Page 1 of 2 1 2 LastLast
Results 1 to 20 of 33

Thread: Arturo from Wages of Wins released his season predictions - Raps looking good

  1. #1
    Raptors Republic Starter
    Join Date
    Jan 2010
    Posts
    338
    Post Thanks / Like

    Default Arturo from Wages of Wins released his season predictions - Raps looking good

    http://wagesofwins.com/2012/10/09/th...early-preview/

    Wins Produced has its problems with overvaluing certain stats and undervaluing others, but if there's one thing you can say about it: it's a remarkably good model for predicting wins at the team level. I'm feeling better and better about this season.

    (Biggest 'lol, yup' moment of the article: Jose ranks as the 6th most efficient offensive player...but he's 319th on D.)

    Other highlights:

    - Lowry ranks 21st overall, without being in the top 30 in any category. What? It's because he rates pretty good at everything.
    - Landry Fields 27th on defense.
    Last edited by tkfu; Tue Oct 9th, 2012 at 05:59 AM.

  2. #2
    Raptors Republic Starter Quirk's Avatar
    Join Date
    Nov 2009
    Posts
    319
    Post Thanks / Like

    Default

    Quote tkfu wrote: View Post
    Wins Produced has its problems with overvaluing certain stats and undervaluing others, but if there's one thing you can say about it: it's a remarkably good model for predicting wins at the team level.
    Hey tkfu. I agree with you, I really like the WP community since they contribute so much data, tools and analysis, but I also have my issues with WP, though still think it's quite valuable.

    The system is remarkable good at predicting wins at the team level because it is based upon regressions from wins to box score stats, the problems arise when you try to attribute wins to individual players.

    I used to think the problem's with WP where based on a lack of efficiency stats for rebounds in boxscore, causing them to overvalue high-rebound, low-shot-attempts players. But now I think the problems are much bigger than just that.

    The trouble is not all teams play the same, and thus league wide regressions may not explain how a given-team's players win games, and what contributions players make to the wins a given team has. To clarify the picture we need something along the lines of a "Relative Wins Produced," regressions based on a given team, not league wide. IMO, that could help clarify individual performance significantly.

  3. #3
    Raptors Republic Starter
    Join Date
    Jan 2010
    Posts
    338
    Post Thanks / Like

    Default

    I mean, I think the biggest problem is the lack of evaluation of defense. The single greatest contributor to the offensive portion of most players' WP is thier shooting efficiency, which is fine (up to a point, shot creation, blah blah blah, &c.) But on the defensive side, an individual player only gets credit for steals, defensive boards, and blocks. Just like the essence of good offense is taking efficient shots, the essence of good defense is making your opponent take inefficient shots, and WP completely ignores that on the level of the individual player. There is a team adjustment added, but it is simply divided among all players, pro-rated by minutes.

    All that said: on the team level, WP is very good, and there is a higher season-to-season correlation for an individual player's WP than for any other advanced stat I know of. Those two facts mean that WP is the best publicly available tool out there for projecting a team's winning percentage.

    (of course, no stat is infallible, predictions are prone to a fairly wide margin of error, &c.)

  4. #4
    Raptors Republic Rookie
    Join Date
    Jul 2010
    Posts
    193
    Post Thanks / Like

    Default

    Quote tkfu wrote: View Post
    Wins Produced has its problems with overvaluing certain stats and undervaluing others, but if there's one thing you can say about it: it's a remarkably good model for predicting wins at the team level. I'm feeling better and better about this season.
    We have one fatal flaw in our team right now.

    According to WP:
    - we have 5 players in the top 50 in the league (Kyle Lowry, Ed Davis, Jose Calderon, Amir Johnston, Landry Fields)
    - we have the best PG pair in the league (Kyle, Jose)
    - we have one of the best front courts in the league (Ed, Amir, Aaron Gray, likely big Val)
    - we have decent options at the wings (Landry, Chris Wright, Dominic McGuire, likely TRoss)

    However, we are likely to play a lot of players 11-15 in our depth chart. Two of them are considered our 'stars'. You know, the two who went a combined 11/29 yesterday against Real Madrid. Fatal. Flaw.

