Joel: unchained.

[Ferry Avenue]

Click on the twitter link, and read the whole thread.

That's pretty shocking, lol. Some people do act as if these stats are designed by god and not put together by people doing their best to define the concept of value. The worst part is, they can't be tested against anything. There's no way of determining how accurate any of these metrics are outside of individuals measuring them against their biases:

I think Ja Morant is good and metric confirms it = metric good
I think Coby White is good and metric refutes it = metric bad

Hopefully we're reaching the end of people taking them seriously. They're fun for fans to look at, and I admit I go to the advanced section of a player's bref page all the time, but I'm a nobody and no one takes my opinion on basketball seriously.

The first step in determining the value or importance of any metric is correlating it with winning, which is the goal of the game. If a metric correlates strongly with winning, it very likely represents something important in the game.

[Stanford]

The first step in determining the value or importance of any metric is correlating it with winning, which is the goal of the game. If a metric correlates strongly with winning, it very likely represents something important in the game.

All of these metrics purport to measure a player's impact on winning. You're not saying anything revelatory or measurable here. Correlates strongly with winning? What does that mean? Embiid and Jokic have essentially the same number of wins this year. Does that mean the difference in their BPM is essentially accurate because it correlates with winning at the same magnitude? And aren't team wins a factor of the team and not an individual player? How can you determine how much of a team win should be credited to one player versus a teammate?

[Mik317]

For once I am looking at this with a positive view

Jo lost it last year because he wasn't healthy. Played the most games of his career this year.

So maybe he takes losing it this year and uses it to become a better passer and more efficient...that way they cannot use that against him.

[mjkvol]

https://basketball.realgm.com/wiretap/266433/Nikola-Jokic-First-Ever-With-2000-1000-500-Season

Game, set, and match.

[Ferry Avenue]

The first step in determining the value or importance of any metric is correlating it with winning, which is the goal of the game. If a metric correlates strongly with winning, it very likely represents something important in the game.

All of these metrics purport to measure a player's impact on winning. You're not saying anything revelatory or measurable here. Correlates strongly with winning? What does that mean? Embiid and Jokic have essentially the same number of wins this year. Does that mean the difference in their BPM is essentially accurate because it correlates with winning at the same magnitude? And aren't team wins a factor of the team and not an individual player? How can you determine how much of a team win should be credited to one player versus a teammate?

I didn't say a division of credit among teammates could be accurately determined. I said a metric's importance is determined by its correlation with winning. All metrics are not created equal -- some have more meaning than others. It isn't just a "sea" of meaningless metrics out there, whose meaning is determined by Joe Shmoe's personal view of them. Meaning is determined by their relationship with the goal of the game -- winning. Joe Shmoe could think bench scoring for example is incredibly important, and that'll be determined by its correlation with winning, not by Joe Shmoe's personal belief.

[Stanford]

I said a metric's importance is determined by its correlation with winning.

Give me an example of how you would quantify that.

[Ferry Avenue]

I said a metric's importance is determined by its correlation with winning.

Give me an example of how you would quantify that.

For example the correlation between FG% and win percentage throughout the NBA in 2021-2022 is 0.63. By contrast the correlation between FT% and win percentage is 0.37. FG% explains nearly 40% of the variance in win percentage throughout the league, whereas FT% explains only 14% of it.

Now that's common sense, but it illustrates the point -- if Joe Shmoe comes along and tells you FT% is more important than FG%, that's belied by their correlations with winning. FG% is far more important than FT%. The matter doesn't rest on what Joe Shmoe thinks -- the matter rests on what's empirically related to winning in the league. That determines the importance of the statistic.

[Stanford]

For example the correlation between FG% and win percentage throughout the NBA in 2021-2022 is 0.63. By contrast the correlation between FT% and win percentage is 0.37. FG% explains nearly 40% of the variance in win percentage throughout the league, whereas FT% explains only 14% of it.

Now that's common sense, but it illustrates the point -- if Joe Shmoe comes along and tells you FT% is more important than FG%, that's belied by their correlations with winning. FG% is far more important than FT%. The matter doesn't rest on what Joe Shmoe thinks -- the matter rests on what's empirically related to winning in the league. That determines the importance of the statistic.

