Old man MJ vs Old man Lebron

[mysticOscar]

My take on Russell is that he was born in an era with not much footage and a league that was viewed by many as not a full blown professional league.

I'm not quite sure what "a full blown professional league" means (or who the "many" are, without any citation) but if there's an implication that it wasn't fully professional (it's hard to see what else "professional" is doing in that sentence)... I think he and the people paying him $100,001 a year in the 1960s might be surprised.

I've got that contract occurring in 1965. Let's say it started then and counts for 1966. which one source says puts him well more than 20x the mean and more than 9x enough to put him in the top 5% of earners (https://dqydj.com/individual-income-by-year/). Obviously not everyone was as well paid as he was but that would be quite the resources for an amateur or semi-pro league.

"Full blown professional league" means a league where majority if players don't need to hold 2nd jobs outside of there sport to live off all year round

[tsherkin]

My take on Russell is that he was born in an era with not much footage and a league that was viewed by many as not a full blown professional league.

I'm not quite sure what "a full blown professional league" means (or who the "many" are, without any citation) but if there's an implication that it wasn't fully professional (it's hard to see what else "professional" is doing in that sentence)... I think he and the people paying him $100,001 a year in the 1960s might be surprised.

I've got that contract occurring in 1965. Let's say it started then and counts for 1966. which one source says puts him well more than 20x the mean and more than 9x enough to put him in the top 5% of earners (https://dqydj.com/individual-income-by-year/). Obviously not everyone was as well paid as he was but that would be quite the resources for an amateur or semi-pro league.

I recall reading it was a big deal when Wilt got his first contract at $50K per year, so for Russ to get a six-figure contract only a few years later would indicate a major acceleration in league salaries in the mid-60s. Which kind of makes sense.

[lessthanjake]

Yeah, It looks that way because you are doing a different calculation than what is on the basketball-reference estimates. I do not know whether you are deliberately lying or you do not understand basic math principles as well as you claim, but for the thread’s benefit, you should detail exactly what you would do to calculate Lebron’s numbers with the same process you used to calculate Jordan’s if you had no access to those on/off estimates.

Perhaps cut down on the rudeness. You don’t need to make personal attacks to discuss basketball. I’ll address the rest of this below, but just wanted to call this out specifically here. There’s just no need for this behavior.

To be clear. What Jake is doing is in essence calculating Jordan’s per 48 average (without posting it because that would reflect poorly on Jordan) and then multiplying it by 100/“Average BBRef Bulls Postseason Pace in applicable years” and then comparing it to Lebron's BBref on/off as opposed to undergoing the same process with Lebron.

Umm, you make this post after I literally posted a per-48-minute average in this very thread, in the very post that you’re responding to (and, spoiler alert, Jordan still looks substantially better). So this is a pretty wild accusation from you.

Anyways, the reason I’m not “undergoing the same process with LeBron” is that we have better data for LeBron so there’s no need to. When we have actual per-100-possession data (which we do for LeBron), then we should obviously use that rather than trying to derive per-100-possession data using per-48-minute data. For Jordan, we don’t have per-100-possession data, so our best option to get that data is to derive it from per-48-minute data using the team’s pace. I’ve been quite upfront with the fact that there’s potential error in that (for instance, if the pace on and off the court was different), but it’s obviously the best method we have for Jordan. It would obviously be dumb to use the same method for LeBron, because we actually have the real error-free numbers.

Of course, even if we *did* look at per-48-minute on-off, saying that LeBron “matches” Jordan is a bit odd. In the data we actually have, Jordan’s per-48-minute playoff on-off is, by my calculation, about +15.7, while LeBron’s per-48-minute on-off even in that specific optimal timespan you identify is, according to you, +13.8. So yeah, LeBron doesn’t match Jordan.

Uh

[note from lessthanjake: cut out the spoiler of homecourtloss since it was breaking the website]

Adding 4 games should not be taking us from sub-14 up to 16.

I don’t have any idea what method or data homecourtloss used to calculate those numbers. The method I used to calculate the per-100-possessions data is extremely transparent, with there being a very long post explaining every bit of my methodology. And I can tell you that what I did to calculate a per-48-minute value for Jordan in my earlier post in this thread was just to take the eyeballed values in the Thinking Basketball video (see the other thread detailing exactly what those eyeballed values were) and the known values from the Squared 1985 playoff data and then calculated “on” and “off” values by doing a weighted average based on the number of “on” minutes and the number of “off” minutes. Pretty simple stuff. And it looks great for Jordan.

