Jump to content


Toggle shoutbox Shoutbox Open the Shoutbox in a popup

@  timetodienow... : (16 August 2013 - 10:35 PM) Has anyone noticed that Morey is more comfortable typing than speaking?
@  2016Champions : (16 August 2013 - 08:53 PM) Daryl Morey's reddit AMA today: http://www.reddit.co...eason_addition/
@  Sir Thursday : (16 August 2013 - 06:54 PM) (Although having said that, Cleveland and Washington being improved mean that there might actually be some competition towards the bottom of the Eastern Conference playoff bracket for once).
@  Sir Thursday : (16 August 2013 - 06:53 PM) They may end up with the 8th seed again this year, but they'll be a team on the up rather than a team that's reached its peak.
@  Sir Thursday : (16 August 2013 - 06:52 PM) That's a lot of shooting, anyhow. And they've filled the bench with serviceable players, too.
@  Sir Thursday : (16 August 2013 - 06:49 PM) (or maybe you put Gary Neal in for Delfino)
@  Sir Thursday : (16 August 2013 - 06:48 PM) It's not a roster completely devoid of talent, anyway. Knight/Mayo/Delfino/Ilyasova/Sanders seems like a reasonable starting lineup.
@  Sir Thursday : (16 August 2013 - 06:45 PM) A lot depends on if they can develop their young guys. Knight, Henson, Antetokounmpo...
@  timetodienow... : (16 August 2013 - 06:41 PM) Although this signing does make them look better to me.
@  timetodienow... : (16 August 2013 - 06:40 PM) Maybe. I don't really care all that much, but it just seems like he's being underpaid. And why would he want to stay in MIL? Have you seen the moves they've made thus far.
@  Sir Thursday : (16 August 2013 - 06:36 PM) A player with 7 years' experience qualifies for a bigger max contract, right? Maybe he's angling for that
@  timetodienow... : (16 August 2013 - 06:33 PM) I hope he has an ETO. That would make the deal make sense in my mind. Otherwise why not play next year and get offered the max by a desperate team and have MIL match?
@  timetodienow... : (16 August 2013 - 06:32 PM) Then why not get a 5th year?
@  Sir Thursday : (16 August 2013 - 06:31 PM) And if Sanders doesn't mind Milwaukee then he would probably want as long a contract as possible...maximise guaranteed money and all that.
@  Sir Thursday : (16 August 2013 - 06:30 PM) Milwaukee probably have a bit more leverage than most teams do in contract negotiations in that [a] it's an extension rather than a resign, so they could just let him to to RFA; and [b] they have a really deep front court rotation.
@  timetodienow... : (16 August 2013 - 06:29 PM) True about Pek, but I don't consider them elite rim protectors like in the vein of Duncan/KG/Asik/Howard/Gasol/Noah/Hibbert(showed improvement last year). Drummond and Sanders are potential candidates for rim protectors in my opinion.
@  timetodienow... : (16 August 2013 - 06:26 PM) Unless he has an opt out clause after two years(he might) this was a dumb deal.
@  timetodienow... : (16 August 2013 - 06:26 PM) If he had to agree to the 11 per to get a 5th year, then I could see it.
@  timetodienow... : (16 August 2013 - 06:26 PM) I agree with you in a vacuum. But with Splitter making 9 mil per and Pek making 12 per and all of the other bigs making money, I don't think so.
@  Sir Thursday : (16 August 2013 - 06:25 PM) Pekovic got $60M/5 years...so a bit more. He's played long enough minutes to be sure it's sustainable though.

Photo

RAPM vs. Win Shares vs. WARP


31 replies to this topic

#1 Rahat Huq

Rahat Huq

    Administrator

  • Administrators
  • 1,199 posts

    Posted 12 August 2013 - 01:34 PM

    Wanted to get the community's take on the respective merits on each of these methodologies.  I also think this would be a good place in which to consolidate ideas for the purpose of educating newer members unfamiliar with the metrics.  I'll probably sticky this at some point if discussion is fruitful.

     

     


    • 0

    #2 timetodienow1234567

    timetodienow1234567

      Advanced Member

    • Members
    • PipPipPip
    • 1,187 posts
    • LocationAlabama

    Posted 12 August 2013 - 01:46 PM

    RAPM -

     

     In "Regularized Adjusted Plus-Minus" (RAPM), the goal is to provide more accurate results by employing a special technique called "ridge regression" (a.k.a. regularization). It significantly reduces standard errors in Adjusted Plus-Minus (APM).

