Value over Replacement vs Absolute Value

[OhayoKD]

Most of my takes and arguments over here have revolved around the former but since seeing how much bigs can create offensively during the RPOY ontop of what they do defensively I've started wondering if taking the former approach is just punishing better players for happening to play the same position as other players who do more.

Is value over replacement a better approach than ballparking how much a player offers in a vacuum?

[lessthanjake]

I think our assumption should be that bigs do not actually systematically offer more value. If bigs simply “do more” then there’d be every reason to play a bunch of bigs at once. But teams don’t do that. If anything, the trend in the league in the last decade or so has been the opposite. If bigs were systematically more valuable, then you really wouldn’t see teams frequently going for small ball, often to great effect. Teams wouldn’t frequently opt out of increasing the value on the court, nor would it be successful if they did.

I also think the reality is that any intuition that one position is more valuable than another has relatively little practical value. If one player type had more value, then you’d want to stack that player type. As mentioned above, the fact that teams don’t stack bigs suggests they aren’t systematically more valuable. In theory, one could try to argue against that by suggesting there’s diminishing marginal returns to adding more players of the same player type, so one big may be most valuable but a second or third big wouldn’t be. But then that leads to a logical conclusion that what’s relevant is value over replacement, because it would tell us that diminishing returns will naturally lead teams to the status quo of having an assortment of different player types/sizes on the court at once. And if that’s the optimal status quo, then it makes a lot of sense to compare players’ value to others of their position, since, due to diminishing marginal returns, the breakdown of player types teams will put on the court will be pretty static regardless of any assessment of the value provided by any one player type.

[Doctor MJ]

Most of my takes and arguments over here have revolved around the former but since seeing how much bigs can create offensively during the RPOY ontop of what they do defensively I've started wondering if taking the former approach is just punishing better players for happening to play the same position as other players who do more.

Is value over replacement a better approach than ballparking how much a player offers in a vacuum?

So, I'm interested in this topic, but I might be confused as to what you mean when you say "Replacement", and I think this because I think that definition is really problematic one to codify.

Based on what you're saying, it seems like you're notion of "Replacement" is position-based, which is a plausible way to do it, and definitely the way to do it in many sports...

But in basketball I'd prefer a replacement level just based on a certain level of player regardless of position. I just don't think the positions are that meaningful the way the game is played today.

To the actual question of Replacement vs Absolute, to me if you can make a good definition for replacement level, that will lead to a more valuable stat an an Absolute. But the "if" is no trivial thing. It has to be based in objective data but feel still subjectively meaningful to us as basketball fans.

[Doctor MJ]

I think our assumption should be that bigs do not actually systematically offer more value. If bigs simply “do more” then there’d be every reason to play a bunch of bigs at once. But teams don’t do that. If anything, the trend in the league in the last decade or so has been the opposite. If bigs were systematically more valuable, then you really wouldn’t see teams frequently going for small ball, often to great effect. Teams wouldn’t frequently opt out of decreasing the value on the court, nor would it be successful if they did.

A great point to bring up. Modern NBA teams are built around players who were the size of 4s and 5s a century ago, so clearly things just scale up as needed and to the extent that we're not in the midst of a major shift to bigger/small, you could say we're in an equilibrium where things are effectively optimized around lineups of 5 equally necessary sets of features.

I also think the reality is that any intuition that one position is more valuable than another has relatively little practical value. If one player type had more value, then you’d want to stack that player type. As mentioned above, the fact that teams don’t stack bigs suggests they aren’t systematically more valuable. In theory, one could try to argue against that by suggesting there’s diminishing marginal returns to adding more players of the same player type, so one big may be most valuable but a second or third big wouldn’t be. But then that leads to a logical conclusion that what’s relevant is value over replacement, because it would tell us that diminishing returns will naturally lead teams to the status quo of having an assortment of different player types/sizes on the court at once. And if that’s the optimal status quo, then it makes a lot of sense to compare players’ value to others of their position, since, due to diminishing marginal returns, the breakdown of player types teams will put on the court will be pretty static regardless of any assessment of the value provided by any one player type.

So the tricky part here is that it's not really about Positions so much as ball dominance and other such constraints. All-Star games are always utterly dominated by players who have the ball in their hands a lot, and frankly it makes for bad fit.

As such, I'm not fundamentally opposed to a VORP design that takes into account player role, I just don't think Position is good enough proxy to role to justify it, and I really think to account for player role properly is no small thing.

[rrravenred]

I suppose one approach is to look at the trends of what "bigs" offer preferentially to other player types, try to break that down functionally, then see:

* to what degrees those functions correlate with wins
* to what degree those functions can be provided by non-bigs (although you'd have to somewhat compensate for outliers)

We don't generally see single-archetype stacked teams, which suggests some form of diminishing returns in operation. There are only so many balls on the court at once, and only so much space within the arc.

[penbeast0]

I think our assumption should be that bigs do not actually systematically offer more value. If bigs simply “do more” then there’d be every reason to play a bunch of bigs at once. But teams don’t do that. If anything, the trend in the league in the last decade or so has been the opposite. If bigs were systematically more valuable, then you really wouldn’t see teams frequently going for small ball, often to great effect. Teams wouldn’t frequently opt out of increasing the value on the court, nor would it be successful if they did.

I also think the reality is that any intuition that one position is more valuable than another has relatively little practical value. If one player type had more value, then you’d want to stack that player type. As mentioned above, the fact that teams don’t stack bigs suggests they aren’t systematically more valuable. In theory, one could try to argue against that by suggesting there’s diminishing marginal returns to adding more players of the same player type, so one big may be most valuable but a second or third big wouldn’t be. But then that leads to a logical conclusion that what’s relevant is value over replacement, because it would tell us that diminishing returns will naturally lead teams to the status quo of having an assortment of different player types/sizes on the court at once. And if that’s the optimal status quo, then it makes a lot of sense to compare players’ value to others of their position, since, due to diminishing marginal returns, the breakdown of player types teams will put on the court will be pretty static regardless of any assessment of the value provided by any one player type.

I think traditionally good/great bigs have had a greater VORP and a greater value to teams partially because:

(a) before the move to spamming 3's, shots closer to the basket were the most efficient shots and big men had more effect on those at both ends of the court. This has obviously diminished greatly since the pace and space era really took hold.

(b) but also, there are a lot more 6' to 6'4 sized players in the human body type pool than there are 6'10 and taller players. Thus, when you cream off the top 200 or so top players in each body type, there will logically be more of a spread among the rarer, taller archtype. While players have gotten bigger over time, it is far more true among the 6' to 6'4 types being replaced by more 6'4 to 6'8 sized players than that the tallest 2-3 men on each NBA roster getting taller. There is some of that, thanks to the move to a worldwide player pool offering more choices, but less so than at the positions other than center.