Yes you know, it is a massive negative but not exactly just how massive, so better throw it out entirely and pretend it is not there.



How about this? Use the nicest derived result you can. There are a few different approaches, but even the most mild is still an applicable drain.

I don’t think you quite understand the situation with 1995. There is no “nicest derived result.” There’s only one derived result, because there’s only one three-year chart in the Thinking Basketball video that includes 1995. And we know from trying other derivations that using one three-year chart is prone to a huge amount of inaccuracy, and this inaccuracy is doubled because we’re trying to derive two values (both “on” and “off”). It’s just not data that should really be thrown into an average, since it’s just incredibly rough. [That said, as you well know, I did make an attempt to do so in a post in the other thread—which makes your accusation here that I am “pretend[ing] it is not there” really odd—and Jordan’s playoff on-off still looked fantastic, at about +15 per 100 possessions.]

Ultimately, I think what we can say is that 1995 would drag down the average, by some unspecified amount. What we can also say is that the optimal timespans you’re using for LeBron involve cutting out playoffs that drag down his average too. The bottom line is that you’ve identified an optimal time period for LeBron that includes just under 90% of his playoff games, and we have decent data on just over 90% of Jordan’s playoff games. And when we compare those, Jordan’s playoff on-off is clearly superior.

[Djoker]

--

I'm sorry but I don't get your point. The BRef ON-OFF data (obviously?) uses their own pace estimates and the MJ data was computed using BRef's pace estimates as well.

Right, so logically, you should be able to recreate BRef’s numbers yourself, no?

Say BRef has an abnormally high pace estimate for Lebron’s teams, and that explains why Lebron’s on/off is different (lower…) than what you would pull from PBPStats or NBA.com. The pace is not so far off that it should result in substantially different numbers, yet here we are.

Of course you're trying to play with numbers to suit your narrative so you're removing Lebron years you don't like.

I only see one group playing with numbers unevenly to inflate the guy they prefer. If you cannot be consistent in your process, then you have no process.

You can calculate Lebron’s career on/off. The larger the sample, the more valid the number, and Lebron did not actually change from being a minus twenty something player in the 2006 postseason to being a positive twenty something in the 2007 postseason. The reality trends toward the larger sample, but it is ultimately a trend.

However, when people watch all your standards suddenly become pretty much exactly one of counting out a number and declaring it a measure of that player’s quality — now that you have some superficial indication that it supports what you believe — yes, I think it matters that you are including “negative” Lebron years where Jordan by comparison was not even present or good enough to lead his teams there. And I think it also matters that now we are abandoning Jordan’s negatives for lack of “confidence” in the precise scale of negativity. HCL gave you a game from 1986; where is that in this very dedicated and principled summary of Jordan’s colossal and very real advantage?

I personally have already conceded that MJ's total career ON-OFF per 100 is probably around +15 or even +14 because like you said 1995 brings him down. That's still noticeably higher than Lebron's career ON-OFF per 100 which is +10.2. It's kind of indisputable at this point even if we assume the worst drop in the remaining games that Jordan is ahead.

Except that is not the case if you use the process evenly. Nothing is indisputable so long as you are inventing a wholly disparate and inflationary process for one and then attempting to pile that entire gap onto marginal pace disparities.

Of course it's also possible that 1986 and 1987 helps Jordan's ON-OFF the way the 1985 series does. I doubt those Bulls teams were doing well against the Celtics when Jordan was off the court.

Yet in the one game you were given from 1986, it was not any real positive for Jordan.

First of all I don't think lessthanjake has demonstrated a lack of transparency and thus it's not fair to accuse him of fabricating anything. That's just me.

Secondly, if as you claim, BRef calculated their ON-OFF numbers differently from how lessthanjake did, can you please post the BRef methodology instead of just saying that's it's different which is highlighting a potential problem but not giving a solution. It's in everyone's best interest to recalculate the numbers if they are indeed inaccurate so please shed light on this.

[AEnigma]

If as you claim, BRef calculated their ON-OFF numbers differently from how lessthanjake did, can you please post the BRef methodology instead of just saying that's it's different which is highlighting a potential problem but not giving a solution. It's in everyone's best interest to recalculate the numbers if they are indeed inaccurate so please shed light on this.