  5. #5
    Raptors Republic Starter
    Join Date
    Oct 2010
    Location
    Virginia
    Posts
    793
    Post Thanks / Like

    Default

    Quote Kuh wrote: View Post
    We have one fatal flaw in our team right now.

    According to WP:
    - we have 5 players in the top 50 in the league (Kyle Lowry, Ed Davis, Jose Calderon, Amir Johnston, Landry Fields)
    - we have the best PG pair in the league (Kyle, Jose)
    - we have one of the best front courts in the league (Ed, Amir, Aaron Gray, likely big Val)
    - we have decent options at the wings (Landry, Chris Wright, Dominic McGuire, likely TRoss)

    However, we are likely to play a lot of players 11-15 in our depth chart. Two of them are considered our 'stars'. You know, the two who went a combined 11/29 yesterday against Real Madrid. Fatal. Flaw.
    I would not put so much faith on a system which ranks Ed Davis (42) higher than James Harden (67), Tony Parker (91), Marc Gasol (119), Russel Westbrook (137), Rudy Gay (157), LaMarcus Aldridge (169), and does not list Kobe Bryant as a top 180 player in the league.

    It fails the eye test, bigggggg time.

  6. #6
    Raptors Republic All-Star Primer's Avatar
    Join Date
    May 2012
    Posts
    1,185
    Post Thanks / Like

    Default

    So the Raps are predicted to be anywhere from a 5-8 seed in the East, with 8 seed being most likely. Sounds about right to me.

  7. #7
    Raptors Republic Starter minks77's Avatar
    Join Date
    Aug 2010
    Location
    T dot Ohhhhh
    Posts
    778
    Post Thanks / Like

    Default

    Quote Kuh wrote: View Post

    However, we are likely to play a lot of players 11-15 in our depth chart. Two of them are considered our 'stars'. You know, the two who went a combined 11/29 yesterday against Real Madrid. Fatal. Flaw.
    Is what you are saying here that AB and DD are in the bottom quarter of our players on a wp scale/depth chart as determined by this system?

    The fatal flaw in this system is it obviously isnt worth shit at predicting what an individual player is worth to the team. Does anyone really believe that Ed or Amir are better players than Andrea? Or that running wright out there for 35min is going to win you more games than DeMar?

  8. #8
    Raptors Republic Hall of Famer mcHAPPY's Avatar
    Join Date
    Jul 2010
    Posts
    19,196
    Post Thanks / Like

    Default

    Highlight of the link:

    The Nets and Raptors are one injury away from contending (Apologies in advance to Bargniani and Lopez)

  9. #9
    Raptors Republic Starter
    Join Date
    Oct 2009
    Posts
    289
    Post Thanks / Like

    Default

    Quote minks77 wrote: View Post
    Is what you are saying here that AB and DD are in the bottom quarter of our players on a wp scale/depth chart as determined by this system?

    The fatal flaw in this system is it obviously isnt worth shit at predicting what an individual player is worth to the team. Does anyone really believe that Ed or Amir are better players than Andrea? Or that running wright out there for 35min is going to win you more games than DeMar?
    Well, yeah, the wins produced guys believe it. It's kind of amazing: in my view a classic case of 'scientism' in action. But they genuinely believe that Ed Davis (and Reggie evans) are amongst the best players in the league. For one thing, WP makes no allowance for the effect players have on their teammate shooting percentages, which are often quite large. So guys like Evans and Davis, who gum up the whole offense because nobody needs to cover them unless they're standing under the rim, don't get penalized for their lack of offensive versatility.

  10. #10
    Raptors Republic Rookie
    Join Date
    Jul 2010
    Posts
    193
    Post Thanks / Like

    Default

    Quote minks77 wrote: View Post
    Is what you are saying here that AB and DD are in the bottom quarter of our players on a wp scale/depth chart as determined by this system?

