Retro Player of the Year Project 1983-84 UPDATE — Larry Bird

[lessthanjake]

To illustrate some of these above-mentioned problems with DTO/EDTO analysis that OhayoKD has tried to proliferate here, I ran this analysis on every possession Michael Jordan played in Game 6 of the 1989 ECF. The full game can be found here:

Watch on YouTube

.

This was a game in the short time period where Jordan was playing PG, and was before the Bulls used the triangle. And, while Pippen definitely wasn’t a primary ball-handler at that time, I note that Pippen went out very early in the game. The Bulls’ game plan here was very heliocentric, centered around getting the ball to Jordan and having him create.

So how much did Jordan create? Well, by my tally, in 86 offensive possessions, Jordan racked up 102 DTOs, and 60 EDTOs. Per 100 possessions that comes out to the following (my notes on this are in a spoiler at the bottom of this post):[/spoiler]

Great post and analysis. I also try to count his DTOs and EDTOs this game after seeing how much you got for him. And by my way of tracking like I did to every other player I do - I got MJ having 69DTOs and 34EDTOs - it really great.

First it's great to see more people get involved in tracking and checking film.

I will say, though, that is an unusually large disparity. Tsherkin had me checking his tracking possession by possession and we disagreed on one edto . When I looked over Lebronny's creations for a game from 2006 I counted 4 less edtos (39 vs 35). When we looked at a game from ewing in 90, i found 3 less. The biggest gap happened when Rubber dubber and elpollo tracked the same jordan game seeing a gap of 10.

But here we have a gap of 26. Might be worth checking the tape to see where the extra edtos are coming from. A possesion that stands out to me here is possession 2:

2. Jordan drives and pulls 4 defenders and passes for what would be an open three, even with a 4v5 with Pippen injured. They win jump ball. Jordan draws two defenders and passes by two more defenders to give Grant a layup. 8 DTOs; 6 EDTOs.

I asked elpollo and he told me he got 6 dtos and 4 edtos. Looking at this closer...

2. Jordan drives and pulls 4 defenders and passes for what would be an open three

Jordan does pull 4 defenders (and 3 extra ones). But does he take out all 4? Well one defender ends up guarding one of jordan's teammates. Is Jordan really the one taking him out here? I would say no as it would be wierd to just leave cartwright open. That said, on this matter elpollo ultimately sided with you

When the drive in semi transition that have 4 defender around him right.

I counted it as 4 DTOs and 3 EDTOs

But I also could count as 2 EDTOs and 1 ADAs because it can be argue the that the 4th defender was on bill Cartwright at the post near Jordan drive

Where I think you lost both of us was this part

They win jump ball. Jordan draws two defenders and passes by two more defenders to give Grant a layup.

.
Laimbeer gives Grant space because he has to worry about Cartwright(and Jordan gets some credit for faking there) and still ends up challenging Grant's shot. That seems pretty weak for "renders a player unable to affect near or entirely single-handedly". Edwards is just rotating out to cover Jordan's teammate. And I think here we figure out where the gap comes from:

I think our first possession before the jump ball are the same

But the difference is after jumpball

I think Jordan only pull 2 defender on him. 1 was in the paint 1 was defending Bill at Post and last one was rotating on number 54

The defender was already tight on Bill even before MJ move off to receive the pass.

I do think if you count players drawn by other attackers as "dtos" it would make sense you'd get higher counts there than most people doing it.

I think this all really goes to the high subjectivity of the exercise. I think my tally for this game was reasonable. At the same time, I think one could reasonably have a very different assessment using the same criteria (and I’m sure elpollo’s very different tally was reasonable too). You give some examples of differences. Those are plays that I think one could reasonably assess differently than I did. And the problem is that it comes down to an assessment of things like who a defender is being drawn to, whether a defender is really “out of the play,” etc. These are not even close to mechanical assessments, and in some cases even require knowledge we cannot possibly actually have (i.e. such as assessing who a defender is focused on—which, in plenty of cases, is something we’d need to literally be that person to really know for sure).

