it ... even if I don't agree. Used here I think it is meant to signify the gap between Lebron/MJ and the average player at that position.
Jordan having the largest spread in that measure is no surprise. Drexler was arguably the #2 SG of a good chunk of Mj's career but most would agree he was a great HOF player but nowhere near MJ ... but before today havent given this much thought. It's why I asked guys like Amb who actually live for this shit ...after PER I just feel this shit has gone to far ...BUT I still find rankings like these and clutch rankings interesting since they differ from what we "expect".

to me it is just an indication of how much a team invested in the replacement. It makes sense that those two are at the top since they basically played non stop. You don't pay much for someone who plays barely 10 minutes. It would be funny to see the number wilt chamberlain would have registered if it was counted the year he never got to the bench :lol ... you can compute that easily

[Silver&Black]

8. Barkley (only he and David are ringless on this list but he was a beast)


:wakeup

[Buddy Mignon]

One last note no offense Kool my brother but this is a a good case for Magic over Kobe as GOAT Laker ...not that I just go by metrics.

You must have missed the 80's. Magics run was only second to Russells. We dominated. Kobe's 20 years have been a roller coaster.

[Killakobe81]

You must have missed the 80's. Magics run was only second to Russells. We dominated. Kobe's 20 years have been a roller coaster.

What? Magic is crazy high on my own list ...

[ambchang]

Some nice thoughts. Some counters:

Some defensive players like KG and David ranked very high.

They put up great offensive numbers as well. Also, both of them got great amounts of rebounds and blocks. The funny thing is, the two were more similar than different. KG was a natural SF playing PF, Robinson was a natural PF playing C. Both had horrible teammates, neither had a real go-to offensive move, both were made to be the defensive anchor of a championship team, but can be an extremely effective scorer, both were ahead of their time in the sense that they were these uber-athletic bigs, neither had great wing support throughout their entire careers, both won championships as the pseudo-lead after their peaks.


Jordan and Lebron are not defenders on that level but also ranked very high actually the highest. They put up great numbers across the board.


Of course stat padders will get rewarded, and with assist being assigned such a high value outside of Jordan the entire top 5-7 is made up of guys who are pass first guys. But I never really considered David or KG stat padders but more stat box "stuffers" even Magic was that to a lessor degree.

WS, ORtg, DRtg, and VORP are not really stat padding friendly. They all have their flaws, but I find WARP to be one of the weakest.


comparing guys across any eras is difficult for any metric unless you adjust for pace. Still leading multiple years in this stat like Jordan and Lebron just confirms what we already knew they are both truly great.

It's not just pace, and I think most advanced stats already took pace into account. The biggest issue is that the game has evolved in that the roles of players have changed. Advanced stats are "made" today so it is constructed with assumptions using today's game, which is very unfair to the old timers.

[Killakobe81]

They put up great offensive numbers as well. Also, both of them got great amounts of rebounds and blocks. The funny thing is, the two were more similar than different. KG was a natural SF playing PF, Robinson was a natural PF playing C. Both had horrible teammates, neither had a real go-to offensive move, both were made to be the defensive anchor of a championship team, but can be an extremely effective scorer, both were ahead of their time in the sense that they were these uber-athletic bigs, neither had great wing support throughout their entire careers, both won championships as the pseudo-lead after their peaks.

They put up great numbers across the board.



WS, ORtg, DRtg, and VORP are not really stat padding friendly. They all have their flaws, but I find WARP to be one of the weakest.



It's not just pace, and I think most advanced stats already took pace into account. The biggest issue is that the game has evolved in that the roles of players have changed. Advanced stats are "made" today so it is constructed with assumptions using today's game, which is very unfair to the old timers.

some great answers and As you could tell by the title ...I wasnt sure if it was good stat or crap. It seems you are squarely on the crap side, and since you enjoy metrics I will trust you on this one. You may have been wrong about my team the past few years (amazing that you think you know my team better than me)...but you have a head for stats. I will file this under crap for now ...maybe one of the ESPN number crunchers will change my mind later.