    The fatal flaw in this system is it obviously isnt worth shit at predicting what an individual player is worth to the team. Does anyone really believe that Ed or Amir are better players than Andrea? Or that running wright out there for 35min is going to win you more games than DeMar?
    Strangely, this system is good at matching individual stats to team wins. VERY good.
    Sure, I believe that Ed and Amir are better players than Andrea.
    I don't know enough about Wright to comment. I would have an open mind.

    One thing Wins Produced does is penalise players for missing shots. I think that's fair. If both teams rebound, steal, etc the same, both have the same number of shots each game. Miss more than half of yours and your team loses. AB and DD miss a high percentage of their shots.

  11. #11
    Raptors Republic Starter
    Join Date
    Jan 2010
    Posts
    338
    Post Thanks / Like

    Default

    Quote malefax wrote: View Post
    Well, yeah, the wins produced guys believe it. It's kind of amazing: in my view a classic case of 'scientism' in action. But they genuinely believe that Ed Davis (and Reggie evans) are amongst the best players in the league. For one thing, WP makes no allowance for the effect players have on their teammate shooting percentages, which are often quite large. So guys like Evans and Davis, who gum up the whole offense because nobody needs to cover them unless they're standing under the rim, don't get penalized for their lack of offensive versatility.
    Yeah, that's definitely a problem. The thing is, it's really hard to address with numbers. The on-off data is much too noisy to be of much use--take a look at adjusted plus-minus and you'll find some pretty ridiculous results too, except that APM is also wildly inconsistent year-to-year.

    I think that there's some really significant value in WP because of its year-to-year consistency. That at least tells you that you're measuring a real effect. To go beyond that you have to apply real-world basketball knowledge. The challenge is in being able to honestly synthesize and use the information, rather than just pick out the pieces of data that support your conclusion from each source. As Feynman said, "The first principle is that you must not fool yourself, and you are the easiest person to fool."

  12. #12
    Raptors Republic Starter minks77's Avatar
    Join Date
    Aug 2010
    Location
    T dot Ohhhhh
    Posts
    778
    Post Thanks / Like

    Default

    I don't know enough about WP/WS to comment but you can clearly see the difference in a player like Harden over Ed so this system is obviously flawed. It's great that it penalizes for missed shots but does it take usage rates and sample size into account? I have no clue where it is they are coming up with these ranks or where it is they are going so laughably wrong but they are. They may believe in their numbers but Casey and others have consistently spoken of Andrea as their best player, bar none. I'm no Andrea fan but he is the best option when you need a bucket from this squad.

    The final predictions it makes may be close (horseshoes and hand grenades) but the individual ranks are ludicrous. I love Amir, up until he fouls a guy while jogging across midcourt. ED is a boarding machine, but he can't hit bunnies in an empty gym. Where is the adjustment for boeheadedness?

    Thanks to injury we've actually seen lineups featuring some of these higher ranked players on the court for extended minutes and the results are ugly ball, anemic offence, mediocre defence and loses with a capital "L".
    LET'S GO RAP-TORS!!!!!

  13. #13
    Raptors Republic All-Star ebrian's Avatar
    Join Date
    Nov 2009
    Location
    Markham
    Posts
    1,535
    Post Thanks / Like

    Default

    I'm a big fan of Wins Produced and Win Probabilities but this is the first time I've seen them so inaccurate from an individual standpoint.

  14. #14
    Raptors Republic Starter
    Join Date
    Oct 2010
    Location
    Virginia
    Posts
    793
    Post Thanks / Like

    Default

    Quote Kuh wrote: View Post
    Strangely, this system is good at matching individual stats to team wins. VERY good.
    Sure, I believe that Ed and Amir are better players than Andrea.
    I don't know enough about Wright to comment. I would have an open mind.
    What we know about Chris Wright:

    24 years old
    Was undrafted
    Split last season between GS and D-league
    Played less than 200 minutes in the NBA last season and did not impress GS enough to be offered a contract.

    Yup, this guy COULD BE better than Andrea Bargnani!!

    While on the subject of keeping an open mind, maybe you should give some thought about coach Casey being able to recognize basketball talent better than you and me.