You often say that normal box stats like assists also are subjective, but there’s just way more guardrails cabining in whatever subjectivity is present there. There’s pretty clear criteria for what an assist is. You must make the pass before a shot is made. On its face, that is a clear bright line rule that is determinative in the vast majority of cases. So usually, absolutely everyone would come to the same conclusion about a given play. The only subjectivity is in the grey area where a pass is made and the receiver of the pass does a lot before scoring. People can differ in their tally in those marginal cases, but it’s still a pretty two-dimensional assessment. For most people, it’s basically just going to come down to how many dribbles the guy took before scoring. So there’s usually effectively zero subjectivity at all, and sometimes there’s basically a two-dimensional assessment that is pretty easy to be consistent with. The assessment of DTO/EDTOs is way more multi-faceted than that could ever be, and involves way more subjectivity and way more scope for inconsistency. It requires assessments of everyone’s positioning on the floor, a judgment of who defenders are paying attention to, an analysis of whether someone is “out of the play” or not, etc. It’s just a way more holistic exercise involving way more factors and way more judgment calls, which inherently makes it way more subjective. This is just definitely a way bigger problem here than it is for normal counting stats.

None of this of course refutes your main point here: that dtos/edtos(and adas) can scale up the more chances you are given to create. However, marking this as flaw for dtos/edtos as a "creation proxy" doesn't really follow. If a player is creating more in a role where they are set up to create more, you should still report that they are creating more, and thus far you haven't really offered a meaningful distinction between "was in a situation to dto/edto more" and "was in a situation to create more."

I think you’re not quite getting what my point is. There is a definite distinction between how much a player is creating in raw individual terms and what effect they’re having on how much their team is creating. The latter is what we should care about, but the former is what this analysis is geared towards.

If a player’s role is set up for them to create more, the flip side of that is that their teammates’ roles are set up to create less. A player in a more egalitarian system will create less but their teammates will be put in positions to create more. And the opposite is true in a more heliocentric system. What should matter most here is what effect a player is having in the context of the system/role they’re in. It will probably often be the case that the guy with more impact in this regard doesn’t actually have higher raw numbers.

For instance, to use a simplistic stylized example, let’s say Player A plays in a very heliocentric system, and Player B plays in a very egalitarian system. Player A gets 20 DTOs, and Player B gets 10 DTOs. Player A is the better creator, right? Well, not necessarily. What if the more heliocentric system means that Player A’s teammates can only be expected to get 5 DTOs, while the system on Player B’s team means that his teammates can be expected to get 16 DTOs? Suddenly, we’d expect Player A’s team to only have 25 DTOs and Player B’s team to have 26 DTOs. Player B’s team is better off (assuming DTOs are actually a good proxy in the first place, which is itself quite debatable, for reasons I’ve mentioned several posts above this). In that scenario, it’s likely that Player B is having better impact on creation, despite the lower number of individual DTOs (though it’s also possible that Player B’s team just has a system that creates more in general—which is another blind spot I’ve noted with your DTO/EDTO analysis). Or at least, Player B is able to raise their team to higher levels of creation within the teams’ chosen systems—which is the most important thing in a sport where the goal is to be the very best team. It may still be the case that if you took Player A and Player B off their teams, Player A’s team would have their creation fall off more—in fact, we may well expect that from taking away a player that the system is more focused on, since way more adjustment would be needed from the team when playing a system that was set up for a now-absent guy to create almost everything. But if Player B working within a more egalitarian system will get better overall team results than Player A working in a more heliocentric system, then their individual DTO/EDTO numbers aren’t really relevant. Player B is raising the ceiling of their team more, at least in this particular regard.

