National Celtics discourse
- Thread starter RorschachsMask
- Start date
Intelligentsia probably isn't the right word and certainly isn't being deployed as a compliment. That's not to say that I don't respect hoop analytics guys--probably could make a case that I'm too tied up in the analytics. I generally mean the group think that forms out of a combination of the twitter/insta heads, the online blogger types, the TV talking heads, etc that seems to water down the honest analysis in order to get eyeballs. Needing eyeballs always fucks honest analysis, because people need to make a living. These aren't scientific academic types fine with riding a bike to work until age 60 if they can get a bacterium named after them or discover a comet or something.
Fwiw, I agree that volume and shooting/scoring efficiency is presently calibrated correctly among analysts and even the group think overall. It's actually a solid achievement and another example of how far hoop analytics has come in just a few years.
But that only covers offense, which despite everyone whistling past the graveyard is still half of the game. Not only are we shit at measuring defense still, we honestly don't even bother (royal we) acknowledging its existence about 99% of the time. If we did, Luka wouldn't be #2 on anyone's list anywhere, unless the ranking included the word "offensive". That's my issue. Giannis is so far down these lists that his name barely comes up around here even, which should be embarrassing.
Or could it be that his defense isn’t being measured properly? If Tatum (ex
was graded on his defense vs 7-foot offensive bigs in Dallas the way Doncic’s against offensive 1’s in Boston the dynamic was be reversed.Edit: Tatum was an “example” not a smiley face lol
Scoring efficiently at volume is the single most difficult thing to do in basketball. It should be viewed as the most important thing
I’ve been telling people for years now, while individual efficiency is obviously important, it has gotten significantly overused, especially in the playoffs.. Team efficiency and what you can do for that is definitely more important.
Individual efficiency matters more if scoring is the only elite skill you bring to the table. You’re able to make up for a couple of extra missed shots per game if you’re strong everywhere else. What is a bigger detriment to their team, Tatum missing an extra two shots per game, or guys like Luka/Hali getting picked on a majority of the possessions on the defensive end? Tatum taking care of the ball also plays into it, a missed shot is less offensive than a turnover.
Individual efficiency matters more if scoring is the only elite skill you bring to the table. You’re able to make up for a couple of extra missed shots per game if you’re strong everywhere else. What is a bigger detriment to their team, Tatum missing an extra two shots per game, or guys like Luka/Hali getting picked on a majority of the possessions on the defensive end? Tatum taking care of the ball also plays into it, a missed shot is less offensive than a turnover.
I'd engage on this if I thought that you really believed what you're shoveling.Or could it be that his defense isn’t being measured properly? If Tatum (ex
It is the most important thing. Nobody questions this. Nobody.
The same as OBP being the most important offensive skill in baseball.
The difference is that the baseball stat nerds factor other stuff into their analysis, and the hoop ones just don't.
The hardest and most important thing for a player is being able to consistently compromise and then exploit the defense.
Scoring efficiently is a decent proxy for this, but it's far from perfect, because a given star only attempts a shot on 15-25 of the ~75 possessions for which he is on the floor during a game.
Potential assists are also a decent proxy, but ignore the quality of shot created by those potential assists, and also only comprise 8-15 of those 75 possessions.
Peak Curry is instructive here: he was the heart of a devastating offensive ecosystem, and his efficiency was really good, but the shot and assist volume wasn't that high relative to some guys now. He was still doing the hardest thing in basketball really well though.
My numbers and observation from last season 100% believe it. I profited quite a bit from it last year post-ASB and playoffs as well as those here who are in my Discord can attest. I posted on this here during second half of season as well.
Glad that you did well gambling honestly. I know that you're a smart guy. I also know that you don't think Brown and Tatum traffic coning Luka for 5 games has anything to do with Luka being weak against "offensive 1s" as you just wrote.
Everyone sucks at covering offensive 1s. If Luka were getting cooked by Ant (when), that's less of a criticism. On this we agree. When he's also getting routinely cooked by 2s, 3s, 4s, and 5s, it's no longer a matchup issue brother. It's a guy who sucks at defense.
This resonates for me, and it's often felt like the primary disconnect (distinction? The incongruity I feel about this is relatively minor) in "rating" Tatum vs. other top stars.