  15. #15
    Raptors Republic All-Star Soft Euro's Avatar
    Join Date
    Jun 2011
    Location
    Utrecht, The Netherlands (Yes, that's Europe!)
    Posts
    1,920
    Post Thanks / Like

    Default

    Quote Kuh wrote: View Post
    Strangely, this system is good at matching individual stats to team wins. VERY good.
    Sure, I believe that Ed and Amir are better players than Andrea.
    I don't know enough about Wright to comment. I would have an open mind.

    One thing Wins Produced does is penalise players for missing shots. I think that's fair. If both teams rebound, steal, etc the same, both have the same number of shots each game. Miss more than half of yours and your team loses. AB and DD miss a high percentage of their shots.
    You got any proof for that?

  16. #16
    Raptors Republic Starter
    Join Date
    Jan 2010
    Posts
    338
    Post Thanks / Like

    Default

    Quote Soft Euro wrote: View Post
    You got any proof for that?
    It's necessary, by dint of the assumptions made in constructing the model. It's normalized to point differential. The question that you have to ask is to what degree you can consider a player responsible for his own stats. For example, does Reggie Evans get a lot of rebounds by taking away boards his teammates would have gotten anyway? Does Bargnani shoot a low percentage because he's forced to take bad shots, or because he chooses to take bad shots? Conversely, does Tyson Chandler keep his shooting percentage so high by forcing Carmelo to take tough shots at the end of the clock?

    The argument in favour of the WP model is that players' per-minute production is remarkably consistent season to season, whether they're on good teams or bad. But that's a general argument; you can always make the case that specific players are exceptions for specific reasons. Additionally, you can argue against the position adjustment they employ and point out that it doesn't make a lot of sense to have a different standard of shooting efficiency for centers than you have for shooting guards if you're also going to make the case that shot creation isn't valuable. (Incidentally, their counter-argument to this criticism is that it's a model of real-world behaviour and results, and it's reasonable to expect it to break down if you extend it ad absurdum, which I find to be at least somewhat reasonable.) The position adjustment also causes problems evaluating players who are used in unusual ways, like Dirk, Bruce Bowen, or Andrea.


  17. #17
    Raptors Republic Starter Fuchan's Avatar
    Join Date
    Jun 2011
    Posts
    239
    Post Thanks / Like

    Default

    Epic post tkfu

  18. #18
    Raptors Republic Starter Quirk's Avatar
    Join Date
    Nov 2009
    Posts
    319
    Post Thanks / Like

    Default

    nice post, tkfu

    the longest thread I ever endured on WP is this one:

    http://thenbehteam.blogspot.de/2011/...on-review.html

  19. #19
    Raptors Republic All-Star Craiger's Avatar
    Join Date
    Jun 2012
    Posts
    1,119
    Post Thanks / Like

    Default

    I'm not sure if this is entirely the place for this or if many will care. But I've long questioned how statistics (as a whole) quantify a 'possession'.

    This is the commonly used equation:

    FGA + (0.47)FTA + TO - OREB = possession used (Where FGA and FTA = made + missed attempts)

    So this seems relatively easy (with the value for FTs comming from studying the number of 'and1' + shooting fouls on 3 pt attempts. ie. Sometimes a player can take 1, 2, or 3 FTs in a possession but only the last FT taken counts as the end of a possession). Basically what happens is any time the ball changes hands, a possession ends for the offense (used) and a new one begins for the defense (gained).

    But where I have a problem is with the theory behind this, which makes me question the -OREB part of the equation. Why does the ball actually have to change hands for a possession to end? Why can't a possession end when a team (ie. the offense) gives up its current control of the ball instead?

    If we look at the rules of the NBA we see something that might help
    - a team gets 24 seconds to use when it gets possession of the ball
    - an offensive foul occurs when the offending team has possession of the ball. Offensive fouls never result in FTs for the defensive team.