To be fair, this is an issue with assist stats too. But, as I mentioned in my earlier post, having done some of this analysis, I think the effect of system is bigger in terms of DTO/EDTOs than it is on assists. This is probably largely because of the outsized effect of plays with multiple DTOs/EDTOs, as well as the effect of counting stuff where someone decides not to even shoot (which both result in a scope for pretty exponential levels of tallies for the guy asked to do all the creation, even for single possessions, in a way that is outsized compared to the impact on those possessions). A guy asked to create everything while his teammates provide more static spacing will naturally just churn out numbers from that stuff to a larger degree than they’d churn out assists (and I think to an outsized degree compared to the actual impact they’re having—I wrote some about this several posts above as well). You have seen this yourself and have concluded that this must mean that assists are actually underestimating creation for heliocentric guys. But I think it’s really just that DTO/EDTO analysis is unduly biased in favor of heliocentrism—in other words, it is overcounting creation for those guys.

[eminence]

Did a tracking study once on which offensive players defenders were tracking with their eyes - really hard to take results from other than that Steph destroyed the field. Depending on the parameters 2nd place was sometimes closer to last (~400th) than they were to 1st.

[OhayoKD]

Great post and analysis. I also try to count his DTOs and EDTOs this game after seeing how much you got for him. And by my way of tracking like I did to every other player I do - I got MJ having 69DTOs and 34EDTOs - it really great.

First it's great to see more people get involved in tracking and checking film.

I will say, though, that is an unusually large disparity. Tsherkin had me checking his tracking possession by possession and we disagreed on one edto . When I looked over Lebronny's creations for a game from 2006 I counted 4 less edtos (39 vs 35). When we looked at a game from ewing in 90, i found 3 less. The biggest gap happened when Rubber dubber and elpollo tracked the same jordan game seeing a gap of 10.

But here we have a gap of 26. Might be worth checking the tape to see where the extra edtos are coming from. A possesion that stands out to me here is possession 2:

2. Jordan drives and pulls 4 defenders and passes for what would be an open three, even with a 4v5 with Pippen injured. They win jump ball. Jordan draws two defenders and passes by two more defenders to give Grant a layup. 8 DTOs; 6 EDTOs.

I asked elpollo and he told me he got 6 dtos and 4 edtos. Looking at this closer...

2. Jordan drives and pulls 4 defenders and passes for what would be an open three

Jordan does pull 4 defenders (and 3 extra ones). But does he take out all 4? Well one defender ends up guarding one of jordan's teammates. Is Jordan really the one taking him out here? I would say no as it would be wierd to just leave cartwright open. That said, on this matter elpollo ultimately sided with you

When the drive in semi transition that have 4 defender around him right.

I counted it as 4 DTOs and 3 EDTOs

But I also could count as 2 EDTOs and 1 ADAs because it can be argue the that the 4th defender was on bill Cartwright at the post near Jordan drive

Where I think you lost both of us was this part

They win jump ball. Jordan draws two defenders and passes by two more defenders to give Grant a layup.

.
Laimbeer gives Grant space because he has to worry about Cartwright(and Jordan gets some credit for faking there) and still ends up challenging Grant's shot. That seems pretty weak for "renders a player unable to affect near or entirely single-handedly". Edwards is just rotating out to cover Jordan's teammate. And I think here we figure out where the gap comes from:

I think our first possession before the jump ball are the same

But the difference is after jumpball

I think Jordan only pull 2 defender on him. 1 was in the paint 1 was defending Bill at Post and last one was rotating on number 54

The defender was already tight on Bill even before MJ move off to receive the pass.

I do think if you count players drawn by other attackers as "dtos" it would make sense you'd get higher counts there than most people doing it.

I think this all really goes to the high subjectivity of the exercise. I think my tally for this game was reasonable. At the same time, I think one could reasonably have a very different assessment using the same criteria (and I’m sure elpollo’s very different tally was reasonable too). You give some examples of differences. Those are plays that I think one could reasonably assess differently than I did. And the problem is that it comes down to an assessment of things like who a defender is being drawn to, whether a defender is really “out of the play,” etc. These are not even close to mechanical assessments, and in some cases even require knowledge we cannot possibly actually have (i.e. such as assessing who a defender is focused on—which, in plenty of cases, is something we’d need to literally be that person to really know for sure).

