RGM GOAT Debate Thread (Part 2)

[levon]

I think I read that the difference between LeBron's total minutes played and Steph's is larger than the difference between Steph's and Ant Edwards. I'm sorry but Steph has zero claim to be GOAT.

[michaelm]

I think I read that the difference between LeBron's total minutes played and Steph's is larger than the difference between Steph's and Ant Edwards. I'm sorry but Steph has zero claim to be GOAT.

He isn’t the GOAT, but not because of FMVP awards.

I agree LeBron’s main claim is longevity. In general.

[michaelm]

Curry on the other hand was excluded from serious consideration for a number of years because he teamed up with Durant and they were too successful.

Nonsense like this is why I stopped bothering with you in the past and why I'm wondering what am I doing with my life. They were too successful? What the hell does that even mean? MVP is a regular season award that is mostly based on how well your team performs. Curry won his unanimous MVP award in a season where his team had the the most successful regular season of all time, but somehow the following team winning less games were too successful? Hell, LeBron's fourth MVP happened during a season when his team had a 27 game winning streak and they won 66 games, one less than the 2017 team did. Hell, the 2018 team didn't even win 60 games. So what the hell are you even talking about here? What part of your ass did you pull this one out of? This would be like me saying I didn't get the job because I did too well in the interview.

I think it’s pretty obvious that WarriorsGM has a valid point about this actually. I’ll articulate the point a bit differently. MVP voters do not take kindly to situations in which a team has two major superstars. Even if the team does incredibly well, the attitude is that the team *should* do that well and that neither one of the superstars should get full credit for how good the team is. This is a large part of why LeBron was a very distant 3rd in MVP voting in 2011, behind two players that everyone knew he was better than. And it’s why Steph and Durant barely got MVP votes in 2017, despite their team being dominant in the regular season. For LeBron, the 2011 Finals combined with a relatively unimpressive 2011 Heat regular season ended up resetting the narrative on this, though. After they lost in the Finals, people stopped really thinking that the Heat *should* just be dominant. And when combined with Wade taking a bit of a back seat to LeBron to avoid some of the issues that occurred with LeBron in the 2011 Finals, this opened up room for LeBron to be given credit for the Heat doing well in the regular season. This reset never happened for the Warriors, because they were dominant from the beginning, so people never moved off of the “This team *should* dominate, so I can’t give too much of the credit to either guy” sentiment. And if people never move off of that sentiment, then neither Steph nor Durant could have a realistic shot at MVP. I feel like people who followed basketball in these years should generally understand that this was the sentiment.

Granted, Steph wasn’t going to get MVP in 2018 anyways, since he missed too many games, and arguably the same thing is true in 2019. So the biggest effect of the above is probably just in 2017 in particular—where Steph was surely the most valuable player in the league, but couldn’t win MVP because he wasn’t being given due credit for the team’s success. Steph probably still was the most valuable player in the NBA in 2018 and 2019 when he played—one need only look at the team’s record when he did and didn’t play to see that—but missing 31 games and 13 games definitely put a damper on his MVP case, such that he was unlikely to get MVP even without the above-discussed issue.

I do think there’s a good chance LeBron doesn’t get the 2012 MVP if the Heat had won the 2011 Finals, though—because that reset of expectations would not have happened, and the Heat’s success would’ve been regarded as a continued expected result of the teaming up of superstars, and LeBron would’ve therefore gotten less credit for their success. Even the 2013 MVP might not have happened. Granted, I think that that would’ve been *wrong*, and merely a result of wrong-headed thinking and narrative-building by MVP voters. But I do think, realistically, this sort of thing is pretty obviously a major factor that affects MVP voting.

You're continued harping on FMVPs shows your insistence on using severely flawed measures. You think KD's FMVPs are a winning argument against Curry? Why should LeBron's be? LeBron outplayed Curry in the finals? But LeBron lost 3 out of 4 times.

Because for one thing, it demonstrates how much more LeBron meant to his team than Curry did. It tells me that there's a legitimate chance Curry could have been out with an injury in the KD Warriors teams, and they'd still probably win their championships. You can't replace LeBron on those teams with someone else and expect them to win. Hell, odds are the Cavs don't even make the playoffs without him.

The KD Warriors played a lot of games without Steph, and they gave us no reason to believe they were very good without Steph. They played 47 games without Steph in the regular season, and they went 24-23, with a -1.18 net rating per 100 possessions. They did play some playoff games without him in those years too, which were all against middling early-round opponents and were disproportionately home games, but if we add those to the mix, we have the Warriors going 29-24 with a +0.28 net rating per 100 possessions in games without Steph. And this is a team that won at a 65-win pace with a +10.18 net rating per 100 possessions in games with Steph. I think it’s very hard to look at things and conclude that Curry didn’t mean about as much as someone could mean to a team. They were not a terrible team without him, but his impact on them was demonstrably enormous and definitely looks GOAT-like. This is the strongest part of a case for Steph IMO.

The Curry off argument is similar to what is often propounded in LeBron’s favour, that his teams collapse when he is not on court or moves elsewhere. This is imo at least partly because the whole system is based on LeBron. While a heliocentric game plan does imo mean a ceiling on how good a team can be, LeBron ball was probably the game plan most likely to succeed for those teams slff Do imo however.

With Curry and the KD warriors it is likely imo a different game plan could be designed based more on KD which would be more successful with than they were when Steph was out for a few games, but Curry was the GSW system as their coach says, so the FMVP awards won by KD and Iguodala were won playing in a system based on Curry as you imply.

[levon]

I think I read that the difference between LeBron's total minutes played and Steph's is larger than the difference between Steph's and Ant Edwards. I'm sorry but Steph has zero claim to be GOAT.

He isn’t the GOAT, but not because of FMVP awards.

I agree LeBron’s main claim is longevity. In general.

LeBron's peak impact was also higher than Curry's and is the highest since we've had impact stats. Curry has neither peak nor longevity, and he doesn't even have the ring edge.

[WarriorGM]

I think I read that the difference between LeBron's total minutes played and Steph's is larger than the difference between Steph's and Ant Edwards. I'm sorry but Steph has zero claim to be GOAT.

He isn’t the GOAT, but not because of FMVP awards.

I agree LeBron’s main claim is longevity. In general.

LeBron's peak impact was also higher than Curry's and is the highest since we've had impact stats. Curry has neither peak nor longevity, and he doesn't even have the ring edge.

How did you determine this? RAPM? A black box stat so unclear someone else could run it and have Curry ahead?