Could you link me to these figures?

[Ferry Avenue]

For example the correlation between FG% and win percentage throughout the NBA in 2021-2022 is 0.63. By contrast the correlation between FT% and win percentage is 0.37. FG% explains nearly 40% of the variance in win percentage throughout the league, whereas FT% explains only 14% of it.

Now that's common sense, but it illustrates the point -- if Joe Shmoe comes along and tells you FT% is more important than FG%, that's belied by their correlations with winning. FG% is far more important than FT%. The matter doesn't rest on what Joe Shmoe thinks -- the matter rests on what's empirically related to winning in the league. That determines the importance of the statistic.

Could you link me to these figures?

Sure:

https://www.nba.com/stats/teams/traditional/?PerMode=Totals&sort=PTS&dir=-1&Season=2021-22&SeasonType=Regular%20Season

The data are there -- I had the correlations calculated based on those data.

[brannigan73]

There is no shame if Embiid doesnt win the MVP award. He had an amazing season. Unfortunately so did 2 other guys. He does need to cut down on the turnovers though with Benamin gone he is now the main source of idiotic turnovers on this team. He has gotten better but not good enough. Last nite his 5 where pivotal in the loss. Look at Siakam the guy basically played point pf and had higher usage rate then Embiid and only turned it over 2 times.

[Stanford]

I'm obviously missing the point here.

The metrics we're talking about (BPM, VORP, LEBRON, etc.) are individual and outright claim to catch everything that makes a player valuable. They're presented in different ways - on a per minute basis, or relative to a replacement player, etc. - but they're all attempting to express the same thing.

We can look at team FG% and discover a strong correlation with team winning percentage, but what does that tell us about individual players other than the how often they make a field goal attempt? More importantly, how does it help improve team building? How does it help us determine which players are good and which player aren't? Is it relevant in choosing the MVP? What use is the knowledge of that correlation?

[Jailblazers7]

Just based on this years MVP convo, my prediction is Ja or Luka win it next year. Giannis has basically surpassed the MVP convo like Lebron did, Jokic will face a major advanced stats backlash, and Embiid will be a “prove it to me in the playoffs” narrative for NBA media.

[Ferry Avenue]

I'm obviously missing the point here.

The metrics we're talking about (BPM, VORP, LEBRON, etc.) are individual and outright claim to catch everything that makes a player valuable. They're presented in different ways - on a per minute basis, or relative to a replacement player, etc. - but they're all attempting to express the same thing.

We can look at team FG% and discover a strong correlation with team winning percentage, but what does that tell us about individual players other than the how often they make a field goal attempt? More importantly, how does it help improve team building? How does it help us determine which players are good and which player aren't? Is it relevant in choosing the MVP? What use is the knowledge of that correlation?

The validity of any statistic that purports to measure individual performance isolated from the surrounding team will be a function of sample size. In the NFL for example, a high or low passer rating in a single game can represent a whole host of variables external to the QB -- receiver drops, yards after the catch, pressure on the QB, facing a weak defense, etc. But a QB's passer rating over years of performance can be reliably attributed to his individual ability, due to the fact that when the sample of play reaches that size, the surrounding variables responsible for his performance have as they say "averaged out."

So when you're looking at LeBron James's career VORP for example, that's likely a reliable measure of his ability. Tyrese Haliburton's VORP over a single-season sample however may not be.

[eyeatoma]

LMAO I really got the GB going with the advanced stats stuff on the MVP thread.

[phillynative]

LMAO I really got the GB going with the advanced stats stuff on the MVP thread.

Lol i bet

[eyeatoma]

Read on Twitter

Read on Twitter

[Bum Adebayo]

**** **** ****

[Da Doctor]

The Ringerification of NBA media has ruined the MVP talk. VORP is more important than PPG, RPG, wins, and keeping your whole team afloat while a malcontent tries to sabotage it.

[Stanford]

He won a scoring title. That's pretty cool.

[Arsenal]

He won a scoring title. That's pretty cool.

Thankfully the NERDS who vote for MVP couldn't steal it from him for whatever arbitrary reason they could come up with.