Meh. I and I imagine almost anyone looking at the existing definition of dto side by side would see "single-handedly or near single-handedly" and pretty instantly rule out players who are already in rotation and running in a different direction towards a different attacker. I agree there's a range of interpretation, but when the gap between two tracking is nearly triple anything that's been compared before, I'd question if the criteria being used is really the same.

You often say that normal box stats like assists also are subjective, but there’s just way more guardrails cabining in whatever subjectivity is present there. There’s pretty clear criteria for what an assist is. You must make the pass before a shot is made. On its face, that is a clear bright line rule that is determinative in the vast majority of cases.

Sure. It also means assists will ignore the vast majority of possessions and the majority of what occurs during a possession. So here's the trade-off. DTOs, EDTOS, and ADAs can be generated every possession and any phase of said possession. We can just throw the majority of that way, or we live with having A count that usually differs by a few dtos/edtos (and until now had never differed by more than 10). I see the latter as worth dealing with to avoid the former.

None of this of course refutes your main point here: that dtos/edtos(and adas) can scale up the more chances you are given to create. However, marking this as flaw for dtos/edtos as a "creation proxy" doesn't really follow. If a player is creating more in a role where they are set up to create more, you should still report that they are creating more, and thus far you haven't really offered a meaningful distinction between "was in a situation to dto/edto more" and "was in a situation to create more."

I think you’re not quite getting what my point is. There is a definite distinction between how much a player is creating in raw individual terms and what effect they’re having on how much their team is creating. The latter is what we should care about, but the former is what this analysis is geared towards.

Recording the former is pretty useful in deriving the latter. Also part of the point of specifically tracking EDTOS is generating a bunch of those is not something most players can reliably do.

If a player’s role is set up for them to create more, the flip side of that is that their teammates’ roles are set up to create less. A player in a more egalitarian system will create less but their teammates will be put in positions to create more. And the opposite is true in a more heliocentric system. What should matter most here is what effect a player is having in the context of the system/role they’re in. It will probably often be the case that the guy with more impact in this regard doesn’t actually have higher raw numbers.

This isn't really any different than arguing "if a player's role sets them up to score more, the flip side of that is their teammates' role is to score less". Of course what's left out in both statements is that teams will generally try to make sure their most effective creators/scorers are the ones being put in situations to create/scorers the most and the capability of said creator/scorer will play a large role in dictating how much opportunity they're given.

That said, to the extent players are "gaming" dtos for reasons outside of goodness, there are a couple better ways to adjust for that than throwing out a(nearly) all-inclusive raw creation proxy.

1. Look at how players raw creation fluctuates in different roles (ala how much does Lebron's raw creation drop in Miami/the 2007 Cavs, how much does Jordan's drop in the triangle?)

2. Track creation relative to opportunity to create which is what elpollo has started doing.

Both seem more fruitful to me than worrying about simplistic stylized thought exercises. And I would say all the dto-watchers to an extent are already baking this into how we evaluate playmakers.

You have seen this yourself and have concluded that this must mean that assists are actually underestimating creation for heliocentric guys. But I think it’s really just that DTO/EDTO analysis is unduly biased in favor of heliocentrism—in other words, it is overcounting creation for those guys

I would say that assumption of yours becomes unlikely if we find the "heliocentric guys" are far outstripping their assist production when paired next to other high-volume playmakers or when they're not used as predominant primary initiators.

And thus far, they (and by they, I mostly mean Lebron) are.

[B-Mitch 30]

I talked about this earlier with Ohayo, but maybe his analysis could be improved if he applied it to teams with a heavy emphasis on distributing the ball equally, like the Beautiful Game Spurs.

[lessthanjake]

Meh. I and I imagine almost anyone looking at the existing definition of dto side by side would see "single-handedly or near single-handedly" and pretty instantly rule out players who are already in rotation and running in a different direction towards a different attacker. I agree there's a range of interpretation, but when the gap between two tracking is nearly triple anything that's been compared before, I'd question if the criteria being used is really the same.