In terms of changing the game LeBron isn't even close to the impact Curry has had.

[levon]

He isn’t the GOAT, but not because of FMVP awards.

I agree LeBron’s main claim is longevity. In general.

LeBron's peak impact was also higher than Curry's and is the highest since we've had impact stats. Curry has neither peak nor longevity, and he doesn't even have the ring edge.

How did you determine this? RAPM? A black box stat so unclear someone else could run it and have Curry ahead?

In terms of changing the game LeBron isn't even close to the impact Curry has had.

Okay, Steph wins the "changing the game of all time" title. Seems more authoritative than a "black box" stat, which isn't actually a black box by the way. https://www.nbastuffer.com/analytics101/regularized-adjusted-plus-minus-rapm/

[levon]

Also, I don't even buy the argument that "Steph changed the game". High volume 3 point shooting was on the way with or without Steph. Steph is 1-of-1 and literally nobody plays like him, yet you have heliocentric all-around perimeter players like Luka and Harden in the LeBron mould, with many more to come.

If I was gonna pick a current player with the most impact on the style of the game, it would be Jokic. And he also has a far better chance at becoming the GOAT.

[WarriorGM]

LeBron's peak impact was also higher than Curry's and is the highest since we've had impact stats. Curry has neither peak nor longevity, and he doesn't even have the ring edge.

How did you determine this? RAPM? A black box stat so unclear someone else could run it and have Curry ahead?

In terms of changing the game LeBron isn't even close to the impact Curry has had.

Okay, Steph wins the "changing the game of all time" title. Seems more authoritative than a "black box" stat, which isn't actually a black box by the way. https://www.nbastuffer.com/analytics101/regularized-adjusted-plus-minus-rapm/

I remember there was a site that calculated those numbers and it had Curry as the player with the highest RAPM. Then you have guys saying that conflicts with the calculation of the guy who introduced it. Going into the weeds the calculation seems to use weights and when that happens that introduces possible biases.

At its most basic RAPM is a derivative of raw plus-minus. In such a case as this I'd rather go with that and if we do, it is Curry who pretty clearly looks better.

I think I read that the difference between LeBron's total minutes played and Steph's is larger than the difference between Steph's and Ant Edwards. I'm sorry but Steph has zero claim to be GOAT.

The logic of total minutes played can be reversed in this argument. If LeBron was as good as Curry and played double the amount of time shouldn't we expect him to have double the number of championships? That he doesn't is further proof Curry is better.

[levon]

How did you determine this? RAPM? A black box stat so unclear someone else could run it and have Curry ahead?

In terms of changing the game LeBron isn't even close to the impact Curry has had.

Okay, Steph wins the "changing the game of all time" title. Seems more authoritative than a "black box" stat, which isn't actually a black box by the way. https://www.nbastuffer.com/analytics101/regularized-adjusted-plus-minus-rapm/

I remember there was a site that calculated those numbers and it had Curry as the player with the highest RAPM. Then you have guys saying that conflicts with the calculation of the guy who introduced it. Going into the weeds the calculation seems to use weights and when that happens that introduces possible biases.

At its most basic RAPM is a derivative of raw plus-minus. In such a case as this I'd rather go with that and if we do, it is Curry who pretty clearly looks better.

I think I read that the difference between LeBron's total minutes played and Steph's is larger than the difference between Steph's and Ant Edwards. I'm sorry but Steph has zero claim to be GOAT.

The logic of total minutes played can be reversed in this argument. If LeBron was as good as Curry and played double the amount of time shouldn't we expect him to have double the number of championships? That he doesn't is further proof Curry is better.

It's of no surprise to me that you prefer stats in which Curry comes out on top. I'm sure you'll figure out the correct set of prior weights that'll produce the right outcome once you run your own ridge regression.

[lessthanjake]

LeBron's peak impact was also higher than Curry's and is the highest since we've had impact stats. Curry has neither peak nor longevity, and he doesn't even have the ring edge.

How did you determine this? RAPM? A black box stat so unclear someone else could run it and have Curry ahead?

In terms of changing the game LeBron isn't even close to the impact Curry has had.

Okay, Steph wins the "changing the game of all time" title. Seems more authoritative than a "black box" stat, which isn't actually a black box by the way. https://www.nbastuffer.com/analytics101/regularized-adjusted-plus-minus-rapm/

RAPM as a general concept isn’t a black box. But the exact details of how it is run are, because there’s a lot of decisions that have to be made about how to run it and what to adjust for. And different ways of running it end up leading to different results. Just a few examples of these sorts of decisions:

- Do you use some sort of prior? Using a prior should result in the RAPM stabilizing better, particularly in smaller samples. But a prior will inherently introduce bias by favoring and disfavoring certain players. If you do you use a prior, what is your prior? Do you use some sort of basic prior based on things like minutes played and player accolades? Do you use a more complicated prior using box and/or tracking data? If you do the box and/or tracking data prior, what data are you using and what is the exact formula? Do you instead use prior years’ RAPM as the prior for each year? If you do that, you end up not adequately catching rapid increases or decreases in a player’s quality, and often end up with lagged effects in RAPM (for instance, if you use prior years’ RAPM as a prior, it ends up telling you LeBron’s most impactful regular season was 2011). If you do use prior years’ RAPM as a prior, how many years do you go back and how do you weigh those years?

- How do you deal with players that played really low minutes? Do you just take them out of the equation entirely? Or do you leave them in, even though the results with them will be really noisy and that could have large knock-on effects on other players’ RAPM? If you do take them out of the equation, what minutes cut-off do you use for this?

- Do you adjust for the rubber-band effect—i.e. the fact that teams tend to do worse when they’re ahead and better when they’re behind? If you do adjust for it, what exact adjustment are you using for it?

- Do you do a “luck” adjustment? If so, what are you counting as luck? Just free throws? Or do you include three-point percentage in the luck adjustment too?

- Do you include some sort of adjustment for garbage time? If so, what adjustment are you making?

- What time horizon do you look at? Shorter timeframes are noisier, but longer timeframes run into the issue of not accounting for players becoming better or worse over time (i.e. for instance, it may not be fair to Shaq’s teammates in Phoenix to be adjusting for him based on a time horizon that includes his impact in his Lakers years).

And those are just some things off the top of my head. It’s virtually never entirely clear what any particular RAPM source is doing for many of these things. And changing these sorts of decisions can and will change the results quite a lot. So, just as one example, EPM uses RAPM + a box/tracking prior, and it has Steph’s peak year well above LeBron’s peak year. But other measures are the opposite. Do we know what is right?