[ambchang]

some great answers and As you could tell by the title ...I wasnt sure if it was good stat or crap. It seems you are squarely on the crap side, and since you enjoy metrics I will trust you on this one. You may have been wrong about my team the past few years (amazing that you think you know my team better than me)...but you have a head for stats. I will file this under crap for now ...maybe one of the ESPN number crunchers will change my mind later.

I don't think it's crap, I think there are flaws in the numbers, as with all the numbers, but taken together, they can really spell out certain things.

Again, the numbers are mostly directionally correct.

And I don't think I know the Lakers more than you do, because I don't know how much you know, but trust me, I know a lot about the Lakers. I loved Magic (as a player).

[Chinook]

I think it's self explanatory, really.. and yeah, win shares suck too. offensive, defensive and net rating are waaaaay better imo.

What? Win-shares are mostly based on net rating. It doesn't make sense to like the latter but not the former.

[Killakobe81]

I don't think it's crap, I think there are flaws in the numbers, as with all the numbers, but taken together, they can really spell out certain things.

Again, the numbers are mostly directionally correct.

And I don't think I know the Lakers more than you do, because I don't know how much you know, but trust me, I know a lot about the Lakers. I loved Magic (as a player).

he was great, the GOAT PG and He and Bird the GOAT SF (pre-Lebron) were both amazing saved the NBa so MJ could move it forward.
So let me ask you this ... could you make a strong case for Lebron over Bird using the WARP metric? At face value it does seem to argue that but as you said ALL of these metrics have flaws ...your thoughts because I know you respect both players game and many on here are so biased one way or another ... love some deep analysis on the subject.

[Raven]

What? Win-shares are mostly based on net rating. It doesn't make sense to like the latter but not the former.

it depends what it is used for, win shares is very brainy and of little practical use for player to player comparison as it uses league average numbers which clearly is strongly dependent on the age in which the player is playing in. correct me if I'm wrong... if what I'm saying is correct, than how can you make a case to use win shares as a metric for players from different ages..

[ambchang]

he was great, the GOAT PG and He and Bird the GOAT SF (pre-Lebron) were both amazing saved the NBa so MJ could move it forward.
So let me ask you this ... could you make a strong case for Lebron over Bird using the WARP metric? At face value it does seem to argue that but as you said ALL of these metrics have flaws ...your thoughts because I know you respect both players game and many on here are so biased one way or another ... love some deep analysis on the subject.

Honestly, if it wasn't for nostalgia, Lebron has already passed Bird. They both have the same heights (really, people forgot how ridiculously dominant Bird was at his absolute peak), but Lebron just peaked for longer. It may have to do with modern medicine, training and time management principles, but Lebron just peaked for longer. I loved Bird, and I am not really that big of a fan of Lebron, but it would be hard for me to say Bird > Lebron, if not impossible, at this point.

That said, Bird is way clutcher than Lebron ever could be, and just carry that sense of arrogance about him that is tough to argue. If I want to build a team around for a decade, I'd pick Lebron, if I want a player to lead me to a championship for 1 or 2 years, it would be prime Bird. If you put the two on the same team, both in their primes, I would have no question that Bird will be the alpha.

Metrics wide, again, it was era dependent. The 80s teams were stacked, so it was tough for one player to really dominate on his own, teams are thinner now, and there is more of a drop from option 1 to option 2 on the same team, so the advanced metrics is tough to compare that way. There is no way you can find a team that had the depth and talent of the 86 Celtics, 80s Lakers, or even the early 80s 76ers. It would be tough to find a team like the late 80s Blazers or even the mid 90s Sonics in today's league. The ones that do are called superteams.

One last note, Bird, as slow and unathletic as he was, was really underrated as a defender. He was so smart, it's almost like he just see things 3 steps ahead of everyone else. We will see another Lebron in our lifetime, we will never see another Bird in our lifetime (Duncan is probably the closest, but he really is his own beast. The other unique player we will never see is Barkley).