Again this is all so subjective. When is someone “already in rotation”? It’s not always so clear. And if someone is running towards a different attacker but is in no man’s land, then surely they’ve been taken out of the play. How do we assess whether that’s what’s happened? At what point in the rotation is that no longer the case? If someone is scrambling in rotation such that they’re easy to sidestep or drive by, then are they not taken out because the attacking player did something in reaction to their close out, or are they taken out because they were rendered ineffective? And where do we draw the line there? It just is undeniably extremely subjective, because it’s a completely holistic analysis.

And I will note that the above stuff goes to the effect of eras on this—earlier eras will have fewer DTOs and EDTOs, and the magnitude of the effect of that will also depend on subjectivity. For instance, let’s say you take a more strict view of those scrambling-in-rotation plays than I do. In other words, you say that if the defender scrambles back to their man, they aren’t taken out of the play, even if they’re easily sidestepped or driven by. And let’s say I take the opposite view of those situations. Well, because of that difference, you’d be systematically less charitable to past eras (relative to more recent eras) than I would. Why is that? Well, with less spacing in earlier eras, a higher percent of the time someone draws a defender towards them that defender will at least be able to scramble back towards the guy who receives the ball, because there’s just less distance to cover to do so. So that subjective difference in how we assess that kind of situation would lead to a systematic difference in how players of different eras shake out. It’s all just kind of a mess of issues, where there’s almost nothing objective about it and decisions will have disparate effects on different players.

You often say that normal box stats like assists also are subjective, but there’s just way more guardrails cabining in whatever subjectivity is present there. There’s pretty clear criteria for what an assist is. You must make the pass before a shot is made. On its face, that is a clear bright line rule that is determinative in the vast majority of cases.

Sure. It also means assists will ignore the vast majority of possessions and the majority of what occurs during a possession. So here's the trade-off. DTOs, EDTOS, and ADAs can be generated every possession and any phase of said possession. We can just throw the majority of that way, or we live with having A count that usually differs by a few dtos/edtos (and until now had never differed by more than 10). I see the latter as worth dealing with to avoid the former.

I’m not saying the assist stat is perfect either. It shares some (but not all) of the issues that your analysis has. But it is absolutely substantially less subjective. In the past, you’ve suggested otherwise, but that seems obviously wrong to me, especially after taking the time to actually do one of these DTO/EDTO tracking exercises.

Anyways, I’m not saying it is completely useless. But I think the huge flaws need to be acknowledged. It is analysis that is absolutely biased against (or in favor of) certain eras, systems, and roles. And it is also very subjective, and being used to draw concrete conclusions despite the analysis being in extremely tiny samples. I think if one wanted to properly contextualize it with the eras, systems, and roles it comes in, and not actually draw concrete conclusions about players from it based on tiny samples, then it could provide for interesting case studies. But you want to use it for much more than that, and it’s not really reasonably fit for that broader purpose.

None of this of course refutes your main point here: that dtos/edtos(and adas) can scale up the more chances you are given to create. However, marking this as flaw for dtos/edtos as a "creation proxy" doesn't really follow. If a player is creating more in a role where they are set up to create more, you should still report that they are creating more, and thus far you haven't really offered a meaningful distinction between "was in a situation to dto/edto more" and "was in a situation to create more."

I think you’re not quite getting what my point is. There is a definite distinction between how much a player is creating in raw individual terms and what effect they’re having on how much their team is creating. The latter is what we should care about, but the former is what this analysis is geared towards.

Recording the former is pretty useful in deriving the latter. Also part of the point of specifically tracking EDTOS is generating a bunch of those is not something most players can reliably do.

Recording the former is potentially useful in deriving the latter, but only really in similar contexts. Comparing the DTO/EDTO tallies of similarly heliocentric stars to each other might actually potentially be useful in determining who is having a bigger effect on how much their team is creating (assuming the heliocentric roles genuinely are similar). But comparing DTO/EDTO tallies of stars in very different systems/roles is just not very probative of a player’s effect on things.