Also, when trying to make a comparison like this, it’s actually by no means clear that it’s methodologically sound to compare values across different years. Different measures are scaled in different ways, but each measure will have its RAPM values inherently be scaled within its own time horizon, making comparisons outside that time horizon a bit tricky. For instance, let’s hypothetically say we use Metric A, and that metric had LeBron at +9 RAPM in 2010 and had Steph at +8 RAPM in 2017. Do we know from this that Metric A thinks LeBron had more impact in 2010 than Steph did in 2017? Not really, actually. Because Metric A probably scales the RAPM values for each year based on standard deviations within that year. So it might tell us that LeBron was more standard deviations above the mean in 2010 than Steph was in 2017, but that doesn’t necessarily mean Steph was less impactful in absolute terms. And, again, how these metrics are scaled is typically a black box, making it really hard to meaningfully compare values across seasons.

[levon]

How did you determine this? RAPM? A black box stat so unclear someone else could run it and have Curry ahead?

In terms of changing the game LeBron isn't even close to the impact Curry has had.

Okay, Steph wins the "changing the game of all time" title. Seems more authoritative than a "black box" stat, which isn't actually a black box by the way. https://www.nbastuffer.com/analytics101/regularized-adjusted-plus-minus-rapm/

RAPM as a general concept isn’t a black box. But the exact details of how it is run are, because there’s a lot of decisions that have to be made about how to run it and what to adjust for. And different ways of running it end up leading to different results. Just a few examples of these sorts of decisions:

- Do you use some sort of prior? Using a prior should result in the RAPM stabilizing better, particularly in smaller samples. But a prior will inherently introduce bias by favoring and disfavoring certain players. If you do you use a prior, what is your prior? Do you use some sort of basic prior based on things like minutes played and player accolades? Do you use a more complicated prior using box and/or tracking data? If you do the box and/or tracking data prior, what data are you using and what is the exact formula? Do you instead use prior years’ RAPM as the prior for each year? If you do that, you end up not adequately catching rapid increases or decreases in a player’s quality, and often end up with lagged effects in RAPM (for instance, if you use prior years’ RAPM as a prior, it ends up telling you LeBron’s most impactful regular season was 2011). If you do use prior years’ RAPM as a prior, how many years do you go back and how do you weigh those years?

- How do you deal with players that played really low minutes? Do you just take them out of the equation entirely? Or do you leave them in, even though the results with them will be really noisy and that could have large knock-on effects on other players’ RAPM? If you do take them out of the equation, what minutes cut-off do you use for this?

- Do you adjust for the rubber-band effect—i.e. the fact that teams tend to do worse when they’re ahead and better when they’re behind? If you do adjust for it, what exact adjustment are you using for it?

- Do you do a “luck” adjustment? If so, what are you counting as luck? Just free throws? Or do you include three-point percentage in the luck adjustment too?

- Do you include some sort of adjustment for garbage time? If so, what adjustment are you making?

- What time horizon do you look at? Shorter timeframes are noisier, but longer timeframes run into the issue of not accounting for players becoming better or worse over time (i.e. for instance, it may not be fair to Shaq’s teammates in Phoenix to be adjusting for him based on a time horizon that includes his impact in his Lakers years).

And those are just some things off the top of my head. It’s virtually never entirely clear what any particular RAPM source is doing for many of these things. And changing these sorts of decisions can and will change the results quite a lot. So, just as one example, EPM uses RAPM + a box/tracking prior, and it has Steph’s peak year well above LeBron’s peak year. But other measures are the opposite. Do we know what is right?

Also, when trying to make a comparison like this, it’s actually by no means clear that it’s methodologically sound to compare values across different years. Different measures are scaled in different ways, but each measure will have its RAPM values inherently be scaled within its own time horizon, making comparisons outside that time horizon a bit tricky. For instance, let’s hypothetically say we use Metric A, and that metric had LeBron at +9 RAPM in 2010 and had Steph at +8 RAPM in 2017. Do we know from this that Metric A thinks LeBron had more impact in 2010 than Steph did in 2017? Not really, actually. Because Metric A probably scales the RAPM values for each year based on standard deviations within that year. So it might tell us that LeBron was more standard deviations above the mean in 2010 than Steph was in 2017, but that doesn’t necessarily mean Steph was less impactful in absolute terms. And, again, how these metrics are scaled is typically a black box, making it really hard to meaningfully compare values across seasons.

You're welcome to normalize by year, go by something as simple as percentiles, or the like. I tend to go 3 year minimum with RAPM, but 5 years is also fine and typically captures a peak.

Any critique of RAPM or EPM or impact stat X pales in comparison to the vibes argument of the poster I was replying to. Of course there is no objective measure. All models are wrong but some are useful, etc.

Also source on EPM favoring Steph?

[michaelm]

He isn’t the GOAT, but not because of FMVP awards.

I agree LeBron’s main claim is longevity. In general.

LeBron's peak impact was also higher than Curry's and is the highest since we've had impact stats. Curry has neither peak nor longevity, and he doesn't even have the ring edge.

How did you determine this? RAPM? A black box stat so unclear someone else could run it and have Curry ahead?

In terms of changing the game LeBron isn't even close to the impact Curry has had.

I said in general deliberately and intended to mean not in comparison with Curry. LeBron’s case against Jordan is mainly longevity imo.

Fine with me for people to have LeBron as superior to Curry, he obviously has more individual attributes. I will continue to be happy with Curry having led the team I suppprt to 4 titles, including a team which was possibly the best of all time.

[WarriorGM]

Okay, Steph wins the "changing the game of all time" title. Seems more authoritative than a "black box" stat, which isn't actually a black box by the way. https://www.nbastuffer.com/analytics101/regularized-adjusted-plus-minus-rapm/

I remember there was a site that calculated those numbers and it had Curry as the player with the highest RAPM. Then you have guys saying that conflicts with the calculation of the guy who introduced it. Going into the weeds the calculation seems to use weights and when that happens that introduces possible biases.

At its most basic RAPM is a derivative of raw plus-minus. In such a case as this I'd rather go with that and if we do, it is Curry who pretty clearly looks better.

I think I read that the difference between LeBron's total minutes played and Steph's is larger than the difference between Steph's and Ant Edwards. I'm sorry but Steph has zero claim to be GOAT.

The logic of total minutes played can be reversed in this argument. If LeBron was as good as Curry and played double the amount of time shouldn't we expect him to have double the number of championships? That he doesn't is further proof Curry is better.

