APBRmetrics

Mike G wrote: ↑Mon Mar 10, 2025 9:59 am OK, just curious. If a team BPM is -2.6, and a -2.0 player enters the lineup, does he have a positive impact?

Right, the vast majority of NBA players are below the average NBA impact. It's not a bell curve, it's a strongly skewed distribution with relatively few players with very strong positive impacts and high minutes and a lot of players between -1 and -3 that get few minutes.

Now, if you're talking about impact on a championship contender, you probably need to be at +0 or above. But if you're just trying to win a few more games and your team's net rating is -12... as Crow says, you can be -2 and improve the team.

DSMok1 wrote: ↑Mon Mar 10, 2025 2:52 pm

Mike G wrote: ↑Mon Mar 10, 2025 9:59 am OK, just curious. If a team BPM is -2.6, and a -2.0 player enters the lineup, does he have a positive impact?

Right, the vast majority of NBA players are below the average NBA impact. It's not a bell curve, it's a strongly skewed distribution with relatively few players with very strong positive impacts and high minutes and a lot of players between -1 and -3 that get few minutes.

Now, if you're talking about impact on a championship contender, you probably need to be at +0 or above. But if you're just trying to win a few more games and your team's net rating is -12... as Crow says, you can be -2 and improve the team.

Last year every team had at least 1 player with 500+ who was negative bpm, so the bar is probably a little below zero.

https://stathead.com/tiny/t1q71

On the other hand, in the long term I think a problem people have in looking at player value at the championship level is that if you have a -1 bpm player he has positive VORP, and he "helps" your team in winning games, but doesnt help it win a championship, as giving minutes to that player "hurts" your team in achieving an SRS of +6/+7 or whatever the lower limit needed to be a chamionship team.

Can evaluate player impact on current team or championship team... or an average team. On a flat average / neutral SRS team, a -1 on BPM would appear to have negative impact on that team. But check the raw +/-... and the RAPM... and then decide if it is enough to call it negative or meaningful negative.

Now If a team BPM is -2.6 on average and a -2.0 player on BPM, raw +/- and RAPM enters the lineup...

decide whether you want to talk about impact on that team relative to its average and / or specific player alternatives or an average team or a championship team or a 10th seed or whatever.

DSMok1 wrote: ↑Mon Mar 10, 2025 2:52 pm...
Now, if you're talking about impact on a championship contender, you probably need to be at +0 or above. ...

Rookies who get minutes are mostly with non-contending teams, though.

The top-heavy 2023 champion Nuggets had 3 players above zero BPM in the season and again thru the playoffs. And they won it pretty handily.
I think any BPM above -5 is contributing more than nothing toward a winning effort. So Jeff Green (-3.8) with Jokic (12.8) total 9.0 and still dominate other frontlines.
Below Green were less reliable players.

It may be only semantics, but when I see some NBA players referred to as negative contributors, I always wonder what happens when a player or coach hears or reads that.
A minus sign applied to mean "not among the top 100 in the world" does not indicate that a player's efforts make a team worse; even the most successful teams utilize such players.

Relatedly, I am pretty sure that replacement level is quite a bit lower than I previously thought. -2 is what was used for VORP, but more recently additional research has shown to me that is quite a bit too high. -3 is probably about right, perhaps even -3.5.

This is because better statistics are showing that the spread in player performance is wider than I previously thought. (Higher highs and lower lows).

I have considered -2.7 or -3 replacement level line but mainly use the label on those clearly under.

2023 Nuggets

https://www.pbpstats.com/wowy-combos/nb ... 999,201145

Jokic with Green, team results 6.7 pts worse than without. Green no Jokic 19 pts worse than together.

Can stretch "positive with" in some other cases but I am not stretching that far.

Jokic with Green, team results 21 pts worse than without in playoffs.

Oddly Green / no Jokic worked in playoffs.
Need to look further at that but my approach would be use labels appropriate for context. Regular season data 4 times the playoff data. Green / no Jokic in playoffs under 4% of total data.


Green no Jokic in playoffs mainly worked with Gordon and at least one of Murray or KCP and was probably mainly positive because of the others. But finding something that worked with a generally negative player is worth something, though in fairly small minutes.

2 yr RAPM had Green at -2.6 in 2022-23. LeBron put him at -3.1. Every other metric had him clear negative. Looked better in 2021-22.

Maybe he had positive value... outside of with Jokic.

6 factor RAPM had him at -2.5 in 2022-23. +0.6 on efg%, huge negative on offensive rebounding, fairly minor impacts elsewhere.

Use everything.

Almost neutral on / off in 2021-22. Went to -15.8 in 2022-23. Playoffs went from almost neutral to -11.8.