As for most players not being able to reliably generate EDTOs specifically, I think that’s generally true. But that’s why more egalitarian systems often don’t focus on doing that. Those systems focus much more on things like off-ball screens that briefly create quick openings, off-ball movement that gets quick separation with a defender or forces a mismatch, and lots of ball movement or off-ball movement that require a ton of coordination to properly defend and therefore creates openings through forcing defensive mistakes. Guys who aren’t stars actually *can* reliably do those things, so more egalitarian systems focus on that stuff much more. And it can be very effective.

And this goes to another bias in your analysis, which I’ve mentioned before. Some systems will also just systematically produce fewer DTOs/EDTOs, and this is particularly true of EDTOs. Systems that rely on the types of actions I mentioned above aren’t designed to get tons of EDTOs. They’re designed to create openings in other ways, which can be exploited by pretty much everyone on the court just with some understanding of what the team is running, where the ball should go, and where people should be moving. Some of these things may still end up tallied as a DTO (depending on the person tallying, I imagine), but players in those systems won’t rack up many EDTOs, because the system is creating the openings, rather than the openings being created by a combination of court spacing and a guy beating his man. Crucially, it’s not that teams just run more egalitarian systems if they are worse. Some of the best teams ever have run egalitarian systems (and vice versa is true as well). And bad teams have run both. One is not better than the other. But one of them certainly will rack up more numbers in your analysis, and that’s indicative of the analysis itself simply being biased in favor of players in certain systems. This Jordan analysis is a good example of that. Jordan unsurprisingly racks up dramatically more EDTOs when played in a heliocentric system for a little bit, compared to when he played in something like the triangle. And I’m confident that the entire Bulls team as a whole would rack up fewer EDTOs in the triangle. But the triangle wasn’t a worse system. It just created offense in a way that gets credited much less in your analysis.

If a player’s role is set up for them to create more, the flip side of that is that their teammates’ roles are set up to create less. A player in a more egalitarian system will create less but their teammates will be put in positions to create more. And the opposite is true in a more heliocentric system. What should matter most here is what effect a player is having in the context of the system/role they’re in. It will probably often be the case that the guy with more impact in this regard doesn’t actually have higher raw numbers.

This isn't really any different than arguing "if a player's role sets them up to score more, the flip side of that is their teammates' role is to score less". Of course what's left out in both statements is that teams will generally try to make sure their most effective creators/scorers are the ones being put in situations to create/scorers the most and the capability of said creator/scorer will play a large role in dictating how much opportunity they're given.

It’s more about comparative advantage and maximizing the overall result with the personnel you have. Let’s assume two teams have the same superstar on their team, and that that superstar is the best on their team at basically everything. Team A has a supporting cast that isn’t very good at shooting or at beating their man, but has lots of players that are good at smartly reading and reacting to situations. You’ll probably end up wanting them to play an egalitarian system with lots of off-ball player movement and ball movement, because those players will move the ball in smart ways to create openings. Meanwhile, Team B has a supporting cast of guys that are good at shooting, but not as smart at reading and reacting. You’ll probably want a heliocentric system with that team, because that limits the reads the supporting cast will need to make, while leveraging their shooting ability. We could come up with a ton of different permutations of this that would lead to different conclusions as to what is the right system. The calculus does of course get even more complicated if you change who the superstar is and what their strengths are and how they prefer to play. If the superstar on Team A is better at beating their man with the ball than they are at anything else, or even if they just prefer to play that way, then Team A may possibly be better off playing a heliocentric system even though it doesn’t best fit the rest of the team. So yeah, the capability of a superstar definitely has an effect on how a team plays. But so does the capability of the rest of the team, as does the specific willingness/preferences of the superstar (which is a distinct concept from the superstar’s capability, but is also very important). The preferences of coaches also matters. There’s a lot that goes into these decisions. The bottom line, though, is that your DTO/EDTO analysis will systematically give better results to players based on what those decisions end up being.

That said, to the extent players are "gaming" dtos for reasons outside of goodness, there are a couple better ways to adjust for that than throwing out a(nearly) all-inclusive raw creation proxy.

