APBRmetrics

I would assume Zach got those on and off stats from
https://www.pbpstats.com/wowy/nba

And I do agree with the premise: if adjusting for shot quality of 3 pointers (who, where, and how much contested), departure from that baseline has been repeatedly down to be almost 100% random noise.

Any studies trying to find signal in that 3pt defense for specific players? I have seen none.

Please post links of best studies finding nothing but noise.

To say it is all noise is deny that any 3 point defense could have any meaningful impact for any player.

On average, maybe. But we were talking about for a specific player.

If no player has negative impact on 3pt shot defense then none have positive impact? So what then about 2pt shot defense? Where are the studies?

It flies in face of reason to think that no player (not a single one, ever??) can have meaningful impact on one aspect of the game but presumably can in every other aspect.

Anybody what to do a RAPM based study of 3pt shot defense? Assuming it does not exist in public, it is needed.

Any with access to the detailed location data and the ability to analyze it compared LeBron positioning and degree of contest on shooters, "his man" and in general compared to average to consider direct and indirect impacts? Going to conclude on this case before either is done and presented and studied? I'm not.

Average time between catch and shot release would be another indication of degree of pressure. Or that compared to shooter's average. Study the issue to the nth degree instead of walking away after a few checks, perhaps finding what they might want to find.

Or study the entirety of 3pt shot related detailed movement data. What 5-10 specific movements & timing by defender and shooter most change observed 3pt results, 3ptas and fg% (and blocks and rebounding)? Then is it casualty? New data (now or soon) with position of arms / hands?

How "real" is individual block data? If blocks are real, couldn't meaningful 3pt deterence above or below average be real? Why not?

How many are saying case closed, don't want or need any more analysis?

Certainly three-point defense has a real component. That should be reflected in the shot quality metrics, which account for location and how tightly guarded a specific player is. Beyond that, the vast majority of any residual will be noise.

Location and degree of contest are "sufficient" when all other considerations have been fully studied and properly and entirely rejected.

Soooooo did anyone else see the guy making this thread that his stat showed Jokic’s 30-20-20 game was the 5th best game THAT NIGHT and then double down on it

DSMok1 wrote: ↑Thu Mar 06, 2025 11:49 am I would assume Zach got those on and off stats from
https://www.pbpstats.com/wowy/nba

And I do agree with the premise: if adjusting for shot quality of 3 pointers (who, where, and how much contested), departure from that baseline has been repeatedly down to be almost 100% random noise.

I think there’s a question on how much the quality of the three can be measured, but I agree a lot of it likely is noise

On 3pt 'luck' style adjustments - very useful at smaller samples imo, more middling at larger samples - it seems to be more prone to having the same players consistently show up as 'lucky' or 'unlucky' than something like FT luck adjustment. This to me would suggest it's capturing something that isn't luck in the adjustment (Jokic is one I remember pretty consistently looking 'unlucky' by opponent 3pt% in the near past, haven't looked in a bit).

TeemoTeejay wrote: ↑Sun Mar 09, 2025 5:09 pm Soooooo did anyone else see the guy making this thread that his stat showed Jokic’s 30-20-20 game was the 5th best game THAT NIGHT and then double down on it

It looks like about the 25th best game of the season so far.
Without the overtime, he was just 26-18-17.
Is 40 points (24-16) in an OT a record?

Mike G wrote: ↑Sun Mar 09, 2025 7:18 pmj

I’m not against the idea that it being an OT game might have inflated the totals a tad compared to what you’d expect, but the dude said Jalen Williams having 16-6-6 was the best performance that night because okc win by a lot or something it was very funny

The role of raw +/- in metric scores, especially in one game measurements, can be heavily influential. Deserved or maybe not.

Single game observations are interesting for chat but prefer to focus on season level and broader.

Both raw +/- and RAPM should be known and considered. RAPM generally more important, more reliable but not always reliable. Consider everything.

Unless there are hugely impacting non box score factors in individual cases, generally speaking I think that one thing that’s nice about box score composites is they tend to do well for very small samples when rapm components in it even with the context of a prior and stuff can be a bit stinky

Within that context though they should be good within a game to game basis as well, which is why something like the fmetric presented in the first post which is nearly a purely box score based thing that was trained on essentially overfitting a player ranking board, isn’t great imo, having Jokic’s 30-20-20 game being less impressive than the raw totals because it was an OT game, and having it be the 5th best game of the night behind a 16-6-6 game in a blowout, are very different things imo

Without wading in to all this discussion of luck and other such things, I feel the need to point out that the initial post in this thread smacks of a failure to understand what RAPM is. Statements such as the idea that RAPM is regularised in order to predict out of sample RAPM are entirely incorrect on their face. RAPM is regularised to best predict out of sample stint data using cross-validation, which isn't the underlying player RAPM values, it is the actual results of the stints. Anybody with even a cursory knowledge and understanding of the space understands this fact trivially.

