Comments on: Adjusting for Position Played

By: John

John — Mon, 16 Jul 2007 16:02:56 +0000

I think a lot of you are missing the point. The point is not that Rashard Lewis is a “bad” power forward. The point is about rational economic activity. If we consider a team to be a rational economic entity, it should apportion it’s resources (salaries) to maximize it’s production (wins).
The point is ultimately that Rashard Lewis is significantly above average for a small forward, but only average for a power forward. Essentially what he is trying to say is that Orlando would maximize their investment in Rashard Lewis if they would simply pay a league-average salary to a league-average power forward to play alongside him. Is his ability to play league-average power foward useful? Certainly. There will be times when he will be forced to play the position and it’s better to play him there than someone who’s below average, but to count on that as part of your plan is silly. The only reason any of this matters is that the Magic have tied up a significant amount of their resources in Lewis, they cannot afford for him to give them average production, they need for him to give above-average production. The answer to doing so is relatively simple, pay an average player to rebound and hit the occasional shot at the power forward position, and allow Lewis to dominate small forwards.

By: Westy

Westy — Thu, 12 Jul 2007 15:27:33 +0000

Good posts. The issue of position adjustment is certainly the seemingly glaring reason Wins Produced is experiencing some hesitation in acceptance, JChan. As Rasta indicates Lewis’ own stats (rebounding) changed based on the position he was changing. While his aggregate totals were better than an average SF, his total while actually playing SF was less than average.
The issue seems to be that while PAWS evaluates each player against his own position the basis is Win Score, and with higher scores at PF and C, Berri seems to be arguing that those positions contribute more to wins.
On the basis of their Win Scores alone, fielding a team consisting of Garnett, Duncan, Howard, Lee, and Boozer would win the NBA championship. Lacking any guards or small forwards, that seems highly unlikely, though. Thus, it would seem some level of matching the lineup to conventional positions is appropriate. But could you get by with a PG and 4 of the 5 above? Who’s to say which positions are required to be successful. Would a team with the conventional positions beat one ‘going big’ or ‘going small’? Would a player’s Wins Produced remain consistent if he was on a team of all PF’s and C’s? It seems a team consisting of the league leaders in PAWS would beat a team of only those leading in WS. How consistent with the traditional positions would the PAWS team need to be, though?
There is reluctance to accept the fact that PF’s and C’s are more important to a team’s success, and thus the cry of overweighting rebounds (which accounts for their greater Win Scores). Obviously these positions get more rebounds based only on their position on the floor. Does this mean they are more important to their team’s success?

By: Tball

Tball — Tue, 10 Jul 2007 20:02:13 +0000

A baseball parallel would be moving Piazza to DH/1B or Nomar to first base. They may be able to put up the numbers to get by at the new position, but Piazza and an average DH/1B contribute more to a lineup than Piazza and an average catcher. If we assume the forward playing along side Lewis is average, would you rather have Lewis playing an average PF role along side an average SF or Lewis playing a superior SF role along side an average PF?

I do think Lewis’s ability to play PF should be awarded some value, not unlike Hines Ward’s ability to be an emergency QB for Pittsburg or Troy Brown being able to fill in at cornerback – it provides situational flexibility, but it is a bad sign if you need to go to that well regularly.

By: Rasta

Rasta — Tue, 10 Jul 2007 20:02:07 +0000

I’ve got several comments to make, but not much time. I apologize in advance if I’m unable to make my point clear.

I understand (and generally agree with) the positional adjustment to determine a player’s marginal value versus an “average player”.

In this case, Dave Berri has taken Lewis’ total stats, and compared them to the “average SF” and “average PF”. Looking solely at rebounds, Lewis’ 8.1 rbs/48 is higher than the average SF (7.6 rbs/48), but lower than the average PF (11.4 rbs/48).

The problem with approach is that Lewis’ 8.1 rbs/48 is really the the combination of all the rebounds he grabbed while splitting time between SF and PF. By the way, using 82games data, it appears that Lewis played 1771 minutes at SF, and 576 minutes at PF.

So how did his rebounding stats look while playing those positions?

At SF, Lewis grabbed 7.2 rbs/48. At PF, he grabbed 10.3 rbs/48. This paints a somewhat different picture when comparing to the “average” players.

However, why stop there? Why should we compare Lewis’ stats against an average NBA SF or PF when we have data that shows how he performed against the player he was actually guarding?

This gets into an entirely new issue which I’d prefer to save for another day. Anyway, for comparison, here are the rbs/48 for Lewis, his man, and Dave’s average player:

At small forward:
Lewis 7.2
His man 5.7
Average 7.6

At power forward:
Lewis 10.3
His man 10.5
Average 11.4

At you can see, Lewis out-rebounds his man at SF, while holding his own versus PFs.

Based on this comparison, I don’t think it’s fair to say that Lewis’ rebounding (or perceived lack thereof) hurts his team.

By: Neil the Ethical Werewolf

Neil the Ethical Werewolf — Tue, 10 Jul 2007 17:58:34 +0000

I’ve read Yglesias’ response to this, and it seems fairly convincing. A good small forward should get credit for being average rather than terrible at other positions, because there will be situations (injuries, fatigue) when the alternative to moving Lewis to PF will be playing someone terrible at PF. Maybe you also have a fairly good SF to fill in for Lewis. Then Lewis’ average PF skills actually give you a better collection of players on the floor.