I don't think Tatum's shooting efficiency correctly proxies how much he bends opposing defenses, or how willing he is to exploit that by being the first passer, but not necessarily the assist passer.
The subjective example I can make is some past conversations with a non-obsessive-diehard level C's fan, about why it's so clear to me that Tatum is the team's best player, and not Jaylen. To his credit, Brown has lots of games where he's finishing more plays, and often more efficiently, than Tatum. But to my eyes, EVERY opposing defense treats Tatum like the bigger threat, puts their best defender(s) on him, is quicker and more likely to send help to him, etc. And it's Tatum exploiting that attention, often by moving the ball, and creating better shots for others (again, including lots of plays where that happens 2-3 passes away) that really makes the C's offense sing. Tatum just has a level of gravity, and ability to exploit that, that's above and beyond even his very-strong scoring volume+efficiency, in a way that isn't always true even for the very, very top guys.
I do think Curry is the platonic ideal of that admittedly very subjective way of thinking about most impactful individual player value; his shooting and movement gravity is otherworldly, and especially combined with a team full of intuitive ball movers, and at least one other all-time-great shooter in Klay (when they also had Durant, it broke the league).
FWIW, I also think Jokic and Giannis do similar things for their teams, probably fairly even more so than Tatum (I think Embiid is a step below the other recent-MVP bigs in that regard).
I'd also point out that matchups matter to that stuff, too, and maybe particularly for guards. Broadly, I'd DEFINITELY consider Luka to a have a ton of that "bend and exploit the defense to the entire team's benefit" to his impact/game. The Wolves were the best defense in the NBA last year statistically, they crushed Phoenix and Denver last year, and then Luka destroyed them in the WCFs. But Boston's ability to avoid sending help to him due to their switching and across-the-roster versatility is pretty unique, and uniquely suited to making Luka look bad (in addition to their ability to exploit his mediocre defense). I think both of these things can be true: A) Luka was probably the 3rd best individual player in the Finals, at best (Tatum and Brown for sure, and I'd accept arguments for Jrue); and B) he still might be more valuable/"better" than Tatum across a full regular season+playoffs, because he is SOO indispensable to a team that's also good enough to make the Finals.
Great post.
Agreed that being the highest efficiency volume scorer is not the whole picture. Bending the defense to your will with gravity is a huge and rare skill. Few guys have it in the league at any one time. Tatum is one of them, as is Luka, Jokic, and a bunch of those other guys on top 10 lists.
Darko thought Tatum was the best offensive player in the league last year (first in O-DPM), and the rest of the top 10 (Luka, Jokic, Booker, Curry, Brunson, Kawhi, Mitchell, Durant, SGA) makes enough sense to not write off the stat. Can't say it's all about the Celtics offense either, given that the rest of the C's starting 5 was just 16th (Brown), 40th (White), 44th (Porzingis) and 57th (Holiday). Unless he either takes a major leap forward in basic counting stats and easy to measure efficiency, he's never going to get a ton of credit for being the engine behind the best offense in history. If it can't be easily measured and quantified, it just doesn't exist.
Tatum over Jokic offensively seems crazy to me, but after that.....the dude did lead the best offense ever. Putting up 122 ortg or whatever as a team is completely obscene, and I don't think people really grok how obscene it is.
On a related note, my contrarian opinion re Luka is that his defense is underrated and his offense is overrated. I think the opposite is true for Tatum.
Yeah, Tatum is merely good on defense, while he's talked about as if he's great.
(n.b. this is just D-DPM, i.e. defensive points saved per 100 possessions. I threw in some comparables to other veteran wings.)
I think a lot of that rep-to-reality-differential is from how our teams were very defense-first in the Marcus Smart years, and Tatum was often a big part of that. But starting with the Udoka year but particularly Mazzulla Year 1 and obviously Year 2, we turned into this offensive juggernaut and it's like nobody in the league noticed.
(n.b. this is just D-DPM, i.e. defensive points saved per 100 possessions. I threw in some comparables to other veteran wings.)
I think a lot of that rep-to-reality-differential is from how our teams were very defense-first in the Marcus Smart years, and Tatum was often a big part of that. But starting with the Udoka year but particularly Mazzulla Year 1 and obviously Year 2, we turned into this offensive juggernaut and it's like nobody in the league noticed.