It's of no surprise to me that you prefer stats in which Curry comes out on top. I'm sure you'll figure out the correct set of prior weights that'll produce the right outcome once you run your own ridge regression.

If you cannot explain something and you want to get to the truth you must develop tools that do. If there is a cure for a disease that works almost automatically but you cannot explain it and conventional wisdom says it shouldn't work do you say it isn't a cure? If a player is winning contrary to all conventional wisdom what do you do? Insist on the conventional wisdom?

If I favor certain measures over others it's probably because they are the best in our current toolkit that explain what Curry, contrary to expectations, has done. They are the ones that come closest to the truth. Curry is the kind of anomaly that can demand such a shift.

Indeed it is the extreme outlier results of Curry defying all established dogma that most simply and eloquently argue that he is the greatest.

[lessthanjake]

Okay, Steph wins the "changing the game of all time" title. Seems more authoritative than a "black box" stat, which isn't actually a black box by the way. https://www.nbastuffer.com/analytics101/regularized-adjusted-plus-minus-rapm/

RAPM as a general concept isn’t a black box. But the exact details of how it is run are, because there’s a lot of decisions that have to be made about how to run it and what to adjust for. And different ways of running it end up leading to different results. Just a few examples of these sorts of decisions:

- Do you use some sort of prior? Using a prior should result in the RAPM stabilizing better, particularly in smaller samples. But a prior will inherently introduce bias by favoring and disfavoring certain players. If you do you use a prior, what is your prior? Do you use some sort of basic prior based on things like minutes played and player accolades? Do you use a more complicated prior using box and/or tracking data? If you do the box and/or tracking data prior, what data are you using and what is the exact formula? Do you instead use prior years’ RAPM as the prior for each year? If you do that, you end up not adequately catching rapid increases or decreases in a player’s quality, and often end up with lagged effects in RAPM (for instance, if you use prior years’ RAPM as a prior, it ends up telling you LeBron’s most impactful regular season was 2011). If you do use prior years’ RAPM as a prior, how many years do you go back and how do you weigh those years?

- How do you deal with players that played really low minutes? Do you just take them out of the equation entirely? Or do you leave them in, even though the results with them will be really noisy and that could have large knock-on effects on other players’ RAPM? If you do take them out of the equation, what minutes cut-off do you use for this?

- Do you adjust for the rubber-band effect—i.e. the fact that teams tend to do worse when they’re ahead and better when they’re behind? If you do adjust for it, what exact adjustment are you using for it?

- Do you do a “luck” adjustment? If so, what are you counting as luck? Just free throws? Or do you include three-point percentage in the luck adjustment too?

- Do you include some sort of adjustment for garbage time? If so, what adjustment are you making?

- What time horizon do you look at? Shorter timeframes are noisier, but longer timeframes run into the issue of not accounting for players becoming better or worse over time (i.e. for instance, it may not be fair to Shaq’s teammates in Phoenix to be adjusting for him based on a time horizon that includes his impact in his Lakers years).

And those are just some things off the top of my head. It’s virtually never entirely clear what any particular RAPM source is doing for many of these things. And changing these sorts of decisions can and will change the results quite a lot. So, just as one example, EPM uses RAPM + a box/tracking prior, and it has Steph’s peak year well above LeBron’s peak year. But other measures are the opposite. Do we know what is right?

Also, when trying to make a comparison like this, it’s actually by no means clear that it’s methodologically sound to compare values across different years. Different measures are scaled in different ways, but each measure will have its RAPM values inherently be scaled within its own time horizon, making comparisons outside that time horizon a bit tricky. For instance, let’s hypothetically say we use Metric A, and that metric had LeBron at +9 RAPM in 2010 and had Steph at +8 RAPM in 2017. Do we know from this that Metric A thinks LeBron had more impact in 2010 than Steph did in 2017? Not really, actually. Because Metric A probably scales the RAPM values for each year based on standard deviations within that year. So it might tell us that LeBron was more standard deviations above the mean in 2010 than Steph was in 2017, but that doesn’t necessarily mean Steph was less impactful in absolute terms. And, again, how these metrics are scaled is typically a black box, making it really hard to meaningfully compare values across seasons.

You're welcome to normalize by year, go by something as simple as percentiles, or the like. I tend to go 3 year minimum with RAPM, but 5 years is also fine and typically captures a peak.

Any critique of RAPM or EPM or impact stat X pales in comparison to the vibes argument of the poster I was replying to. Of course there is no objective measure. All models are wrong but some are useful, etc.

Also source on EPM favoring Steph?

The source on what I said about EPM is that, if you look at the EPM website, Steph peaks out at +10.5 in actual EPM, while LeBron peaks out at +9.2.

(Note: There’s now two versions of EPM, and the “actual EPM” version is the long-standing version, while the other version is a new thing that’s designed to prospectively predict how a player will do in the future, rather than retrospectively evaluating how they did in the past. Obviously, for our purposes, the retrospective one is the most relevant one).

[MavsDirk41]

I think I read that the difference between LeBron's total minutes played and Steph's is larger than the difference between Steph's and Ant Edwards. I'm sorry but Steph has zero claim to be GOAT.

He isn’t the GOAT, but not because of FMVP awards.

I agree LeBron’s main claim is longevity. In general.

James main claim is longevity and a culmination of stats. His peak is up there too but he has won less than the other goat contenders and has played with more talent than Jordan.

[levon]

I remember there was a site that calculated those numbers and it had Curry as the player with the highest RAPM. Then you have guys saying that conflicts with the calculation of the guy who introduced it. Going into the weeds the calculation seems to use weights and when that happens that introduces possible biases.

At its most basic RAPM is a derivative of raw plus-minus. In such a case as this I'd rather go with that and if we do, it is Curry who pretty clearly looks better.



The logic of total minutes played can be reversed in this argument. If LeBron was as good as Curry and played double the amount of time shouldn't we expect him to have double the number of championships? That he doesn't is further proof Curry is better.

It's of no surprise to me that you prefer stats in which Curry comes out on top. I'm sure you'll figure out the correct set of prior weights that'll produce the right outcome once you run your own ridge regression.

If you cannot explain something and you want to get to the truth you must develop tools that do. If there is a cure for a disease that works almost automatically but you cannot explain it and conventional wisdom says it shouldn't work do you say it isn't a cure? If a player is winning contrary to all conventional wisdom what do you do? Insist on the conventional wisdom?