I am asking you guys to think critically about how we can have a series where Lebron is objectively +10 on the floor in a series with an objective Lakers point differential of -39… and then have basketball-reference tell you that the “per 100 on” is +1.0, and the “per 100 on/off” is +36.7. Switch over to NBA.com and it will say “net rating on” is +2.3 and “net on/off” is +38.1.

Again, across six games, playing 224 minutes, Lebron is +10, and his team is -49 in the 64 minutes they spend without him. You can fiddle with those numbers yourself. Do they match what Basketball-Reference says? Basketball-Reference says the series was played at a 92.9 pace… but then when you flip over to to the playoffs page, it says the Lakers played at a 93.8 pace. NBA.com says the pace was 95.08. And I am sincerely asking how you can blend together all those numbers and replicate the value Basketball-Reference spits out.

The claim is that the data is better. It is very obviously not. And it is one thing if you are just taking BRef numbers equally for both players, because then there can be a wrong result, but it is not in itself biased. However, when you manufacture a number one way and the compare it with a completely different calculation and call it objectively better, then no, you have stopped engaging in honest statistics.

You do not need basketball-reference as a crutch. If you have Lebron’s plus/minus and his team point differentials, you can apply the same exact Jordan process to his numbers. And there could still be flaws in that e.g. incorrect pace estimates (which is why per 48 is a bit more consistent, albeit disadvantaging to a small degree for Jordan because of the slight disparity in their respective career average postseason pace), but at least the process would be even. And at that point, once you have per 48 numbers, you can adjust relative to each other based on that average pace disparity. If Lebron played 2% faster, fine, boost Jordan’s value up 2%. That does not skyrocket him ahead. And when both these paces are well below 100, it should immediately ring as false when Lebron is apparently plummeting from his per 48 values (by basketball-reference…) and Jordan’s per 48 is ballooning by an entirely different process. The most base level common sense stuff here, all ignored.

And by the way: there are two graphs with 1995. Ohayo even image posted both of them. That too evidently went ignored.

[Owly]

My take on Russell is that he was born in an era with not much footage and a league that was viewed by many as not a full blown professional league.

I'm not quite sure what "a full blown professional league" means (or who the "many" are, without any citation) but if there's an implication that it wasn't fully professional (it's hard to see what else "professional" is doing in that sentence)... I think he and the people paying him $100,001 a year in the 1960s might be surprised.

I've got that contract occurring in 1965. Let's say it started then and counts for 1966. which one source says puts him well more than 20x the mean and more than 9x enough to put him in the top 5% of earners (https://dqydj.com/individual-income-by-year/). Obviously not everyone was as well paid as he was but that would be quite the resources for an amateur or semi-pro league.

"Full blown professional league" means a league where majority if players don't need to hold 2nd jobs outside of there sport to live off all year round

Okay well I think there's probably a misconception here then.

Having a second job doesn't mean you "need" one to "live off all year round". Jerome Williams did acting during his career. Trey Murphy III was doing draft coverage this year, players have podcasts whilst active ... I assume even during the season. Just randomly off the top off my head.

One could even argue endorsements or Space Jam or rap careers constituted secondary incomes for additional work. Or an author. Or a pro-wrestler.

One primary question facing people doing off season work is: "Does this contract give me lifetime security?" And there, more than now (though differing for differing players) the answer may have been no. (Another might be about the costs and benefits of "working on their game")

But even then the answer need not be "no" for someone to choose not to permanently retire at ... 35.

I've just seen an article beginning thusly ...

Johnny Davis played in the N.B.A. in the 1970s and 1980s. In those days, before the league became so lucrative, players often had second jobs.

Perhaps the game didn't professionalize until the 1990s or later.

Looking things up apparently Mark West became a licensed broker in 1992. James Donaldson ran a dating magazine. Would one say MJ played in "a league that was viewed by many as not a full blown professional league"? I don't know what percentage of players have had second jobs over time, nor the mean league salary versus national norms, some of these will be dependent on context (how much do they have to play to qualify, what constitutes a second job). One could argue Larry Bird did not consider himself professional enough that he should not be doing manual labor.

If there's actual evidence that Russell's era was filled with those who "need to hold 2nd jobs outside of there sport to live off all year round" then I'm open to hearing it. Given the aforementioned sums (and national norms) that would not be my expectation. Perhaps there is some great line that marks Jordan's era as "professional" (and Russell's in some way not) but I haven't seen it.