There was a time when Green was not that harmful but he was mostly harmful in my evaluation of 2022-23. Other parts of team were so good that his minutes did not change the outcome. But Nnaji, Jack White and Cancar all had better raw on and on / off. 2 of the 3 significantly less negative BPM, 1 about the same.

Green's performance in 2021-22 may have colored his impression on Malone too much. But he didn't have "good" options. Probably better ones but still negative.

Braun at PF was / is a possibility but untried based on BRef accounting. He only had modestly better raw on and on / off +/- than Cancar then. Great +/- now. I'd test him at PF now to inform playoffs. Better than Watson. Was better than Nnaji but look closer at recent data. Probably would still be better. Good to test and have another option.

DSMok1 wrote: ↑Mon Mar 10, 2025 7:53 pm Relatedly, I am pretty sure that replacement level is quite a bit lower than I previously thought. -2 is what was used for VORP, but more recently additional research has shown to me that is quite a bit too high. -3 is probably about right, perhaps even -3.5.

This is because better statistics are showing that the spread in player performance is wider than I previously thought. (Higher highs and lower lows).

Is at least part of the spread due to the fact there are so many more players in the league now? And less players are in total playing a lot more minutes?

Median at 345 of 689 players for BPM at BRef is -2.2 currently.

-3 hits at about 390. -3.5 at 415.

Medians could change if a minimum minutes was applied.

Replacement level at 11.5th man or 13-14th?

The estimate matters more to me than the application of labels.

From a spreadsheet of last season, I ranked the 450 players with the most (>160) minutes -- 15 per team -- by each of these four summary stats. The 15th ranked player would be the median #1 player for a team. Each subsequent 30th rank could be #2, #3 etc. https://www.basketball-reference.com/le ... anced.html
Leader in most stats was Embiid, who played just about 1/3 of possible minutes. So other players replaced his minutes; and the #6 guy moves up, often to the 3 or 4 spot, sometimes higher.
Sixers currently are close to what you see when starters are #6-10 in this picture.

Looking at any of the positive stats (and knowing they are the same-level players across the columns), I don't see "negative players". There are ~5 above-avg players per team, perhaps 8 above-median.

When I do 'PER wins', I subtract that 6.3 from the PER. For BPM wins, I add 5.9 to BPM. Zero win-contribution can therefore be (15.0-6.3=) 8.7 PER or -5.9 BPM. Less can be considered negative.

So, in your perspective, the lowest minutes qualified player / worst 15th man in league has no negative impact.

We have very different frameworks.

Why would a coach play someone for their negative impact? Why would they be on the roster, unless they are a project for development?
In playoffs, nobody uses 15 players. Those who get playing time are expected to do some good.

The numbers indicate median 1st thru 15th, not best or worst.

Players with negative impact compared to league average are on the roster and on the court because there aren't enough positives and there naturally is a left tail to the distribution around average.

If you want to compare everyone who plays only to those that don't / shouldn't more than everyone that does, then you have a different answer, different meaning.

Being exactly at replacement level means that you are worse / more negative than everyone above that line. It also means you are better performing than everyone below that line. But the original concept was that players around that level are common and pretty easily replaceable.

There are 104 total players between -2 and -4 on BPM. Just 25 of those get enough playing time to be "minutes qualified".

Today BRef is showing 110 others below -4. Yesterday there were 140 more. Not sure what changed, maybe they are players who haven't actually played. But clearly there is near some definition of replacement level and then a lot of players measured below that. Perhaps could be better with more trial. To replacement level or perhaps beyond it. But lots of them, given low minutes, low regard, low security.

One can easily compare a 1.20 to a .80, or a 17 to a 13, +1.5 to -1.5, etc.
If coach has to give Jokic some rest, a .30 sub is 7 times as 'below average' as a .90 sub.
But do the math: referring to the chart above, .90 e484 is equivalent to -0.5 BPM, and .30 = -3.5 BPM.

A .90 backup center is a great asset, and a .30 is pretty marginal. But if the difference is 3 pts/100 poss, there is little difference in a 5 or 10 minute (10 to 20 poss.) stint. Like 0.3 to 0.6 points shaved off your lead.
Either player, however, gets Joker some needed rest. Without a break, he's not a 2.95 or +13.0 or 31. This is distinctly positive.

Mixing up the lineup also introduces some uncertainty in the opposition. Porter and Gordon and Murray are likely to dig further into their bag of tricks, make more assertive drives to the basket, etc.

If you have -1 and -2 BPM players coming off the bench, you have a deep squad. It's an embarrassment of riches.
It's also easier to say/write/explain "below average" than "with negative impact compared to league average".

Every team digs down to their 10th man and beyond at some point. Having a better-than-the-avg-10th-man is an asset.