    Conventional adjusted plus-minus is shown to do a poor job of predicting the outcome of future games, particularly when fit on less than one season of data. Adding regularization greatly improves accuracy, and some player ratings change dramatically. The enhancement with the RAPM is a Bayesian techniquein which the data is combined with a priori beliefs regarding reasonable ranges for the parameters in order to produce more accurate models. That is what ridge regression (a.k.a. regularization) does.

     

    Win Shares - 

    A Win Share is worth one-third of a team win. Win Shares are assigned to players based on their offense, defense, and playing time. If a team wins 60 games, there are 180 Win Shares to distribute among the players. This is always true; if a team wins n games, then there are 3n Win Shares to allocate to the players.

     

    Here is a link that explains IN DETAIL about win shares. http://www.basketbal...m/about/ws.html

     

     

    WARP - 

    WARP, a metric which is invented by Kevin Pelton, 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 also draws heavily on the work of 
    Dean Oliver. Using replacement level shows the value of players who 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, WARP does a better job of incorporating defensive value. Learn more about WARP calculation.

     

    Here is an in depth article detailing everything about WARP. http://sonicscentral.com/warp.html


    • 0

    Why so Serious? :D


    #3 Steven

    Steven

      Advanced Member

    • Members
    • PipPipPip
    • 1,132 posts

      Posted 12 August 2013 - 02:46 PM

      So if a team fields 15 replacement players, how many games does it win?
      • 0

      #4 BenQueens

      BenQueens

        Advanced Member

      • Members
      • PipPipPip
      • 47 posts

        Posted 12 August 2013 - 03:06 PM

        So if a team fields 15 replacement players, how many games does it win?

         

        From the SonicsCentral link (emphasis added):

         

        The last step is critical: introducing replacement level to the equation. I define replacement level on a theoretical basis, starting with the theory that a team comprised of replacement-level players would win 10 games. (This figure is borrowed from my work adjusting the late Doug Pappas' Marginal Wins/Marginal Payroll measure to the NBA.) Therefore, I adjust replacement level (using the Goal Seek function in Excel) to match the total marginal wins produced by the entire league above 10 wins per game (adjusted for the percentage of the schedule complete during the season). In practice, this generally produces a replacement-level winning percentage of around .415, or 83 percent of league average. This is similar to the 80 percent figure used by Baseball Prospectus and often used by Football Outsiders in other sports.


        • 0

        #5 2016Champions

        2016Champions

          Advanced Member

        • Members
        • PipPipPip
        • 2,425 posts
        • LocationVirginia, USA.

        Posted 12 August 2013 - 04:43 PM

        Personally, I'm not big on WARP and Win Shares. They don't even pass the smell test. Offensively they're an okay way of looking at a player's stats in one convenient number, but defense is not something that can be accurately measured with blocks, steals, and rebounds. 


        Edited by 2016Champions, 12 August 2013 - 04:49 PM.

        • 0

        Debates are destructive as they present one-sided opinions and demolish reasoned arguments, whereas discussions are constructive and encourage the expression of opinion.

         

        Debate is the death of conversation.


        #6 timetodienow1234567

        timetodienow1234567

          Advanced Member

        • Members
        • PipPipPip
        • 1,187 posts
        • LocationAlabama

        Posted 12 August 2013 - 04:45 PM

        They're all equally useful in forcing us to reevaluate our opinions more thoroughly. None of them are perfect and each has their flaws.
        • 0

        Why so Serious? :D


        #7 Steven

        Steven

          Advanced Member

        • Members
        • PipPipPip
        • 1,132 posts

          Posted 12 August 2013 - 05:37 PM

          Thanks Ben.
          • 0

          #8 blakecouey

          blakecouey

            Red94 Chief of Police

          • Moderators
          • 1,049 posts

            Posted 12 August 2013 - 11:06 PM

            They're all equally useful in forcing us to reevaluate our opinions more thoroughly. None of them are perfect and each has their flaws.

            ^This says it best.  

             

             

            WARP gets a more whole representation of value than the others(IMO), but it is not an end all statistic nor is it perfect.


            • 0
            Have a Twitter? Post your info here and have it added to gain more followers and faster Rockets and Red94 news.
            http://www.red94.net...ic/310-twitter/

            #9 2016Champions

            2016Champions

              Advanced Member

            • Members
            • PipPipPip
            • 2,425 posts
            • LocationVirginia, USA.