[Djoker]

If as you claim, BRef calculated their ON-OFF numbers differently from how lessthanjake did, can you please post the BRef methodology instead of just saying that's it's different which is highlighting a potential problem but not giving a solution. It's in everyone's best interest to recalculate the numbers if they are indeed inaccurate so please shed light on this.

I am asking you guys to think critically about how we can have a series where Lebron is objectively +10 on the floor in a series with an objective Lakers point differential of -39… and then have basketball-reference tell you that the “per 100 on” is +1.0, and the “per 100 on/off” is +36.7. Switch over to NBA.com and it will say “net rating on” is +2.3 and “net on/off” is +38.1.

Again, across six games, playing 224 minutes, Lebron is +10, and his team is -49 in the 64 minutes they spend without him. You can fiddle with those numbers yourself. Do they match what Basketball-Reference says? Basketball-Reference says the series was played at a 92.9 pace… but then when you flip over to to the playoffs page, it says the Lakers played at a 93.8 pace. NBA.com says the pace was 95.08. And I am sincerely asking how you can blend together all those numbers and replicate the value Basketball-Reference spits out.

The claim is that the data is better. It is very obviously not. And it is one thing if you are just taking BRef numbers equally for both players, because then there can be a wrong result, but it is not in itself biased. However, when you manufacture a number one way and the compare it with a completely different calculation and call it objectively better, then no, you have stopped engaging in honest statistics.

You do not need basketball-reference as a crutch. If you have Lebron’s plus/minus and his team point differentials, you can apply the same exact Jordan process to his numbers. And there could still be flaws in that e.g. incorrect pace estimates (which is why per 48 is a bit more consistent, albeit disadvantaging to a small degree for Jordan because of the slight disparity in their respective career average postseason pace), but at least the process would be even. And at that point, once you have per 48 numbers, you can adjust relative to each other based on that average pace disparity. If Lebron played 2% faster, fine, boost Jordan’s value up 2%. That does not skyrocket him ahead. And when both these paces are well below 100, it should immediately ring as false when Lebron is apparently plummeting from his per 48 values (by basketball-reference…) and Jordan’s per 48 is ballooning by an entirely different process. The most base level common sense stuff here, all ignored.

And by the way: there are two graphs with 1995. Ohayo even image posted both of them. That too evidently went ignored.

How is it a completely different calculation? You haven't explained that. lessthanjake's methodology is in the OP of that thread. Tell him specifically what he did wrong with the calculations. What should he have done?

He can't use pace estimates he doesn't have. We don't know the pace with Jordan ON and Jordan OFF. But the odds that over 100+ games this would significantly inflate the data is extremely unlikely. His teams would have to consistently play fast with him on the court and slow without him for it to make a difference.

[Djoker]

I tried replicating the calculations using the the different pace estimates and I'm slightly off the BRef per 100 numbers but I'm also slightly off the NBA per 100 numbers using their pace estimate. It's not a big difference though and over a huge number of Jordan playoff games it's extremely unlikely to make a sizable difference. I'd be very confident that MJ's +17.0 is somewhere between +16.5 and +17.5. Problem is we don't know the pace when any player is ON the court vs. OFF the court so it will be hard to replicate the official data on NBA.com or BRef.

[AEnigma]

How is it a completely different calculation? You haven't explained that. lessthanjake's methodology is in the OP of that thread.

So recreate it. You think there is no meaningful difference? Test it out. Do the same methodology for Lebron and see just how much it deviates from those very legitimate basketball-reference estimates.

You are assuming the methodology matches basketball-reference. If it does, should be easy to show. Test and confirm.

Tell him specifically what he did wrong with the calculations. What should he have done?

He should have taken the same approach for both if he wanted to make a comparison. As it is right now, you have one Jordan calculation with a number, and a basketball-reference page pull for Lebron with a different number, and you see the first is larger and declare that a real advantage (to whatever extent we suddenly want to pretend that is what on/off means).

He can't use pace estimates he doesn't have. We don't know the pace with Jordan ON and Jordan OFF. But the odds that over 100+ games this would significantly inflate the data is extremely unlikely. His teams would have to consistently play fast with him on the court and slow without him for it to make a difference.