I think people see that Tatum clearly is a top contributor to winning, and they naively look at raw stats/efficiency and think that it can't be coming from the offensive end, so they say "defense" to explain his impact.
Tatum also is a significantly more impactful defender in the playoffs (he's not matching up with the likes of Mobley/Turner/Gafford/Lively in the regular season), which is when most people are watching. He's generally not locking guys up, his greatest asset defensively is his versatility and his ability to provide competent defense against 95% of the league. That's just less useful in the 82 game grind than it is in the final sprint to 16 wins.
Does D-DPM take rebounding into account? For Tatum it's not just that he provides competent defense against most centers, it's that he can generally keep them off the boards. I've seen a couple of media folks talk about the Knicks gaping hole at center, and they've all suggested that OG could end up guarding opposing big men and he is pretty good at that, but he's a pretty mediocre rebounder at best for his size. It's a start contrast to Tatum who is one of the best rebounding forwards in the league.
Does D-DPM take rebounding into account? For Tatum it's not just that he provides competent defense against most centers, it's that he can generally keep them off the boards. I've seen a couple of media folks talk about the Knicks gaping hole at center, and they've all suggested that OG could end up guarding opposing big men and he is pretty good at that, but he's a pretty mediocre rebounder at best for his size. It's a start contrast to Tatum who is one of the best rebounding forwards in the league.
I don't fundamentally disagree with the conclusion (I personally think he's more like "very, very good" but not considering him truly elite is fair, IMO).
I also believe you, and others here, know more about the underlying statistical methods that produce things like DARKO's DPM, so I'm mostly looking to learn more about the strengths and limitations.
But at a core level, any plus-minus-based stat is actually lineup data, not individual data, and I'm a bit skeptical of the reliability of the methods used to adjust purely line-up +/- into something individual. It seems self-evident to me that the datapoints that separate individuals out of lineup-based metrics are pretty small samples, and over any reasonable timeframe that allows for and incorporates player improvement (or decline), I suspect that those small samples can be over-sampled to the point that the outcomes aren't super-reliable.
Like, IIRC Wayne Winston used to claim that Adjusted-plus-minus wasn't predictably reliable until you had 3 years of regular rotation minutes to work with for every player, and that's a HUGE sample, that is going to hide or smooth out a TON of player improvement/decline, if those changes aren't happening pretty close to the standard model of historical age-based skill development.
That seems especially true for defensive plus-minus systems, where there aren't nearly as many box score numbers that come even close to proxying core defensive value (i.e. actually making it harder for the other team to make shots, which the large majority of isn't reflected in any box-score numbers at all).
Given all of that, I feel like team/lineup context has to be pretty heavily weighed in evaluating defensive plus-minus-based metrics (including DARKO's D-DPM, which I admittedly don't know much of anything about "under the hood" in terms of how it's actually calculated). In Tatum's case, I think there are several factors along those lines that likely suppress his D-DPM (and might also inflate his offensive DPM just a little. Two things jump out in my head:
1) He's never been on a team that WASN'T really good defensively, with other players around him who are also really, really good defenders, and who have also changed teams/roles much more than Tatum has. I'd wonder whether other guys going to different situations, and improving other teams' defenses, gets "over sampled" in adjusting lineup-based plus-minus to individually-adjusted metrics like DPM, and indirectly hurts Tatum (and other guys who have always been on the "very good defensively" C's, including Jaylen, who rates below Tatum in D-DPM and D Box-DPM), because the guys who left and remained good defenders are being credited with a little too much of the team/lineup defensive impact from when they were in Boston.
2) Tatum spends a significant amount of non-starting 5 minutes with bench units that tend to feature the C's least-effective defenders Pritchard and Hauser, playing less with Porzingis last year and Horford previously in bench lineups, etc.) That seems like it might be oversampled in adjusting Tatum's individual plus-minus impact out of the best defensive lineups (which have tended to be the C's starters, I think). Similarly, though, those Tatum-bench lineups have historically rained fire offensively, especially compared to non-Tatum bench units, so the same over-sampling might contribute to elevating Tatum to #1 in last year's O-DPM (I buy top-5 fully, and potentially #2, as lovegtm says, but I'm skeptical of him being ahead of Jokic in that regard).