1. Look at how players raw creation fluctuates in different roles (ala how much does Lebron's raw creation drop in Miami/the 2007 Cavs, how much does Jordan's drop in the triangle?)

2. Track creation relative to opportunity to create which is what elpollo has started doing.

Both seem more fruitful to me than worrying about simplistic stylized thought exercises. And I would say all the dto-watchers to an extent are already baking this into how we evaluate playmakers.

As to #1 above, this doesn’t really necessarily follow at all. The raw “creation” will fluctuate in different roles more or less based on how different the role is. LeBron’s raw “creation” between 2007 Cavs and Miami would surely fall less than Jordan’s would fall from this game and the triangle, because the difference in Jordan’s roles is way bigger.

As for trying to “track creation relative to opportunity to create,” that’s theoretically a way to adjust for this issue. But the usefulness is really just hypothetical. I think the way this would actually be applied is just to discount possessions where a guy doesn’t get the ball. But that doesn’t really properly adjust for the issue, because it doesn’t adjust for the fact that the opportunity to get DTOs/EDTOs is different in different systems even assuming a guy has the ball. It is relatively easy to create DTOs/EDTOs if you get the ball and your teammates are providing static spacing for you to drive, or giving you a spread pick and roll. It is much harder to create DTOs/EDTOs if your teammates are occupying space in the middle with off-ball screens and/or off-ball movement. So even holding the number of involved possessions constant, you’d expect a lot more DTO/EDTOs in a heliocentric system. Which, for instance, means that we’d still definitely expect more DTO/EDTOs from a star in a your-turn-my-turn heliocentric situation than one in a more egalitarian system. A star in that situation may not get the ball every play, but when they do get the ball the offense is set up in a way that maximizes generation of DTOs/EDTOs. If there were some way to quantify how to adjust for system/role, then that’d be great, but I don’t think there is. Which just leaves the exercise being inherently biased in favor of players with certain roles/systems.

You have seen this yourself and have concluded that this must mean that assists are actually underestimating creation for heliocentric guys. But I think it’s really just that DTO/EDTO analysis is unduly biased in favor of heliocentrism—in other words, it is overcounting creation for those guys

I would say that assumption of yours becomes unlikely if we find the "heliocentric guys" are far outstripping their assist production when paired next to other high-volume playmakers or when they're not used as predominant primary initiators.

And thus far, they (and by they, I mostly mean Lebron) are.

See above on why we’d expect a your-turn-my-turn heliocentric system to result in more DTOs/EDTOs for a star than a more egalitarian system. If the system still inherently involves having the off-ball guys doing things that maximize the on-ball guy’s ability to get DTOs/EDTOs, then they’ll get more of them, even if sometimes another star is the one in that position in the offense. The fact that the system itself is much more conducive to DTOs/EDTOs is a massive factor (not to mention that even in your-turn-my-turn situations, the stars still end up with the ball more than guys in more egalitarian systems).

[Elpolo_14]

See above on why we’d expect a your-turn-my-turn heliocentric system to result in more DTOs/EDTOs for a star than a more egalitarian system. If the system still inherently involves having the off-ball guys doing things that maximize the on-ball guy’s ability to get DTOs/EDTOs, then they’ll get more of them, even if sometimes another star is the one in that position in the offense. The fact that the system itself is much more conducive to DTOs/EDTOs is a massive factor (not to mention that even in your-turn-my-turn situations, the stars still end up with the ball more than guys in more egalitarian systems).

I think this mislead. 07 Cavs very far your turn my turn. Lebron not getting opportunity much. DTO probably underrate not over. Lebron creates alot relative to how little his opportunities onball action are.

[OhayoKD]

I talked about this earlier with Ohayo, but maybe his analysis could be improved if he applied it to teams with a heavy emphasis on distributing the ball equally, like the Beautiful Game Spurs.

Open to doing 2014 duncan (creation) after I do another game of 2003 duncan (defensively).