    When we look at what happens after any shot is taken we see:
    - whatever team controls the ball after the shot gets a new 24 sec shot clock. This happens after a shot is made, and on either an offensive or defensive rebound (the only exception being if the ball doesn't hit the rim)
    - a foul on a rebound attempt is considered a lose ball foul and can result in FTs for whatever team the action was committed against. (ie. a foul on any rebound is not an offensive foul)

    When we look at that, I think a fair argument can be made that a possession ends (is used) when:
    - a turnover is committed
    - a shot goes in
    - a shot hits the rim

    Which then in turn would mean a possession is gained when
    - the defense obtains the turnover
    - defense allows the a shot to be made
    - any rebound (offensive or defensive) is made*

    *an exception here could ofcourse be on an airball, but these are rather rare events and would have minimal statistical impact. Although an average number of airballs could be caluculated and that value could be included in the equation

    This ofcourse would significantly change the equation for a possession:

    FGA + (0.47)FTA + TO = possession used

    There are a number of impacts here
    1) more possessions occur in a game than is assumed (because OREB are new possessions) and therefore the pace teams play at will be significantly different
    2) teams can have a different number of possessions in a game (not all teams will have an equal number of OREB)
    3) the value for a rebound could be significantly different than it is currently valued in any statistic, missed rebounds could be considered (any offensive rebound is a missed rebound by the defense, and any defensive rebound is a missed rebound by the offense)
    4) the above would all have a significant impact on the any individual players value.


    I'm not saying with certainty that this should be the case, and there are many more intelligent people than myself who have come to the conclusion that the currently used equation is the right one. But this seems a much more logical way to me to view a possession (a possession ends when a team gives up control, and NOT necessarily when the ball changes hands between teams). Ofcourse this also means any non +/- stat has an error in its equation, which ofcoures means they'd also have to be redone right from their very initial possessional equations (which I doubt many would like to do) and how they value certain actions (missed shots, defensive rebounds, offensive rebounds) may be mistaken.

    Any thoughts?
    Last edited by Craiger; Thu Oct 11th, 2012 at 09:24 AM.

  20. #20
    Raptors Republic Starter
    Join Date
    Jan 2010
    Posts
    338
    Post Thanks / Like

    Default

    Yeah. The reason possessions are defined like that is basically because it's more useful to think that way. As you said, if we defined possessions the way you suggest, teams would have different numbers of possessions each game. That would make per-possession efficiency numbers useless until you added in the possessions-per-game stat.

    It's usually more revealing, when you're looking at team-level stats, to have a measure for overall efficiency. You can then break that down into its component pieces if you want to drill down and see why the efficiency is what it is. So you see your team has an overall offensive efficiency of, say, 100.0 pts per 100 possession. That's a good measure for comparing how good your offense is overall compared to other teams. But now you want to see what is helping you and what's hurting you, so you check what percentage of possessions end in turnovers, or assists. If you want to know about rebounding, you check to see what percentage of available rebounds your team got, breaking it down into off/def if you want. If you want to know about scoring efficiency, you look at TS%.

    Regarding your four points above:

    1 & 2:

    Well yeah, it would be different. But it wouldn't be better, because all you've changed is removing offensive rebounds. You'd have a stat called "possessions" whose only function would be to tell you about rebound differential. We already have that stat, it's called rebound differential. We also already have a stat for scoring efficiency, TS%. And if, for some reason, you really wanted a stat for offensive or defensive efficiency that included turnovers and shooting efficiency but not rebounding, well, the math is easy enough.

    Another side-effect of your proposed change would be that any time you wanted to look at per-possession numbers as a measure of efficiency, which is what we use per-possession numbers for most of the time, you'd have to adjust them for rebound rate. It just doesn't make sense.

    3 & 4:

    No, how a possession is defined on the team level wouldn't have any effect on how an individual player is valued. Any box score-based stat that attempts to measure a player's overall efficiency already chooses a weight to assign to offensive and defensive rebounds. Those valuations differ between the different stats, but they all have their reasons (well, except PER, which seems to be pulled straight out of Hollinger's ass).
    Last edited by tkfu; Thu Oct 11th, 2012 at 10:39 AM.

Page 1 of 2 1 2 LastLast

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •