Joined: 05 Jan 2005 Posts: 506 Location: Columbus, OH
Posted: Wed Feb 16, 2005 3:55 pm Post subject: Win Shares
Some of you may have noticed that I changed the player pages on B-R a bit. I added effective field goal percentage, true shooting percentage, usage rate, and Win Shares to the "Advanced" section. Win Shares? Yes, Win Shares. I came up with a (potentially lame) method for calculating Win Shares for basketball players. I posted a description of the Win Shares method on the site. Any feedback you can provide would be appreciated. _________________ Regards,
Justin Kubatko
Basketball Stats!
Joined: 31 Dec 2004 Posts: 23 Location: North Carolina
Posted: Wed Feb 16, 2005 10:45 pm Post subject:
This really hampers players on bad teams and favors players on good teams. You have instances where a player has more win shares than an entire team. That isn't very believable.
This really hampers players on bad teams and favors players on good teams. You have instances where a player has more win shares than an entire team. That isn't very believable.
Doesn't WIN % do the same to some degree? At least in the sense where bad teams with players who use a lot of possesions at a solid but not great efficiency level get penalized in a sense?
When I use WIN% I try to keep some perspective about Usage Rate of a player, and if I am going to compare players, I try to compare players who play a similiar role for one team that is on the same talent plain as the other. I guess there is no systematic reasoning, but in my head I usually compare players in similiar systems, who use similiar amount of possesions, while also factoring quality of the team, and some of my own subjective analysis of players.
BTW Yyzlin how do you know a player has more WIN SHARES than an entire team?
Justin, is there any chance when you have spare time you could apply your stats on the player profiles onto the teams as well? It would be immesnly helpful when looking at team snapshots of the past. I am confused as to how [b]Yyzlin[b] came to the conclusion that certain players on good teams have more win shares than entire BAD teams?
Thanks and great job adding those stats so far! You are doing an awesome job with your website, and I find myself using it more and more with each update! I use to frequent BR.com all the time, now I pretty much only using B-R.com!
And so on; it's apparent that subtracting out replacement value strips something away from the bench warmers. And that defensive credit for Ginobili, Bowen, and Horry gives more WS credits than my system does.
As with Player Wins, my heart and mind agree with the premise and the results. I like it.
Joined: 05 Jan 2005 Posts: 506 Location: Columbus, OH
Posted: Thu Feb 17, 2005 11:33 am Post subject:
Mike G wrote:
I, too, was wondering about the differences between PW and WS. And whether WS could be placed on the team page.
I'll get to the questions later, but I just wanted to let everybody know that Win Shares are now on the team pages as well. _________________ Regards,
Justin Kubatko
Basketball Stats!
Last edited by jkubatko on Thu Feb 17, 2005 3:47 pm; edited 1 time in total
This really hampers players on bad teams and favors players on good teams.
Given that a 25 year old Reggie Miller on a .500 Pacers team has 4 more win shares than Duncan did last season I'm not sure about that.
I do think it benefits people who are the one star on teams with a variety of average teammates. [checks]
Actually I see that's not the case.
Look at san antonio (http://www.basketball-reference.com/teams/SAS/1996.html) in a year when David Robinson averaged 25,12,3, and 3.
Robinson has twice as many win shares as the next best player on the team but that's on a season where he has almost as many net player wins as the next 4 best players on the team put together.
Looking at a couple of teams/players I have to say that the win shares numbers look reasonable.
Joined: 13 Jan 2005 Posts: 202 Location: Iowa City
Posted: Thu Feb 17, 2005 2:25 pm Post subject:
Mike G wrote:
How was it you decided to allott 3 WS per team win? Why not 5? or 1?
As with Player Wins, my heart and mind agree with the premise and the results. I like it.
He's probably following Bill James. That begs the question, but I think that's what it is. Personally, I'd go with straight wins and include a decimal place. I really like the idea of dividing up team wins and like Justin's results generally, but IMHO, it looks like the importance of scoring efficiency is exaggerated in his results.
Justin, please, please, please lose the 3 WS = 1 win thing. It adds needless complexity for zero gain. A win is a win -- if a win share is not directly analogous to a win, lets call it something else. If it is -- and I think it's supposed to be -- lets make it a win. This is just about the silliest idea Bill James ever came up with -- lets just steal his good ideas, and leave the bad ones alone.
Quote:
Marginal points allowed are 1.08 times expected points allowed minus points allowed.
Where did the 1.08 come from? Did I miss it? Was it derived empirically, or theoretically? Is it similar to the 0.52 multiplier in the baseball win shares calculations, which were derived (IIRC) empirically? _________________ ed
Joined: 05 Jan 2005 Posts: 506 Location: Columbus, OH
Posted: Thu Feb 17, 2005 4:02 pm Post subject:
Ed Küpfer wrote:
Where did the 1.08 come from? Did I miss it? Was it derived empirically, or theoretically? Is it similar to the 0.52 multiplier in the baseball win shares calculations, which were derived (IIRC) empirically?