            Posted 12 August 2013 - 11:13 PM

            I agree, none of them are perfect and each has their flaws--that should be a given. 


            Edited by 2016Champions, 12 August 2013 - 11:16 PM.

            • 0

            Debates are destructive as they present one-sided opinions and demolish reasoned arguments, whereas discussions are constructive and encourage the expression of opinion.

             

            Debate is the death of conversation.


            #10 timetodienow1234567

            timetodienow1234567

              Advanced Member

            • Members
            • PipPipPip
            • 1,187 posts
            • LocationAlabama

            Posted 12 August 2013 - 11:16 PM

            I agree that none of them are perfect and each has their flaws. I thought that was stating the obvious though. 

             

            Yes we know that you are a RAPM junkie. 


            • 0

            Why so Serious? :D


            #11 2016Champions

            2016Champions

              Advanced Member

            • Members
            • PipPipPip
            • 2,425 posts
            • LocationVirginia, USA.

            Posted 12 August 2013 - 11:17 PM

            Yes we know that you are a RAPM junkie. 

            That response had nothing to do with my post troll! lol jk


            Edited by 2016Champions, 12 August 2013 - 11:18 PM.

            • 0

            Debates are destructive as they present one-sided opinions and demolish reasoned arguments, whereas discussions are constructive and encourage the expression of opinion.

             

            Debate is the death of conversation.


            #12 timetodienow1234567

            timetodienow1234567

              Advanced Member

            • Members
            • PipPipPip
            • 1,187 posts
            • LocationAlabama

            Posted 12 August 2013 - 11:21 PM

            I like RAPM, but I also like looking at Winshares and WARP. I look at efficiency, effective field goal percentage, true shooting percentage, per 36 minute stats. I look at all of them when coming to conclusions(in addition to actually watching the games). Maybe I'll come around on RAPM someday.


            • 0

            Why so Serious? :D


            #13 2016Champions

            2016Champions

              Advanced Member

            • Members
            • PipPipPip
            • 2,425 posts
            • LocationVirginia, USA.

            Posted 12 August 2013 - 11:36 PM

            I like looking at PER and TS% to see how efficient someone is, but so many things can't be measured by stats. I like looking at RAPM to see how much they improve their team offensively and defensively. 


            Edited by 2016Champions, 12 August 2013 - 11:38 PM.

            • 0

            Debates are destructive as they present one-sided opinions and demolish reasoned arguments, whereas discussions are constructive and encourage the expression of opinion.

             

            Debate is the death of conversation.


            #14 Losthief

            Losthief

              Advanced Member

            • Members
            • PipPipPip
            • 65 posts

              Posted 12 August 2013 - 11:37 PM

              I also like looking at units numbers rather than individuals in basketball alot. Different guys play better with other guys sometimes, and it can make a big difference in a sport like basketball. Think an Offensive Line in Football, one that been together/has complimentary guys may work better than the one in the Pro Bowl. I especially do this for defensive, as the league has turned into a defensive league were its 5 guys defending not just one or two with 3 standing around. All the switching, packing the paint, etc etc makes individual defense I think a little defunct, and 5 man unit defensive numbers bigger (except for isolation plays), at least as far as plus-minus goes for me.


              • 0

              #15 2016Champions

              2016Champions

                Advanced Member

              • Members
              • PipPipPip
              • 2,425 posts
              • LocationVirginia, USA.

              Posted 12 August 2013 - 11:43 PM

              Ah, I like looking at unit numbers too. One thing I recently looked into is whether or not the Pelicans are much better offensively when Anthony Davis and Ryan Anderson share the floor together. Unfortunately the sample size is too small, but I'm pretty confident that a Davis-Anderson front court is a match-up nightmare, because it will be tough to defend the Davis-Holiday/Evans pick and roll without giving Anderson some open looks. The defense goes down because Davis is too skinny to play center, but the offense goes up more than the defense goes down which makes it a good trade off. 


              Edited by 2016Champions, 12 August 2013 - 11:45 PM.

              • 0

              Debates are destructive as they present one-sided opinions and demolish reasoned arguments, whereas discussions are constructive and encourage the expression of opinion.

               

              Debate is the death of conversation.