I agree. These should be subtle relative changes. So why does that process fail to match basketball-reference? Why does basketball-reference lop off a point of “impact” from Lebron with a pace adjustment while Jordan is adding a point or two? You need to stop just accepting numbers that correspond to what you want. 2007-21 Lebron’s value by any consistent process should be very similar to what you have for Jordan — again, even using the smallest value for 1995 you can glean. Different people will disagree on the “fairness” of comparing total career values to shorter career values, so if that is what you end up needing to insist upon, fine, but right now I am talking about specific calculations, unjustified assumptions of “accuracy”, and the fundamental basis of comparison.

[AEnigma]

I tried replicating the calculations using the the different pace estimates and I'm slightly off the BRef per 100 numbers but I'm also slightly off the NBA per 100 numbers using their pace estimate. It's not a big difference though and over a huge number of Jordan playoff games it's extremely unlikely to make a sizable difference. I'd be very confident that MJ's +17.0 is somewhere between +16.5 and +17.5. Problem is we don't know the pace when any player is ON the court vs. OFF the court so it will be hard to replicate the official data on NBA.com or BRef.

I am not asking about pace data. Either way you are going to be using some estimate. What I am asking about is the deviation. You do not think it should make a difference, but if you are seeing deviated estimates for a six game sample, are you truly confident that would not compound over 250 games?

[falcolombardi]

Using different metrics for players then comparing them is pretty bonkers

Is the minimum expected to actually make sure a comparision of two players on a specific stat is the same stat

if a 6 game sample is already showing difference in results that makes the bigger less sample -more- likely to keep differing, not less

[MrVorp]

According to this squared2020 article, https://squared2020.com/2017/07/10/analyzing-nba-possession-models/, BBref’s possesion model is a biased estimator that, on average, overestimates possessions. So if a aplayer’s on sample is positive and off sample is negative, like Jordan and Lebron most of the time, BBref’s estimate would hurt their On/Offs.

[lessthanjake]

And by the way: there are two graphs with 1995. Ohayo even image posted both of them. That too evidently went ignored.

See, this is where we can see why deriving from these multi-year graphs is silly. I realize you’re right actually that there’s also that 1991-1996 data point on a graph in the video. And if I calculate using that (eyeballing it with a +8.7 on and +7 on-off), it gives us a +8.43 “off” value per 48 minutes instead of the +16.94 “off” value per 48 minutes that I’ve previously noted that I got using the other chart. Using those multi-year charts is evidently just a really flawed method—which is why I altered my methodology in the other thread to avoid using them (despite the fact that doing so actually hurt Jordan’s numbers).

Anyways, if your position is that Basketball-Reference per-100-possessions on-off data is incorrect, then I think that’s a much broader discussion, since those numbers are widely accepted here. I’ve never seen anyone here dispute them except now that they make LeBron James’s on-off per 100 possessions look lower than what has been estimated for Michael Jordan.

[Djoker]

I think you guys don't understand the idea of a random error. Any pace discrepancies are likely to be random i.e. they'll overestimate the number in one data sample and underestimate in another data sample. The odds that ON pace is consistently above and OFF pace is consistently below over a huge sample is extremely unlikely.

Using different metrics for players then comparing them is pretty bonkers

Is the minimum expected to actually make sure a comparision of two players on a specific stat is the same stat

if a 6 game sample is already showing difference in results that makes the bigger less sample -more- likely to keep differing, not less

We aren't using different metrics. We just aren't explicitly given ON possessions and OFF possessions for Lebron on BRef. If we were, we couid recreate their Lebron numbers. Unless we're accusing BRef of fabricating numbers to hurt Lebron which I'm sure we're not.

I'm almost tempted to go through the PBP data for the 2021 series and count the possessions that Lebron was ON and OFF and prove that it would match the ON-OFF number given on BRef. Of course we still can't use that for Jordan prior to 1997 so it's impossible to do the same calculation for Jordan that we have for Lebron. Blaming lessthanjake for being dishonest is honestly insane to me. He's been completely transparent about what he did.

[lessthanjake]

According to this squared2020 article, https://squared2020.com/2017/07/10/analyzing-nba-possession-models/, BBref’s possesion model is a biased estimator that, on average, overestimates possessions. So if a aplayer’s on sample is positive and off sample is negative, like Jordan and Lebron most of the time, BBref’s estimate would hurt their On/Offs.