I fully agree with the above---and specific to Darko, the precise challenge above (the intersection of lineups, matchups that are related to those lineup choices, and sample size) was discussed with bowiac way back when and at least at that point in time, his best response was that you use the data that is there even if it isn't perfect. So we can infer the limitation is a known one, imo.
A specific challenge for Tatum is that he is the starter/scorer who most often is playing minutes with the 'bench unit' which both is worse defensively and will make his adjusted plus-minus look worse by comparison. As Celtics shift to Jaylen playing more of those 'with the bench unit' minutes (as they did later in year) I will not be surprised if Tatum's numbers go up even without any change in his actual game - just deployment and impact on metrics given relatively small sample sizes.
FWIW, as a subjective assessment, I would put Tatum in 'very good, not elite' level defense. He got serious-person exploration for all defense, which is reasonable but probably about his ceiling. I agree with others that part of his defensive value is his versatility---he's at least ok against virtually everyone, and that enables switching and multiple lineups around him. But he is not Jrue, or even dialed-in one-on-one Jaylen defensively.
I'll need to give a think on the rest of it, but I believe Tatum is rated more highly on D-DPM than Jaylen is. They're both well behind our starting guards, of course, but Tatum is given more credit. As he should be, IMO - I think his size/strength combo lets him contest more shots more fully, better resist getting bodied by bigger guys (although Jaylen vs Zion on 3/30 was an eye-opener, certainly), and contest more shots at the rim. His switching also seems to me to be more instinctive and less susceptible to a momentary indecision than Jaylen is (and we see that in JB's ball-handling on offense sometimes too). So I would expect a well-tuned defensive metric to rate him more highly than Jaylen. And, indeed...
I threw in Giannis here because he was noted earlier as being hugely discounted in discussions on best players, and for similar reasons to Tatum - people forget just how good a defender he really is. But Hauser rated the same as Jaylen is an eye-opener too - I'm sure some people could scoff at that, while others argue that it reflects that the data is finally capturing how Sam is a credible NBA defender.
I threw in Giannis here because he was noted earlier as being hugely discounted in discussions on best players, and for similar reasons to Tatum - people forget just how good a defender he really is. But Hauser rated the same as Jaylen is an eye-opener too - I'm sure some people could scoff at that, while others argue that it reflects that the data is finally capturing how Sam is a credible NBA defender.
I defini agree with the overall point that plus minus metrics struggle much more on defense with disaggregating team numbers to individual numbers but i don’t understand what you mean by “over sampling” here.
What's happening probably isn't a correct use of that term, from a statistical standpoint (it comes up a lot in the only real way I interact with formal statistics professionally, which is using the results of opinion polling, where smaller demographics are actually over-sampled in order to make the resulting demographic comparisons statistically-significant).
What I was trying to say is probably more like the opposite of that, where a small sample ends up being given more weight than the larger sample. In other words, the amount of minutes/resulting data points that distinguish individuals within regular 5-man lineups from each other are pretty small samples, and because they're the only real data points that do that, especially defensively, they seem likely to hold a LOT of weight in the adjustments that individualize lineup-level plus-minus. Potentially an outsized weight, compared to how small a sample they actually are, and how reliable the conclusions drawn from them should actually be considered.
EDIT: But again, I don't know how the adjustments are actually made in DARKO DPM, so I'm asking questions about this, more than trying to make conclusions.
Hauser has enough size that most wings can't go through or over him, moves his feet really well and keeps his hands up, rarely gambling. Teams just attack him constantly, and it feels incredibly rare that any non-all star level players make him look bad. Hauser led the league in point differential in both the regular season and in the playoffs, and that just doesn't happen if you're bad at defense.
I do think him being on the same level as JB says more about Brown than it does Hauser. Jaylen is still not great off ball and goes through large stretches of the regular season where he's just ok. It's good to know that he can dial it up the way he did in the playoffs (and at times in the regular season vs guys like Zion and Jamal Murray, who are about as different offensively as it gets), because that's the only time we need him expending the extra effort.
I don't disagree with any of this, but I'll add that bench guys under CBS and CJM often get minutes in situations where they are more likely to succeed than random minutes. That's not a knock on Sam at all. He's a very underrated defender by the general public. He's probably playing with more bench guys than he would be as a starter, so he's not getting cover from as many elite defenders. But he's also more likely to get quick hooked if the opponent rolls out a 3PG or 3Big lineup. Pritchard too. That's a luxury that the starters don't get.