Good question. I figured a team made up of marginal players would win about 10% of their games. So, I solved the following equation for x:
Code:
(1 - x)^14 / ((1 - x)^14 + (1 + x)^14)) = 0.10
The value of x is roughly 0.08, so 1 - x = 0.92 and 1 + x = 1.08. I then tested this to see if marginal points would do a good job of predicting team wins. The relationship between marginal points and wins was:
Joined: 05 Jan 2005 Posts: 506 Location: Columbus, OH
Posted: Thu Feb 17, 2005 4:58 pm Post subject:
Yyzlin wrote:
You have instances where a player has more win shares than an entire team. That isn't very believable.
Why not? The 1993 Dallas Mavericks had just 11 wins, or 33 win shares. Why is it such a stretch to believe that the 1993 Chicago Bulls (57 wins, 171 win shares) had one player (Michale Jordan) with more win shares than the Mavericks? Keep in mind, that doesn't mean that Jordan would have beaten the Mavericks by himself. _________________ Regards,
Justin Kubatko
Basketball Stats!
Joined: 05 Jan 2005 Posts: 506 Location: Columbus, OH
Posted: Thu Feb 17, 2005 5:30 pm Post subject:
Ed Küpfer wrote:
Justin, please, please, please lose the 3 WS = 1 win thing. It adds needless complexity for zero gain. A win is a win -- if a win share is not directly analogous to a win, lets call it something else. If it is -- and I think it's supposed to be -- lets make it a win. This is just about the silliest idea Bill James ever came up with -- lets just steal his good ideas, and leave the bad ones alone.
You know, I didn't like the 3 WS = 1 win approach when James first preseneted it, but it grew on me. Here are two reasons why I like it:
1) It summarizes the player's season value into an integer. It's easier to remember that a player had 25 win shares as opposed to 8.33 wins
2) It makes it easier to see meaningful differences in players. If Ed has 34 win shares and Justin has 32 win shares, I think in most cases it would be safe to say that Ed had the more vaulable season. The difference of 2 win shares makes that fairly easy to see. But if I said Ed had 11.33 wins and Justin had 10.67 wins, I think it's harder to determine if that difference is meaningful.
You might respond "So what?" to both points above. If you really hate the 3 WS = 1 win approach, there's probably not much I can do to change your mind. _________________ Regards,
Justin Kubatko
Basketball Stats!
Joined: 31 Dec 2004 Posts: 23 Location: North Carolina
Posted: Thu Feb 17, 2005 8:40 pm Post subject:
Quote:
Doesn't WIN % do the same to some degree? At least in the sense where bad teams with players who use a lot of possesions at a solid but not great efficiency level get penalized in a sense?
When I use WIN% I try to keep some perspective about Usage Rate of a player, and if I am going to compare players, I try to compare players who play a similiar role for one team that is on the same talent plain as the other. I guess there is no systematic reasoning, but in my head I usually compare players in similiar systems, who use similiar amount of possesions, while also factoring quality of the team, and some of my own subjective analysis of players.
You act as if Win Shares and WIN% are trying to accomplish the same thing, when in actuality, the purposes of both statistics are entirely different.
Quote:
BTW Yyzlin how do you know a player has more WIN SHARES than an entire team?
Just browsing the statistics, you find that league leaders usually end up around 40+ win shares. Thus, any team with 13 or so less wins has less win shares than some individuals.
Joined: 31 Dec 2004 Posts: 23 Location: North Carolina
Posted: Thu Feb 17, 2005 8:45 pm Post subject:
jkubatko wrote:
Yyzlin wrote:
You have instances where a player has more win shares than an entire team. That isn't very believable.
Why not? The 1993 Dallas Mavericks had just 11 wins, or 33 win shares. Why is it such a stretch to believe that the 1993 Chicago Bulls (57 wins, 171 win shares) had one player (Michale Jordan) with more win shares than the Mavericks? Keep in mind, that doesn't mean that Jordan would have beaten the Mavericks by himself.
Michael Jordan, perhaps. But Brad Daugherty, who ended up with 34 WS? No. The point I was trying to make is, they have biases because of the team record. I think Tracy McGrady serves as a good case study here. In 2002-2003, he had 43 WS. In 2003-2004, he had 19 WS. Maybe you do, but I certainly don't think his performance had dropped that much.
I kind of look at them the same. Both kind of give some sort of value to a player.
Theoretically I guess they are different, but I still use them in a similiar context. They are actually quite close if you factor in WIn % and Usage Rate of a player. The best players usually get rewarded in both systems by either having a high # of total wins or high win shares. Of course the team has to WIN to get the benefit of the doubt in the stats as well, but if Tmac can only produce 19 wins, then I guess he should be bashed in the stats. After all, teams with hardly any talent at all have won more games. Just goes to show how Tmac probably dogged way too many games, and put up some empty stats in 0304.
I have to look into Win Shares more, but by looking at different players, I like the way the numbers stack up for some reason.
As an aside, Horace is still better than Scottie when you factor Win Shares. Pippen has more player wins in most seasons, but is less efficient, but uses many more possesions. Unfortnely the Win Shares method doesn't really give Pippen enough credit, but this is a rare example that appears as NOISE. Safe to say that Horace was just an ultra efficient role player, as good as it gets. A true championship role player who just does not makes mistakes and produces on both ends efficiently.
Pippen's intagibles probably do not show as much in the stats, but by 90-91 I would say Pippen was better than Horice. I guess in 89-90 you could make the case Horace was slightly more valuable and efficient without Scotties extra usage rate being a big factor at that point.
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