              #16 timetodienow1234567

              timetodienow1234567

                Advanced Member

              • Members
              • PipPipPip
              • 1,187 posts
              • LocationAlabama

              Posted 13 August 2013 - 12:47 AM

              I like looking at lineups +/- as well. I look at those numbers and measure effectiveness. Not necessarily better, but more effective. And if somebody is in multiple numbers of the best lineups then generally that person is usually a good player with that group of guys.


              • 0

              Why so Serious? :D


              #17 timetodienow1234567

              timetodienow1234567

                Advanced Member

              • Members
              • PipPipPip
              • 1,187 posts
              • LocationAlabama

              Posted 13 August 2013 - 12:51 AM

              Also, Parsons is a crap player if you only look at RAPM.


              • 0

              Why so Serious? :D


              #18 Losthief

              Losthief

                Advanced Member

              • Members
              • PipPipPip
              • 65 posts

                Posted 13 August 2013 - 01:20 AM

                I like looking at lineups +/- as well. I look at those numbers and measure effectiveness. Not necessarily better, but more effective. And if somebody is in multiple numbers of the best lineups then generally that person is usually a good player with that group of guys.

                 

                a la shane battier his whole career, his stats don't do much for you...but somehow he ends up in all the most effective lineups for his teams. (going from memphis to houston to miami).

                 

                 

                Also, Parsons is a crap player if you only look at RAPM.

                 

                average (which he was till playoffs)..but i get your point.


                • 0

                #19 RollingWave

                RollingWave

                  Advanced Member

                • Members
                • PipPipPip
                • 282 posts

                  Posted 13 August 2013 - 01:22 AM

                  well TS% is not really an advanced stat, unless you think slugging percentage or OPS in baseball are advanced stats. I think this is a great comparison though, Effective field goal rate = slugging percentage (accounting that 3s are more valuable than 2s, just like a home run is more valuable than a single.) while TS% is essentially OPS because it adds on the free throw shooting part (which is almost exactly like walks in baseball. thus on base percentage.)

                   

                  Yeah, I think Winshare's a flawed, especially on the defensive end. but on offense it's actually very good, but that just kinda goes back to the dilemma that we really don't have defense figured out in basketball. (this by the way, is also true for baseball, but it helps a lot that pitching makes up for most of the defense in baseball so that's easier to assess.)

                   

                  RAPM is interesting, but really must be viewed in a larger sample to be really relevant . I think they need to do more to break down that stats in terms of 3 year average or splits. WARP is really still in the developing stages,

                   

                   

                  it should be noted that in baseball WAR / WARP has really became the mainstream among stats savvy fans, also, the 2nd post's definition is wrong, replacement level player is not average player, a team with 14 average player will have a 41-41 season obviously, replacement level player are guys who are fringe NBA rotation players, guys who you wouldn't go "omg he should totally be out of this league" all the time. (assuming you are not totally biased against him.) . but would keep going that "I wish he only played in garbage time" as the later post pointed out, the current definition of replacement level in basketball, if you have 14 such player , is like a 10-72 win team, aka the 76ers next year.

                   

                   

                  I think WARP in basketball is more problematic though, because in baseball players aren't really influenced as much in their end results by teammates, but in basketball, it is quite obvious the role players' result is heavily dependent on the team they're playing for. this is reflected well in Winshare, where Kevin Martin / Mario Chalmers had a 7 Winshare last year, which is borderline star. but in reality if they were on the Rockets they'd probably be more of a 4-5 win share type player . a true lead star player like Lebron / Harden / Love 's line won't be effected that much by their team (example, Love still had a 10 + WS on bad wolf teams.) unless they have too many stars who are better that forces them down into a more role playing role (aka Chris Bosh.)


                  Edited by RollingWave, 13 August 2013 - 01:25 AM.

                  • 0

                  #20 2016Champions

                  2016Champions

                    Advanced Member

                  • Members
                  • PipPipPip
                  • 2,425 posts
                  • LocationVirginia, USA.

                  Posted 13 August 2013 - 02:02 AM

                  I like looking at lineups +/- as well. I look at those numbers and measure effectiveness. Not necessarily better, but more effective. And if somebody is in multiple numbers of the best lineups then generally that person is usually a good player with that group of guys.


                  This is a very intelligent way to use +/- and Eric Spoelstra said he does the exact same thing.
                  • 0

                  Debates are destructive as they present one-sided opinions and demolish reasoned arguments, whereas discussions are constructive and encourage the expression of opinion.

                   

                  Debate is the death of conversation.





                  0 user(s) are reading this topic

                  0 members, 0 guests, 0 anonymous users