If I favor certain measures over others it's probably because they are the best in our current toolkit that explain what Curry, contrary to expectations, has done. They are the ones that come closest to the truth. Curry is the kind of anomaly that can demand such a shift.

Brother if you got a better statistical model, produce it for us, share it with the community, and have people evaluate. But don't pit narrative elements like "change the game" with numbers. LeBron arguably has many more narratives going for him. For instance, winning a title with three different franchises, coaches, and playing styles. His last title was won as a point guard in which he led the league in assists.

Imagine a player who can be your point guard, wing defender, spot up shooter, post defender, and roll big in the same game. That's LeBron at 40. This dude has played 8 more seasons worth of basketball than Curry. He's like the closest thing to if you maxed out every single achievement and stat in the skill tree. Seems pretty anomalous to me.

[levon]

RAPM as a general concept isn’t a black box. But the exact details of how it is run are, because there’s a lot of decisions that have to be made about how to run it and what to adjust for. And different ways of running it end up leading to different results. Just a few examples of these sorts of decisions:

- Do you use some sort of prior? Using a prior should result in the RAPM stabilizing better, particularly in smaller samples. But a prior will inherently introduce bias by favoring and disfavoring certain players. If you do you use a prior, what is your prior? Do you use some sort of basic prior based on things like minutes played and player accolades? Do you use a more complicated prior using box and/or tracking data? If you do the box and/or tracking data prior, what data are you using and what is the exact formula? Do you instead use prior years’ RAPM as the prior for each year? If you do that, you end up not adequately catching rapid increases or decreases in a player’s quality, and often end up with lagged effects in RAPM (for instance, if you use prior years’ RAPM as a prior, it ends up telling you LeBron’s most impactful regular season was 2011). If you do use prior years’ RAPM as a prior, how many years do you go back and how do you weigh those years?

- How do you deal with players that played really low minutes? Do you just take them out of the equation entirely? Or do you leave them in, even though the results with them will be really noisy and that could have large knock-on effects on other players’ RAPM? If you do take them out of the equation, what minutes cut-off do you use for this?

- Do you adjust for the rubber-band effect—i.e. the fact that teams tend to do worse when they’re ahead and better when they’re behind? If you do adjust for it, what exact adjustment are you using for it?

- Do you do a “luck” adjustment? If so, what are you counting as luck? Just free throws? Or do you include three-point percentage in the luck adjustment too?

- Do you include some sort of adjustment for garbage time? If so, what adjustment are you making?

- What time horizon do you look at? Shorter timeframes are noisier, but longer timeframes run into the issue of not accounting for players becoming better or worse over time (i.e. for instance, it may not be fair to Shaq’s teammates in Phoenix to be adjusting for him based on a time horizon that includes his impact in his Lakers years).

And those are just some things off the top of my head. It’s virtually never entirely clear what any particular RAPM source is doing for many of these things. And changing these sorts of decisions can and will change the results quite a lot. So, just as one example, EPM uses RAPM + a box/tracking prior, and it has Steph’s peak year well above LeBron’s peak year. But other measures are the opposite. Do we know what is right?

Also, when trying to make a comparison like this, it’s actually by no means clear that it’s methodologically sound to compare values across different years. Different measures are scaled in different ways, but each measure will have its RAPM values inherently be scaled within its own time horizon, making comparisons outside that time horizon a bit tricky. For instance, let’s hypothetically say we use Metric A, and that metric had LeBron at +9 RAPM in 2010 and had Steph at +8 RAPM in 2017. Do we know from this that Metric A thinks LeBron had more impact in 2010 than Steph did in 2017? Not really, actually. Because Metric A probably scales the RAPM values for each year based on standard deviations within that year. So it might tell us that LeBron was more standard deviations above the mean in 2010 than Steph was in 2017, but that doesn’t necessarily mean Steph was less impactful in absolute terms. And, again, how these metrics are scaled is typically a black box, making it really hard to meaningfully compare values across seasons.

You're welcome to normalize by year, go by something as simple as percentiles, or the like. I tend to go 3 year minimum with RAPM, but 5 years is also fine and typically captures a peak.

Any critique of RAPM or EPM or impact stat X pales in comparison to the vibes argument of the poster I was replying to. Of course there is no objective measure. All models are wrong but some are useful, etc.

Also source on EPM favoring Steph?

The source on what I said about EPM is that, if you look at the EPM website, Steph peaks out at +10.5 in actual EPM, while LeBron peaks out at +9.2.

(Note: There’s now two versions of EPM, and the “actual EPM” version is the long-standing version, while the other version is a new thing that’s designed to prospectively predict how a player will do in the future, rather than retrospectively evaluating how they did in the past. Obviously, for our purposes, the retrospective one is the most relevant one).

Let's take your advice of not comparing individual seasons. By my count, the number of times LeBron has been top 5 in EPM is 12. Steph 9. Does that match your count?

Let me post 5 year RAPM here. I think it's important to screenshot it so people can visualize the difference in ranks:

Steph:


LeBron:


I love Steph and was actively rooting for him to beat LeBron in the GS years. But like, come on now. IMO this is the best publicly available resource we have, and it's pretty clear. Until someone puts out something more useful according to consensus, I'm going to trust this and not worry too much about luck-adjusting and tweaking 5 year career samples.

I think I can only get Steph as high as 5 in my all-time list, and even that might be too high. LeBron is at worst second.

[sp6r=underrated]

I'm not going to read through this entire thread but the NBA GOAT debate to me is dead.

The gap between Lebron and # 2 is considerably larger than the gap between 2-5. Lebron has a great argument for best peak and best 4 year stretch but you can argue for others but he just overwhelms everyone else with elite season due to his early start followed by extreme durability.

And coming from the Jordan generation, I have to say we are embarrassing in denying this.

[Iwasawitness]

You can use odds just before the playoffs but pre-season odds do not have the biases introduced in the regular season. For example some players in certain situations do not take the regular season seriously. Second Cavaliers stint LeBron is one of them. He did not fear his opposition in the weak Eastern Conference so he did just enough to get by. Conversely some teams that perform very well in the regular season were exposed as paper tigers that no one since has taken seriously. The 60-win 2015 Atlanta Hawks for example. With their record that year they got 4 all-star selections. LeBron's Cavaliers swept the Hawks with basically the same roster they faced the Warriors with in the finals but you never hear about the win against the Hawks anymore lest it detract from the narrative that LeBron playing without Kyrie Irving and Kevin Love was such a massive disadvantage.