By: MT

MT — Tue, 10 Jul 2007 15:12:25 +0000

Working off a box score makes granular analysis hard. I would expect that individual teams do this kind of analysis all the time, just as they should track FG% based on shot selection, or whether a player scores going right or left.
(That reminds me of a funny story in Pistol that illustrates the limits of analysis of athletes based on past performance: an opposing coach noted before his first game against Maravich that Maravich was always going right and scoring. So he told his team, force him to his left. They did, all night, and Maravich went left all night. For 55 points.)

One modeling idea I’ll throw out is to create 2 or 3 tiers at each position based on minutes played at the position. KG might play 36 minutes at the 4. Leon Powe might play 2. Rather than evaluating all the other 4’s based on an average of a range that includes both of them, if they were tiered into 2 or 3 groups, it would enable full time 4’s to be measured against other full time 4’s, backup 4’s to be measured against backup 4’s, and perhaps allow for players to be gauged at each of their multiple positions. Also, tiering would address the (admittedly minor) issue of outliers who get garbage minutes.

By: Col Bat Guano

Col Bat Guano — Tue, 10 Jul 2007 03:46:25 +0000

He still isn’t going to play any PF with Orlando no matter how many times you say it. I don’t even know where you got statistics for him playing PF for Seattle since he rarely did it and only when playing small ball.

By: JChan

JChan — Mon, 09 Jul 2007 22:55:01 +0000

Great comments all around, in the post and the following comments. I don’t really have anything significant to add to that discussion, but I think the position adjustments are something that need to be explored more and more. It seems like most arguments I’ve seen against the WOW methodology attack the position adjustment as a weak spot.

Is there another way to present this so that people who have just discovered WOW don’t get turned off right away by the “magic numbers” of the position adjustment? And is there an easy way to decide on adjustments for multi-position players? It seems like until these questions have quick simple answers, then a good portion of the public will not take the time to see the wisdom in the method.

Or maybe do we accept that not everyone will agree with the WOW method and not worry about it? I’m rambling, but the position adjustment has always seemed to me to be the sticking point for wide acceptance.

By: TK

TK — Mon, 09 Jul 2007 22:43:44 +0000

More agreement with MT and Joe. Being flexible enough to fill in occasionally at another slot seems like a big benefit, both in terms of WinScore and anecdotally.

In terms of WinScore, it can mean avoiding big minutes for a really substandard backup PF by shifting Lewis over and putting a decent SF backup in the game.

But I have to say, it works much better — both anecdotally and in terms of WinScore — when the flexibility is to move down a position instead of up. As discussed a few weeks ago, Magic Johnson being able to run the point at 6′9″ means that he gets far more boards than a typical point. And Scottie Pippen working as a Point Forward means that the Bulls were never forced to give major minutes to a really substandard backup point guard. (If memory serves, Darrell Walker was essentially the unused backup point guard on one of those early championship teams, averaging basically zero minutes a game.) Instead, Pippen runs the show and a better backup SG or SF comes in. That *downward* flexibility is huge!

By: Jason

Jason — Mon, 09 Jul 2007 22:27:56 +0000

I’ve struggled with the notion of positions and their importance. What’s clear is that certain tasks need to be accomplished and some players accomplish more than others. Certainly the scoreboard doesn’t care if you’re a 7 foot center or a 6′ point guard scoring a bucket and it doesn’t care what hypothetical position Rashard Lewis will be considered have played when he scored a bucket. But WP cares, and it seems it cares because it assumes that all positions are equal and there’s an expected average output for each position that is equal in terms of its contributions to a win.

I think that this axiom needs to be examined a bit as it does not need to be true. Certainly the distribution of heights suggests that there are fewer potential 4’s and 5’s and thus it may be that substandard athletes will get more minutes in these positions. If true, the standard deviation on performance at these positions should be greater and some teams will have a harder time getting the assumed average output and may do better going with another strategy.

Most or all of this compensation seems to come from Lewis himself. The comparison of his averages for the season don’t tell this story, but again, if the 82games.com stats are to believed, in the minutes that he played PF, he was a better scorer, better rebounder, and took fewer shots (though more FT attempts) to get these points. They provide per 48 averages for all of the Win Score stats by position he played. (I’ll exclude center because the number of minutes he played there are unlikely to be statistically significant.) By my quick (read: possibly errant) calculations, his per48 Win Score as a ‘3′ was 8.8, while as a ‘4′ he produced 12.8 per 48. On a per minute, this means he was at .183 and .267 at the two positions. If the averages of .152 and .215 are to be used for calculating the average production, his production, relative to the average was actually better as a ‘4′ than as a ‘3′. Since the limited supply of potentially adequate (to say nothing of above average) 4s should be lower than the same for 3’s, if Lewis can improve relative to the average when he slides over, it’s probable that a replacement closer to the average 3 can be found easier than a replacement 4. Given that the 82games stats suggest that the Sonics did better in terms of net points while he played the 4, it further suggests that this may be the case. It may be that the Sonics picked their moments to put him at the 4 when the matchups favored him being a better rebounder, or something else was favorable, but no matter, it looks, from the statistical breakdown, that Lewis at the 4 *can* be a more successful strategy given the opportunity for him to improve his personal stats in this capacity and the likelihood of a better sub coming in at the 3 than at the 4.

What’s clear is that the stats for the team need to be there. What’s not clear is that it matters who gets them, though wins produced does care when it comes to crediting the success of a team to its individual players.