Consider me a scoffer. Can you imagine what would happen if Hauser had to chase Curry around the court for a quarter, much less a game? Plus, Hauser can be pretty good defensively so long as he has the shot clock plus help behind him. The Cs may be okay in allowing Hauser to be switched on to a scorer, but they aren't sticking him out there on an island like they do with the rest of the Cs.
Without knowing the details behind DARKO's model, I presume that KP being ahead of JT and JB by so much is a function of KP getting a ton of blocks at the rim. The biggest problem is that if a guy does his job perfectly - he stops the offensive player so well that offensive player gives up the ball and doesn't shoot - I would imagine its hard for any model to figure out how to count for something that doesn't end up in a steal, a block, a made FG, or a missed FG.
I don’t think sample weight is the phenomenon making it difficult here, and the loss function usually used in these sorts of models isn’t going to be more heavily weighting certain data points purely by how rare they are unless someone tuned it to do that which i haven’t heard of.
All on off models do some type of regularization to try to account for heavily correlated data (teammates play together etc) but generally to do that effectively you need sufficient data to separate out individual effects, i think what you are getting at is the low amount of data of, for example, player A playing without player B, is causing high variance.
extremely pedantic point but high variance is not the same as over weighting / underweighting samples. but it leads to the same idea which is these models try to correct for these correlations but can only do so much.
to take an extreme example supposed player A played if and only if player B played your model would have no way to distinguish their contributions, assuming it was a pure on/off model.
i do like models that use box score + on off data. like, there’s no reason for us to toss out solid prior knowledge about how basketball works based on box scores. like guys who can shoot high volume 3s well are generally gonna be helping the offense.
Jaylen is such a tricky player to give one singular rating. It matters how much weight is placed on individual iso-defense and off the ball team defense. He’s the type of player where sample size isn’t going to matter as his performance in these areas is so binary. He’s a very good individual perimeter defender, probably top 10% in league, and is just as poor of a weak side team defender. If it wasn’t for his athleticism making up for some of his poor reactions/reads he’d be close to the bottom 10% in this area. With this knowledge I can see how Hauser can be rated similar or even ahead of Brown depending how each form of defense is weighed.
Since Luka’s defense has come up in here. These clips are just from a couple of games, too. And Tatum missing a couple more shots a game is supposed to be nearly as detrimental as this?
View: https://twitter.com/JahGoated/status/1836355720939192349
View: https://twitter.com/JahGoated/status/1836355720939192349
Tatum is a vastly better team and off ball defender than Jaylen, and is somewhat more versatile - including reboudning - and thus is a better overall defensive player. But Jaylen can be an elite on-ball defender when he is dialed-in, better than Tatum (and better even than White, I'd argue). Jaylen is still a lot weaker off ball than those guys, and a lot less consistent so he isn't in White's class, or even Tatum's, in terms of overall defense....but as an on-ball guy he can be really dominant.
Also from a couple games….4th in league in loose ball recoveries, top10% in deflections. Lowlights though. People won’t watch this in it’s entirely as it doesn’t fit their narrative.
View: https://youtu.be/dmq7GRugkO4?si=8F9GOyaTO3mR08Wy
Fantastic call on Dallas pre-Playoffs by yourself but their run wasn't driven by Luka's defensive highlights.
Luka floats & gambles on D, completely undisciplined. His steals, loose ball #s, & highlight reels, from playing passing lanes & not guarding his man make sense. BUT it also leads to open driving lanes when teams slow down in the playoffs and play half-court offense. Luka holds the 3 worst Blow By % in playoff series over the last 20yrs; our eyes weren't lying.
No narrative is needed, his playoff defense was gawd awful because he's slow, out-of-shape, undisciplined, not held accountable & was partially injured. His size makes him slightly better than shrimpy small PGs like Dame, Hali, Trae, Celtic Kemba but he still stinks defensively.
As long as Luka shows up out-of-shape, Dallas poses absolutely zero problems to Celtic title chances. We should be openly rooting for them against WC teams like OKC, Minnesota, & Denver.
https://www.cbssports.com/nba/news/luka-doncics-defense-hits-shameful-levels-with-mavericks-on-brink-of-being-swept-by-celtics/
Luka really seems to have the numbers of Keldon Johnson and Darius Bazley.