Now, I am not a huge RAPM fan, I have made that pretty obvious. I believe it is overused as the dependent variable in order to try to come up with AIO metrics.

Now, what does have merit but isn't how I'd do it, is this idea of a player metric which is pointing at something else, whether that be a player ranking or otherwise.

I posited an idea along these lines around a decade ago. I got pretty roundly ignored, but the crux of the idea was simple:

Use teammate-marginal player per100 box score stats to predict the teammate-marginal percentage of game time a player is allocated by a coach.

Why is this maybe a good idea? Well coaches and their staff inarguably see more of what their players are, than we ever will. Coaches must allocate minutes in an attempt to maximise team success, and so this marginal allocation of time would seem to be very likely to correlate with actual ability, or at least contextual impact.

Now, there are some issues here: what about tanking teams? What about poorly managed teams? What about the fact that the box score is missing a lot of the game?

Well, for that last part, raw plus-minus is available for use within the teammate-marginal metric, and in the minds of a coach and their staff it is likely a better refined and understood thing, even if only in the abstract, than it is on its own sat there in the box score, propping up players riding the coattails of others.

As for poorly managed teams? Much like an ensemble metric, the coefficients found by such a regression (or whatever statistical puzzle mapper you choose to use) are taking the ensemble over time of a great many opinions, of a great many teammate combinations, and so we can hope they represent bias only in the sense that coaches, on average, likely do have a bias of some sort, toward a certain type of player. Over the span of time we can hope that this averages out tidily enough.

Finally, tanking teams? Well, one simple solution in all of this is to look only at the top-half of the league to perform this analysis. Where is the real competition for playing time? Where are the most difficult decisions being made? In the front offices where winning now matters.

So, if you want to build a metric along the idealised lines of the original post, I would strongly seek to persuade you to have a look in that dimly lit corner.

"Why is this maybe a good idea? Well coaches and their staff inarguably see more of what their players are, than we ever will. Coaches must allocate minutes in an attempt to maximise team success, and so this marginal allocation of time would seem to be very likely to correllate with actual ability."

Disagree with inarguable and very likely. The evidence of substantial apparent departure of coach given time and performance is obvious by every known / attempted metric under the sun and many many eyetests.

I concede that they have more information and impressions but not that they are always / exactly correct in how they use them. Some of the information is misleading or simply used wrong or imprecise or used very sloppily or not used. Some of the impressions are wrong / inconsistent with facts.

The degree to which you think coaches are "right" in the details will vary and reasonably for a range. But what perspective is most reasonable / right and what view is too much / wrong will vary and will sometimes be argued.

I have a ton of comments over many years contesting the results and the wisdom of coach driven lineup management and I stand behind it. That's just me.

Contracts / politics and comfort / habit and failure to fuĺly and properly use performance data pull minutes allocation from optimum. In my view, by a lot with not much doubt. But others can and do assess it differently.

I mean, if you genuinely think you see more of what players are and are not than the staff who work with them, then I honestly don't know what to say to that.

The point you've missed, or failed entirely to understand, is that in the aggregate these will net out. That does not mean the estimator won't have bias, only that that bias won't belong to any single coach or staff, but to the broader space of front office thinking.

In any case I can say, factually, that there is value in that line of thinking.

However, if you wish to sit and postulate, whilst never actually taking up the mantle of doing any of the work yourself, then you will remain as you are.

I have taken up a heck of a lot of work on understanding NBA performance and understanding lineup management. And drafting. I could do more employed by a team.

The point is to get specific calls "right", not settle for averaging out on aggregate.

What works on average in betting is not necessarily as useful for running a team.

There could be other models, useful models still to develop but a lot can be done with those already out there. Analysis with existing tools is what I do best.

I don't recall your idea from a decade ago. But if it is a good idea, develop it. Share it if you want. I'd listen, evaluate, probably use. But you haven't done it and I can't use and don't feel negligent for not having done what you postulate loosely, especially with my deep seeded, deep supported skepticism about the optimality of coaching minute allocation. With your perspective on the wisdom of coaching minute allocation, you do it. But try to find the truth, not just your supposition.

I genuinely think one can see somewhat more clearly what players are and are not in many cases with rigorous use of existing analytic tools than what the staff who work with them apparently think and actually do with them in chosen minutes, usages and roles.

If I thought they were always right or always to be deferred to, I wouldn't have commented for decades and think I have proven more right on many many occasions.


Do you think coaches who use 500-1000 lineups are right and should be deferred to? I don't. Do you think coaches who continue to use a majority of negative performing lineups in their most used set are always doing the best possible? I don't.