Starting this off really strong aren't we?

1. The things you're claiming apply here apply also to pre-season odds, only you're going off of much more inaccurate detail and have much less to work with. Again, there is no argument for pre-season odds being more reliable than pre-playoff odds. They just don't work like that. Again, the only reason you're using it is because Curry won two titles where his team wasn't the favorites going into the season. They just don't help your argument the way you think they do.

2. Your example doesn't work here. Atlanta was barely favored going into that series and there were a lot of things people didn't take into consideration. For starters, Atlanta's perimeter defense was pretty weak. Teague/Korver was a terrible defensive backcourt and Carroll was the primary defender for LeBron. And... well, it's LeBron, so Carroll got lit up the entire series. For another, their frontcourt couldn't handle Cleveland's frontcourt. Despite Cleveland having multiple injuries to key players, they still managed to be a significant matchup problem for the Hawks. In other words, this more than anything comes down to the greatness of LeBron, which if anything helps prove my point.

3. You are correct that some players do not take regular season seriously, but if anything that just helps further prove my point. People absolutely take the playoffs seriously, so someone who isn't favored to win actually winning looks a hell of a lot more impressive, depending on circumstance of course. There are seven games in each series, you get a much better idea of just how good teams really are in the postseason. You don't get that in the regular season.

Lots written down little really said. What is your criteria then?

Basic common sense things when determining how great any basketball player is... how great of a player are you, what kind of impact do you have on your team, were you able to do things the like of which we've never seen before, are you able to step up when it matters most, can you help bring out the best in others... things like this.

To go into further deep analysis, I look at what players accomplished in their careers and compare it to their peers, asking myself whether or not anyone's managed to pull these things off in the kind of fashion that player A did. In the case of LeBron, one of the biggest things I've always factored in were his all time playoff performances, something I brought up earlier. As mentioned before, LeBron was consistently dominant on a scale the likes of which we haven't seen since Michael Jordan, and tbh, I thought he was more dominant than him.

Does Curry usually lose in games where he had an "underwhelming" game? Curry's detractors characterize the 2015 finals as an underwhelming performance from Curry despite Curry having the highest scoring finals 4th quarters in a decade in that series. The Warriors won in 6.

So before I answer your question, this is once again another example of why I can't take your arguments seriously. Curry had the highest scoring fourth quarters in a finals in a decade? Why does that suddenly make up for everything else? Curry shot well below his typical averages in that series, shot his team out of game 2 and had trouble defending Delly of all people. His 2015 Finals are still underrated but they absolutely were underwhelming for him.

To answer your question, yes, he usually does. Game 2 of the 2015 Finals are a perfect example, but others include game 6 vs Toronto in 2019, game 4 vs Cleveland in 2017, multiple games in the 2016 Finals, and I could dive even deeper into the playoffs if I wanted.

Remembering only two games from Curry's numerous trips to the finals does not change the assessment. You are ill informed.

No, it just paints a picture about how good he's actually been in the finals compared to his peers. And if he's had so many, why is it that this is the second time you've addressed this point now and yet haven't given me a single game outside the two I named? Either you're ill informed or they don't exist. So which is it?

Steph at this point was playing without Durant. The Trail Blazers had just eliminated the Nuggets. The Warriors-Trail Blazers series was the Western Conference Finals. LeBron's series against the Raptors was an Eastern Conference Semi-Final. In any event this is just more support for using pre-season odds where the Cavaliers had +515 odds to win a championship and the Raptors had +15000 odds. The difference should have made expecting a sweep a distinct possibility. If you thought the result was an upset whatever method gave you that idea looks suspect.

Okay, again, need to break down another paragraph from you because there is just so much wrong with what you are saying here. Christ, why am I bothering with this...

1. It was a well known fact at this point that the Warriors were still an elite, finals contending team without Kevin Durant. Again, there's a reason the Warriors were favorites going into this. You're trying to create a narrative that doesn't exist here. No one in their right minds thought the Trail Blazers were winning that series. That wasn't the case with Toronto/Cleveland.

2. Why the hell does it matter if it was just the Eastern Semis? If anything that just further proves my point. You do realize the Cavaliers were the fourth seed going into that series, not the first, right? It sounds to me like you didn't know that, otherwise... I just can't fathom why you thought this would be a good point to bring up.

3. Beating the Nuggets at that point didn't really mean anything. You're acting like the Blazers beat a juggernaut or something. In reality they beat a team who needed seven games just to beat the Spurs... who were led by Demar Derozan and LeMarcus Aldridge. This was Jokic's first time being in the postseason mind you. Of course the Blazers beat them... they absolutely should have.

4. Again, no. Pre-season odds don't just make odds for who the most likely team is to win a championship... they also make predictions for what a teams record is going to be. And they are rarely ever correct in that regard. Houston, for example, went ten games over their predicted w/l record. Going by pre-season odds, a team like the 2011 Spurs who had the best record in the league overachieved... even though they were upset in the first round by the Grizzlies.

At some point you need to use logic. You can't just outright ignore it for agenda sake. If you're trying to argue for the favor of pre-season odds because of one season when we have a multitude of them proving they're bull ****, then all that tells me is that you could care less about what reality has to say if it doesn't align with what you believe in. That isn't how this works and that's why people from even your own fan base laugh at you on a regular basis.

All I see there is a lot of narrative. You seem to be suggesting Curry having chemistry with his team makes his victory less impressive while LeBron having a mess of a team makes his victory more impressive. I fail to see a reason why LeBron making moves that cause turmoil in his team should be a reason to be impressed.

Again, why am I even bothering at this point...

What did I say here that was all narrative and no facts? I said that Curry had a great team surrounding him with good chemistry. That's a fact. They were an elite defensive team. That's a fact. LeBron's teams were a mess. That's a fact. They struggled against the Pacers because they couldn't defend a team that had all shooters in the starting lineup. That's a fact. I'm not suggesting that Curry's victory is less impressive because of it. You made the argument that Curry's victory over the Celtics in 2022 is more impressive than LeBron's victory over them in 2018. Again, I'm explaining why that isn't the case, nor should it be. The Warriors were favored to beat the Celtics in the Finals. The Cavaliers on the other hand weren't favored to beat the Celtics. This isn't a narrative driven statement. It's based on reality, something you seem to have a lot of trouble living in.

Also, what the hell are you talking about? LeBron making moves that caused turmoil? LeBron isn't the one who asked for Kyrie Irving to be traded. Irving did that on his own and LeBron was blindsided by it. LeBron wasn't the one who asked for Wade to be traded and for them to get rid of Thomas and co. Koby Altman did that on his own.