Haven't gotten through the whole thing (but I will!) but it seems like a good portion of the highlights involve Luka getting steals because he's in the correct place. He obviously has pretty quick hands - which he would need to be so good on offense.
Still, you are correct that there are instances where Luka moves his feet and stays in front of smaller, quicker guys. A couple that stood out was staying in front of Haliburton, Dosumno (2:15 mark; pretty impressive since he had to flip his hips on that one) and Reggie Jackson.
So question: does it speak better or worse for Luka that he's capable of really moving his feet on defense when he spends so much time not doing it? I mean remember when JKidd said that Luka had to "participate" on defense? He wouldn't have to do that if Luka was moving his feet on defense for most of the games.
When watching Luka in the finals it seemed like the majority of his problems on defense were because of lack of conditioning and lack of effort.
He was often able to move his feet and stonewall the offensive player on their first try…and almost never was able to stop them after they pulled back, regrouped and did it again.
Unless he gets in better shape and starts giving a shit on that end, he’s going to be someone that good perimeter players can isolate and pick on
He was often able to move his feet and stonewall the offensive player on their first try…and almost never was able to stop them after they pulled back, regrouped and did it again.
Unless he gets in better shape and starts giving a shit on that end, he’s going to be someone that good perimeter players can isolate and pick on
I mean those are definitely two of the top 450 guys in the NBA.
Perk continues acting out because the organization told him to fuck off
View: https://twitter.com/ESPNNBA/status/1838659419988512836
View: https://twitter.com/ESPNNBA/status/1838659419988512836
Think there’s a good chance that this thread is quite active this season lol.
Media has shifted from “Celtics were lucky with injuries” to “they’re destroying the game!!”.
Media has shifted from “Celtics were lucky with injuries” to “they’re destroying the game!!”.
The Celtics are basically who they were all of last year, but apparently now folks are noticing. Better late than never, I guess.
The NBA media becomes much simpler to understand in their takes when you realize that they are always a year behind what is actually happening. So by this logic, this train of thought reflects more upon 2023-24 than this season.
Very good point but doesn't explain the Luka love from last year's playoffs. NBA media is a year late AND loses all perspective over certain star players?
Luka love is totally understandable in its simplicity and inexplicable in its ubiquity.
This is even more converse to what we do in forecasting what is to happen before it actually does.
Would you have said this prior to the Celtics series?
I can't speak for JTT, but most of this forum was rolling its eyes at the Luka ballwashing before the finals and wondering out loud why the national media was humping Luka as the clear best player on either team. I'm 100% certain that the search function will bear that out.
With the caveat that I only skimmed the first installment of the series and didn't read anything else, the Athletic/MYT did some sort of ranking that was to include "Reliability" into the factoring. I think it's supposed to be best player factored by playing the most games. Or something like that (it's not really important).
When factoring in availability, JT ranks #1. Jokic is #2. After that, I'm not sure how things are decided but I do know that I don't really care.
Any rate, if anyone is interested, here's the last installment: NBA Reliability Tiers: From LeBron to Luka, league’s top stars populate our top 10
When factoring in availability, JT ranks #1. Jokic is #2. After that, I'm not sure how things are decided but I do know that I don't really care.
Any rate, if anyone is interested, here's the last installment: NBA Reliability Tiers: From LeBron to Luka, league’s top stars populate our top 10
This one is going to be fun if HRB's claim is that people here weren't skeptical of Luka pre-Finals, because we have receipts and then some.
(Obligatory and very tedious-to-constantly-give disclaimer that Luka Doncic is an elite NBA player, and that this conversation is about a) his relative ranking b) his projected performance against Boston, pre-Finals)
Ok I believe you. I don't recall specifics but that is about as weird as the Brunson hate I'm ready here. Guys carry their team through playoff series and are the anti-christ doesn't make much sense to me.
Skeptical vs the Celtics in The Finals and ignoring what he did in the playoffs to get the Mavs there are two COMPLETELY different things. Shit, I had money on the Celtics sweeping Dallas for these same matchup reasons at 11-1 because we all could see the problems that he would have against this team. Everyone talks about Doncic in The Finals while ignoring everything that he did leading up to that or critical of him while he was carrying them....which is even more nuts to me. Simplistic though. /eyeroll