Interesting. Given the pace adjustment that I did for most of the data using Basketball Reference pace data (i.e. their estimation of possessions per 48 minutes), it seems that, if this were correct, as you suggest, it would’ve also underestimated most/all of the Jordan data I provided, since the pace adjustment inherently used the same Basketball-Reference possession model (and if there were actually fewer possessions, then the pace adjustment would not be going far enough).

Meanwhile, for the Squared data, I used Squared’s possession data for the “on” possessions. But the Squared data I used also ultimately uses the Basketball Reference possessions data too. That’s because I took the number of “on” possessions but then derived the number of “off” possessions using BBREF pace data for the given games. If BBREF overestimates possessions, then that means that the number of “off” possessions was overestimated by a decent bit actually (particularly since the “on” possessions wouldn’t be overestimated and I derived the “off” possessions by subtracting the “on” possessions from the total possessions derived from BBREF possession data). This would also mean that the “off” value for those data points would be too low.

In any event, it doesn’t seem this is suggesting the number of possessions is overestimated by much. Looking at the chart of differences in that article, the average difference seems to be less than 1%. So, I wouldn’t expect it to make a big difference. And, as noted above, it’s something that would be biasing down everyone’s data. But it is interesting.

[ShaqAttac]

Using different metrics for players then comparing them is pretty bonkers

Is the minimum expected to actually make sure a comparision of two players on a specific stat is the same stat

if a 6 game sample is already showing difference in results that makes the bigger less sample -more- likely to keep differing, not less

comparin career stuff to short stuff when a dude is already winning with longer stuff is also pretty sus.

also look to me mj squad making up numbers again. i thought the 86 rox thread woulda been the end of it, but they really have no shame

[AEnigma]

And by the way: there are two graphs with 1995. Ohayo even image posted both of them. That too evidently went ignored.

See, this is where we can see why deriving from these multi-year graphs is silly. I realize you’re right actually that there’s also that 1991-1996 data point on a graph in the video. And if I calculate using that (eyeballing it with a +8.7 on and +7 on-off), it gives us a +8.43 “off” value per 48 minutes instead of the +16.94 “off” value per 48 minutes that I’ve previously noted that I got using the other chart. Using those multi-year charts is evidently just a really flawed method—which is why I altered my methodology in the other thread to avoid using them (despite the fact that doing so actually hurt Jordan’s numbers).

Yeah no one is disputing that the margin for error is high on that, but you do not get to toss it out. The offer has been extended multiple times to take the lowest plausibly calculable negative bound (thus most favourable to Jordan), and you have declined.

Anyways, if your position is that Basketball-Reference per-100-possessions on-off data is incorrect, then I think that’s a much broader discussion, since those numbers are widely accepted here. I’ve never seen anyone here dispute them except now that they make LeBron James’s on-off per 100 possessions look lower than what has been estimated for Michael Jordan.

You have been active for what, two months, and seem to dig up or rediscover established or previously rejected statistical markers every three days. It has been a known “issue” by people who pay attention, to the point that it is frequently necessary to specify sources for that data because of how much it differs.

It only stopped being an issue when it let you pretend there is some sizeable gap where there is none. Usually there at least can be common ground in the data used: someone pulling from BRef will do so consistently, although someone who prefers taking numbers from NBA.com might prefer that when possible and it may make for a marked difference (and to be clear: there are objections to NBA.com too). However, we are not standing on common ground here, which is the core of my point and the one from which you have shied away. If you are confident in Jordan’s advantage, then it should not be a problem or a concern for you to treat them equally here, because you already know that Jordan’s advantage maintains even with 1995… right?

I think you guys don't understand the idea of a random error. Any pace discrepancies are likely to be random i.e. they'll overestimate the number in one data sample and underestimate in another data sample. The odds that ON pace is consistently above and OFF pace is consistently below over a huge sample is extremely unlikely.

We are not hung up on the pace differences. Yes, it may well be that Jordan’s pace is overestimated and Lebron’s is underestimated. The point is the “adjustment” is producing an artificial divide entirely because the “adjustment” is a distinct process from what you see on the BRef play-by-play page.