2016 Western Conference Finals

Game 5
Curry 31 7 6
Klay 27 5 2

Game 6
Curry 31 10 9
Klay 41 4 0

Game 7
Curry 36 5 8
Klay 21 5 0

Klay's Game 6 performance in this series earned him the moniker Game 6 Klay. Deserved. Still to put things into perspective Curry had a higher gamescore in game 6 and it was a regular enough game from him that you for one don't bother to deem it worth remembering.

Again, READ THE THINGS I SAY.

From the very post that you quoted:

Now granted, I can't really speak for why Klay should get primary credit, because I don't think he should.

I mean... is it really this hard for you to read and comprehend what you are reading? I was never arguing that Klay should have gotten all the credit for the Warriors 3-1 comeback. In fact, I outright state that he shouldn't have. What part of this are you having such a hard time understanding? Do I need to spell it out for you?

Last quarter heroics in a game 7 your measure? Game 7 4th quarter had Curry tripling Klay's scoring 15 to 5. For the entire second half Curry likewise tripled Klay's scoring output 24 to 8.

Should be worth noting the Warriors were trailing by double digits somewhere around the halfway mark in both games 6 and 7.

But yeah: Game 6 Klay! Game 6 Klay!

Again, read what I said above. Feels like I'm talking to a brick wall at this point.

LeBron taking out the baby Thunder with his superteam is comparable to Curry taking out the mature Thunder with his OG teammates? Seems like a pattern.

1. I'm going to go on record now to say that the "Thunder were babies" narrative is some of the dumbest **** I've ever heard on this site and I'm sick of hearing it. Kevin Durant had been in the NBA for five seasons at this point, Westbrook had been in the NBA for four. They had both already been to the playoffs three times at this point, including a WCF appearance the previous season. They went through the Kobe Lakers and Duncan Spurs just to make it to the Finals. They had more postseason experience at this point than LeBron James did when he first made an NBA Finals at a younger age no less. Durant averaged 30 PPG in this series, Westbrook averaged 27/6/6. They did this while surrounded by postseason veterans such as Derrick Fisher, Kendrick Perkins, and Nick Collinson. This idea that the Thunder were an inexperienced squad with no idea how to play in the postseason is a straight up lie, made up by people who have no clue what they are talking about. They weren't babies and they played great, the Miami Heat just played better. This isn't complicated.

2. I knew you'd mention the "err LeBron had a superteam" nonsense. Yes, LeBron had a superteam. But ironically enough, Curry still had a much better team, with two other players making all NBA (which LeBron's super duper team didn't have) and a much, much better bench. But yeah, you're right, the two situations aren't comparable. Curry did it with a much better team against a worse version of the Thunder. The 2012 Thunder had a perfect mix of veterans and youth who had enough post-season experience to not be tempted by the bright lights (except for James Harden, but he's a career playoff choker so he doesn't count). So LeBron doing it is actually more impressive.

Similar situation to the 2016 finals isn't it? Should we start characterizing LeBron's victory as him benefitting massively from the Warriors playing down a starter the last three games?

By all means, do explain why the Rockets losing Chris Paul is similar to the Warriors losing their worst starting player who only averaged 15 MPG. I can't wait to hear this one. Oh and btw, the Cavaliers had lost Kevin Love for most of the series due to a concussion, but by all means, pretend like that didn't happen.

Is that the case? I thought the series was virtually tied at two games a piece and a tied score with a quarter to go when Durant suffered his injury and went down in game 5. You could hear the player commentators commenting on the series licking their chops expecting Curry and the Warriors to get their comeuppance at the hands of the Harden-Paul Rockets. Didn't come. Curry even game them a full half head start in the next game before he started scoring. Curry still won beating the Rockets on their home floor. Didn't even get to a Game 7.

That doesn't jog you memory? But you remember LeBron beating Victor Oladipo?

Why are you pretending like I don't remember that? I didn't mention it because I didn't need to. Again, the Rockets had a lot of issues that year and weren't the same team that they were in 2018. But I guess if I shut my brain off and don't think at all, I guess I can make sense of why the Warriors beating the Rockets whose best player took a grand total of one shot for the last seven minutes of that game is impressive. But that isn't how I do things, and you should stop doing that too.

This "reputation" you speak of was a concocted narrative just like the current one you champion that Curry has no case being called the greatest. Both are false. Shouldn't it bother you that the same people spinning the previous narrative are the same people spinning the current one?

Whether or not a reputation is deserved or not isn't relevant. The fact of the matter is, Curry had that reputation and it existed since 2015. You not agreeing with it doesn't matter. Sometimes they are true (like Curry having no GOAT case), sometimes they are false (like LeBron choking in 2011, which he didn't). But to answer your question... no not really. It all comes down to whether or not a narrative makes any sense. Going back to the LeBron choking in 2011 one for a minute, that one doesn't make sense to me because LeBron had already been through a postseason on numerous occasions and never once showed that he was scared of the spotlight. And when you factor in Dallas shutting him down in the regular season and actually watching the games to see how they played out and how they game planned for him, the more logical conclusion here is that Dallas simply planned their defense around slowing him down and succeeded in doing so.

Now I look at something like Curry having no GOAT case... and yes, that one makes sense to me. He doesn't have an argument for being a better player than Jordan, LeBron or Kareem, so that automatically disqualifies him. Again, this is all a matter of logic. It's not complicated.

Then please, by all means, show me the instances where Curry made it to a Finals without a stacked team filled with multiple all stars, multiple all defensive players, good solid roleplayers that fit him to a T and great coaching. LeBron has done this three times. Curry on the other hand has never done it.

2015? 2022?

Wrong on both accounts.

In 2015, both Klay and Draymond made the all star team. Draymond and Bogut both made the all defensive teams. They had the likes of Iggy (who was a sixth man of the year candidate that sesaon), Livingston, and Barbosa coming off the bench. Steve Kerr built an incredible system that helped bring out the best of his star players.

In 2022, Draymond and Wiggins made the all star team, Draymond made all defense, and they had the likes of Poole, Porter, Payton II and Kuminga coming off the bench.

For clarity, in what years do you believe did LeBron do this? It must also be pointed out that all-star and all-defense selections are largely a function of great regular season records. Leading teams to a great regular season record is punished by this evaluation method while regular season underperformance is rewarded.

