Player Win-Loss Records. W: # games this season where Rockets outscores opponent while player is on the court. L: # games this season where opponent outscores the Rockets while player is on the court. Code: [B]Player W L WL%[/B] Carlos Delfino 9 4 0.692 Toney Douglas 12 9 0.571 Omer Asik 11 9 0.550 Chandler Parsons 10 11 0.476 Patrick Patterson 10 11 0.476 James Harden 9 11 0.450 Marcus Morris 9 11 0.450 Greg Smith 7 10 0.412 Jeremy Lin 7 14 0.333 Daequan Cook 3 7 0.300 Terrence Jones 2 5 0.286 Cole Aldrich 3 8 0.273 Donatas Motiejunas 1 3 0.250 [B][I]TEAM 11 11 0.500[/I][/B]
these stats aren't very meaningful especially after 20 games. Better stats to look at over the long run is to see the scoring efficiencies of other players while a point guard is on the floor. Take out the defense which depends so much on the opposing individual players.
Thanks for the stats durvasa! Appreciate it. It's just further proving what most of us think that Harden + Douglas is more effective than Harden + Lin at the moment. Hopefully the coach will do something about those plays... because it's doubtful Douglas will be around next season..
Is there any two player combo comparisons? I'd really like to see Lin + Smith's output, they've been having some highlight plays together so far. :grin: Also, a surprise that Delfino is a positive contribution. We probably need to see a larger sample size based on pure -/+.
carlitos needs a max level contract! all he does is win and drain threes when he is not missing them. Is outscoring the opponent the most accurate way to get the W-Ls? wouldn't traditional PGs get outscored on a nightly basis against the new age point guard, whose mission is to score? I think guys like rubio and rondo will get outscored on most nights but that doesn't mean they suck, even with rubio not being the best of defenders.
Its not looking at individuals outscoring eachother, but rather teams while the player is on the floor. If Rubio's and Rondo's teams are consistently getting outscored when they face scoring PGs, then maybe they're not as good as we think they are.
We need adjusted value, but that needs at least a full season of data to be of use (getting rid of some noise). Raw +/- is pretty much useless imo.
I'd like to see Omar Asiks scoring efficiencies of when TD is on the floor vs JL vs. both vs. neither. Go through each player. Then you can get an understanding of which players work better with whom. Grouping all of it together packs too many other variables. Even what I am suggesting is imperfect, but I feel it would be a lot more insightful than +/- particularly for a PG.
Who cares how each player does. The goal is not to pad individual stats, but to outscore the other team.
The link provided in the first post does give stat breakdowns of teammates with Lin on the floor and off the floor, if that's what you're interested in. But I don't really understand how what you're suggesting doesn't pack in an equal set of variables. Who's to say Omer Asik's scoring efficiency depends on the PG rather than a host of other things? It may be interesting to break things down into smaller details, but its not necessarily going to provide a clearer picture unless you have an idea of how to put it all together and make sense of it.
*raising hand*....might be a dumb question but... Example...If Jlin works his but off for 3 quarters and help neutralized opponent and/or extend the lead...did not play the 4th Q...his contributions will not be counted?
In a hypothetical situation, Player A is on the court for 2 quarters a game, and while on the court the score is tied at the end, say 343-343. But the team lost every game. Then he has a win% of 50%. It has absolutely nothing to do with what actually happened to the outcome of the games.
One of the worst stats in basketball: Unadjusted Plus/Minus. Case in point from today's Raptors game. Jose Calderon had a triple double today but his +/- is 0. The rest of the starters are all positive. It doesn't matter how you break down Plus/Minus, by quarter, by player, by playing pairings, you will still get quite useless information, not to mention further reducing the sample size. It's like, splitting a bad apple in quarters and all you get is 4 quarters of a bad apple. However, it's not a sample size issue with this stat. It's a flaw in the foundation logic. What people are trying to show with +/- is whether or not a player is contributing to their team at a given moment during the game. But the problem is that there's 9 other players on the floor, plus coaches, plus refs, that act as variables that have tremendous influence over a single player's +/- rating, all of which is constantly changing over the course of the season and even over the course of a single game. In the stats world, that's called variance. The more variance you have, the more useless the stat, especially if you can't account for it. +/- also suffers from something called collinearity, but that's beyond the scope of this response (look it up if you're interested though). What you think you're seeing is a single's player contributions, but what you're actually seeing a that's player contributions plus what his teammates are doing minus whatever the other team is doing. So it's impossible to get individual data from that without further and more complicated analysis. (That something more complicated is Regularized Adjusted Plus/Minus) Don't want to listen to me because I'm a newbie (even though in real life I have a degree in Applied Statistics). This article more or less says the same thing: http://wagesofwins.com/2011/03/05/deconstructing-the-adjusted-plus-minus-model/ Moral of this post: Stop quoting raw +/- to make an argument about players.
This would be a flaw in interpretation. Just as I would be interested to know my team's win/loss record or margin of victory to get a sense of their overall level of play, I would similarly want to know how the team is playing when certain players are on the floor. Yes, there's a lot of factors that go into it. The more analysis and more in-depth you want to delve into it, the stronger the opinions you can form from it. And note that in the original post, I did not say that the +/- of a player is a direct measure of his contribution. It is, rather, a description of how the team played while that player was on the floor. It may suggest chemistry for certain player combinations, or that some players are simply "better" individually, or maybe its just random chance. The point is that, regardless of why a team has played better with certain players, a coach will often factor in the successes and failures of lineups when choosing who to play and when. Therefore, if we want to understand the substitution patterns which seem to run counter to what we'd expect, looking at +/- may provide a clue.
I love durvasa threads, always good info and good discussion. Would be excellent if he could write some front page statistics-based articles for Clutchfans. Just a collection of his posts (and related posts of other BBS'ers in this thread) would make a great article IMO.