Using different metrics for players then comparing them is pretty bonkers

Is the minimum expected to actually make sure a comparision of two players on a specific stat is the same stat

if a 6 game sample is already showing difference in results that makes the bigger less sample -more- likely to keep differing, not less

We aren't using different metrics. We just aren't explicitly given ON possessions and OFF possessions for Lebron on BRef. If we were, we couid recreate their Lebron numbers. Unless we're accusing BRef of fabricating numbers to hurt Lebron which I'm sure we're not.

You do not need explicit on and off. You have raw plus/minus and team point differentials. No different from what Squared/Ben had for Jordan.

I'm almost tempted to go through the PBP data for the 2021 series and count the possessions that Lebron was ON and OFF and prove that it would match the ON-OFF number given on BRef.

So just ignoring how BRef does not actually match, huh. +10 on, -49 off, and somehow you end up with +1 per 100 on and that makes sense to you.

Blaming lessthanjake for being dishonest is honestly insane to me. He's been completely transparent about what he did.

I am “blaming” the choice to use disparate methodologies and then pretend they represent the same thing.

[lessthanjake]

Using different metrics for players then comparing them is pretty bonkers

Is the minimum expected to actually make sure a comparision of two players on a specific stat is the same stat

if a 6 game sample is already showing difference in results that makes the bigger less sample -more- likely to keep differing, not less

comparin career stuff to short stuff when a dude is already winning with longer stuff is also pretty sus.

also look to me mj squad making up numbers again. i thought the 86 rox thread woulda been the end of it, but they really have no shame

Umm, the data I provided goes year by year and provides the data we have, so I don’t know what you’re talking about. And it’s worth noting that in the years where it’s only a portion of the year (i.e. the Squared stuff), the Bulls actually did *worse* in the sampled data than in the full seasons. So, if anything, the data as presented is biased in favor of games where the Bulls were winning less.

Meanwhile, the only “career” stuff I included was a career playoff number, but the data set includes over 90% of Jordan’s playoff games, so it’s by no means “short stuff.”

Also, stop trying to stir things up. You appear to think a substantial portion of your role on the forum is to make snide posts referencing past discussions and suggesting that someone in the current discussion was wrong in those past discussions. It’s something you consistently do, and often in posts that have virtually no other content. That’s not productive posting in any way, and it needs to stop.

[Djoker]

I am “blaming” the choice to use disparate methodologies and then pretend they represent the same thing.

It is not possible to use identical methodologies when we don't have ON and OFF pace for Jordan. We are just using the overall Bulls playoff pace to compute the Jordan ON-OFF and that's also why we can't match the BRef data.

However the discepancies are extremely small.

For the 2021 playoffs BRef has Lebron's ON-OFF as +36.7 and PBP stats has it +36.9. Aren't you just splitting hairs here blaming BRef pace estimates and what not?

[ShaqAttac]

Using different metrics for players then comparing them is pretty bonkers

Is the minimum expected to actually make sure a comparision of two players on a specific stat is the same stat

if a 6 game sample is already showing difference in results that makes the bigger less sample -more- likely to keep differing, not less

comparin career stuff to short stuff when a dude is already winning with longer stuff is also pretty sus.

also look to me mj squad making up numbers again. i thought the 86 rox thread woulda been the end of it, but they really have no shame

Umm, the data I provided goes year by year and provides the data we have, so I don’t know what you’re talking about. And it’s worth noting that in the years where it’s only a portion of the year (i.e. the Squared stuff), the Bulls actually did *worse* in the sampled data than in the full seasons. So, if anything, the data as presented is biased in favor of games where the Bulls were winning less.

Meanwhile, the only “career” stuff I included was a career playoff number, but the data set includes over 90% of Jordan’s playoff games, so it’s by no means “short stuff.”

Also, stop trying to stir things up. You appear to think a substantial portion of your role on the forum is to make snide posts referencing past discussions and suggesting that someone in the current discussion was wrong in those past discussions. It’s something you consistently do, and often in posts that have virtually no other content. That’s not productive posting in any way, and it needs to stop.

i got no clue what why youre bringing up squared. you cut 95 coz you knew bron would look as good over more years. you doing different calcs so for bron and mj so you can get mj ahead in a stat most ppl dont even care that much about and you then told me i needed to fact-check when u were tryna regress the 86 rox to the mean patrick mahomes style.

you seem to think your job is to make things up and fight with everyone before calling everyone else toxic for responding. so many different ppl have pointed out all the ways youre cappin, but you still act like everyone else is the problem