So let me get this straight... regular season only matters when it comes to pre-season odds but they don't matter when it comes to awards. Please do tell me how that makes any sense, especially when we consider the fact that players often perform worst in the postseason than regular season?

But I digress.

2007, 2015 and 2018.

In 2007, LeBron didn't have a single teammate make the all star team or all defensive squad. And I don't need to explain why his bench was laughably bad.

2015 is a unique circumstance. On one hand, he did have only one teammate make the all star team in Irving, but he did have Kevin Love so it kind of counts. The problem is, Kevin Love didn't play in the Semis or ECF, or even the finals for that matter, so it's irrelevant. And as we know, Irving missed portions of the Semis and ECF. He did not have any teammate make the all defensive team, and by the time they got to the ECF their depth was pretty much shot. He did have a great coach in David Blatt though, who did make key adjustments in the finals to make the series closer than it had any right being. Hell, I'd even argue that he outcoached Kerr in that series.

2018 saw only one teammate of LeBron's make the all star team in Kevin Love. As with the previous two, no one on his team made all defense. And while they did have a decent amount of options, they just didn't have good chemistry at the time and were inexperienced in the postseason.

Funny all I heard during the Warriors years with KD is that forming a superteam disqualified Curry and Durant from being a serious MVP contender.

So now narratives are correct and we should take them as gospel? But you told me they were bull ****. So which is it?

The fact of the matter is neither Curry nor Durant ever finished top 4 in MVP voting while they played together. Neither of them ever got so much as a first place vote during that period. LeBron even when the Heatles backlash was at its fiercest in 2011 got multiple first place votes. The highest rank in MVP votes Steph got in that time was 5th in 2019. The highest Durant ranked got was 7th. In 2020 after luring yet another star player to join him LeBron finished 2nd in MVP voting and Anthony Davis finished 6th higher than Durant ever did on the Warriors. BSPN's ranking list even listed LeBron and Davis as the top two highest ranked players for 2020.

None of that makes sense. I'll leave it as a logical exercise to figure out under what circumstances anyone could come up with such BS.

This really isn't that hard to figure out.

The Durant/Curry Warriors were still very much deep in talent and didn't require their two best players to be at their best to win games. Klay and Draymond were still elite players at this point and they still had a great bench. It was inevitable that neither Curry or Durant would get serious MVP consideration, because they didn't deserve it. Hell, you even pointed out earlier that the Warriors won a WCF without Durant. That's how deep in talent they were.

That wasn't the case in Miami for LeBron, or LA for that matter. Miami wasn't built like Golden State was. They were very top heavy and still relied a lot on the play of LeBron to win them games. Without him they were pretty much screwed. Now granted, the 2013 Heat did manage to get a really good bench, but it still pales in comparison with what GS had and they were still very much dependent on LeBron. One of the main reasons they even won 66 games is because LeBron had a historically great season, something a lot of people don't take into account. Not to mention, Wade was declining as a player after 2011 and it accelerated in 2013 due to injury. LeBron pretty much had no choice but to continue to carry a good deal of the load like he did in Cleveland (albeit on a smaller scale).

I don't even know why I need to explain the LA one. That was very much a two superstar system that still relied heavily on LeBron. They did a very good job balancing out the rest of the team with high end role players and really good veterans, but in the end that's all they really were. They couldn't be anything more than that.

FMVPs demonstrate nothing except the state of mind of its 11 voters. They cannot prove anything on the court.

So you're saying their votes aren't consistent with what's happening on the court? You're saying it always goes to the least deserving player? Because that's the only way a statement like this can be valid.

Your argument that LeBron being more essential to chances of victory is more valuable is rational but it rewards putting all one's eggs in one basket and punishes teams and players for being more versatile and robust. It also fatally undermines your suggestion that Iguodala was deserving of an FMVP.

Huh? One of the major reasons LeBron even has four FMVPs is because he's versatile and robust. Not to mention, it doesn't undermine Iggy's FMVP award win. It proves exactly why he deserved it. He impacted the game in multiple ways and made the biggest impact on the Warriors winning.

So we've been told over and over again since LeBron first emerged. Why is it then that it feels less and less believable?

It doesn't... it only does to you because you don't want it to be the case. And as I've said numerous times... what you want to be the case and what actually is the case are two different things.

[WarriorGM]

It's of no surprise to me that you prefer stats in which Curry comes out on top. I'm sure you'll figure out the correct set of prior weights that'll produce the right outcome once you run your own ridge regression.

If you cannot explain something and you want to get to the truth you must develop tools that do. If there is a cure for a disease that works almost automatically but you cannot explain it and conventional wisdom says it shouldn't work do you say it isn't a cure? If a player is winning contrary to all conventional wisdom what do you do? Insist on the conventional wisdom?

If I favor certain measures over others it's probably because they are the best in our current toolkit that explain what Curry, contrary to expectations, has done. They are the ones that come closest to the truth. Curry is the kind of anomaly that can demand such a shift.

Brother if you got a better statistical model, produce it for us, share it with the community, and have people evaluate. But don't pit narrative elements like "change the game" with numbers. LeBron arguably has many more narratives going for him. For instance, winning a title with three different franchises, coaches, and playing styles. His last title was won as a point guard in which he led the league in assists.

Imagine a player who can be your point guard, wing defender, spot up shooter, post defender, and roll big in the same game. That's LeBron at 40. This dude has played 8 more seasons worth of basketball than Curry. He's like the closest thing to if you maxed out every single achievement and stat in the skill tree. Seems pretty anomalous to me.

LeBron is an anomaly in the sense of a physical specimen. But he is not the kind of anomaly Curry is. LeBron's success on the basketball court is in line with expectations. Curry's is not. Curry operates on a different far less understood paradigm. As such his upper limit is harder to gauge.

In my view 2016 should have made it clear to everybody that Curry was on the fast track to being a top 10 player ever. It shocks me people keep trying to deny what is so plainly obvious to me. Curry's performance was so stunning that year he locked up the MVP in February. He went unanimous and it was perfectly justifiable. He was 5 points away from a year that could be seen as clearly superior to Jordan's 1996 and it's quite likely it was getting injured that stopped him.

You want a statistical model? Curry led a team to the best single regular season record. Curry led a team to the best single season playoffs record. Curry led a team to 67+ wins in three consecutive seasons. Curry led a team that was at the absolute basement of the league to a championship within 2 seasons. Record. Record. Record. This is all one-of-one stuff. Records related to what is supposedly the goal of all players that play the game: winning. That is the definition of greatest.