What's new
By:
Filters

Let’s talk about NBA stats

N

NAOS_pt2

Well-Known Member
I’ve been meaning to get more analytical with basketball for a while now, so I’d appreciate this board’s insights on stats.

There’s obviously no best answer here. And it’s obvious that you always need at least a small set of different data points to get an adequate measure of a player’s impact on the game.

So which statistics do you think are best? If you have time to say why you think so, that’d be dope.

FYI, while I was away from the board for a while, I developed some data analytical skills that I’d like to use as contributions to this board, so I’ll be taking this conversation seriously. I’m thinking id like to produce some insights at the 1/3rd point, AS Break, and end of the season. Can probably take some requests, too.

Thanks for your insights.
 
For anyone getting into NBA analytics, I recommend familiarizing yourself with RAPM. It's not important to understand all the details about the math, but rather the process that gets you there. It is the core to all of the "all in one" type metrics you'll see. I wouldn't say that these numbers (or any other numbers) are the best, but understanding the process and how these things works give good perspective.

If you're more into the micro perspective, I think pbpstats.com is the most useful site out there. The interface can take some time to get used to, but once you get the hang of it there are a ton of insights to be gained from the tools.
 
NAOS_pt2 said:
I’ve been meaning to get more analytical with basketball for a while now, so I’d appreciate this board’s insights on stats.

There’s obviously no best answer here. And it’s obvious that you always need at least a small set of different data points to get an adequate measure of a player’s impact on the game.

So which statistics do you think are best? If you have time to say why you think so, that’d be dope.

FYI, while I was away from the board for a while, I developed some data analytical skills that I’d like to use as contributions to this board, so I’ll be taking this conversation seriously. I’m thinking id like to produce some insights at the 1/3rd point, AS Break, and end of the season. Can probably take some requests, too.

Thanks for your insights.
Click to expand...
Dont have any requests but applaud the idea and any effort to bring something more than eyeball test data to discuss. I will try to do my part as well.
 
I've downloaded all player data from basketball-reference for last year. I'm gonna play around with this stuff for a bit. They don't have RAPTOR, but I can probably add it without much problem.
 
538 has frozen their site which produced RAPTOR and WAR. Dont know if those stats are dead now or if some paywall stat business bought them...
 
HermanG said:
538 has frozen their site which produced RAPTOR and WAR. Dont know if those stats are dead now or if some paywall stat business bought them...
Click to expand...
These kind of all-in-one stats are usually just calculated fields using other publicly available data. If we can figure out the methodology, then we can probably reproduce them.
 
A very simple lesson RAPM:

Image you have two urns and inside of these urns are chips of different colors and each color has a numeric value. You play a game where you take 5 chips from urn #1 and 5 chips from urn #2. The result of each round is the total count from #1 minus #2. Each round, you redraw 5 colors. Some may be the same, some may be different....but you write down which colors were pulled from each side and the final result of each round.

If you have enough rounds, you would eventually be able to estimate the value of each color using different kinds of regression techniques. It will take a LOT of rounds to get an accurate, but with enough of a sample size you will be able to have a good estimate and say things like if you pull a yellow chip you are expected to be +2 given that everything else is neutral. You've probably figured it out at this point, but in an NBA context the chips are different players, the rounds are possessions, and the score is scoreboard.

But what do you do when there are not enough rounds? This is where all your BPM, EPM, RPM, LEBRON, RAPTOR ect all come into play. An NBA season is not long enough for RAPM to stabilized. However, RAPM was reasonably stabilized over a very long sample size. We're talking 10+ years (I think the common stable set was a 14 year sample). Using a giant data set, we can be reasonably confident in the estimates. From there, we can use more standard numbers from the big data set to estimate RAPM. An estimate on an estimate? Yes, that's exactly right. All of these different adjusted +/- stats are just a different flavor of predicting RAPM. It's good to know the differences between them all because then you know what's producing those numbers. BPM uses numbers from the box score to predict RAPM. EPM uses box score numbers with a bayesian prior to predict RAPM. LEBRON introduces luck based +/-. RAPTOR incorporates some spectrum data to predict RAPM. I wouldn't say one is better than the other, but when you see these numbers disagree with each other it's good to know the different methods and why they disagree.
 
NAOS_pt2 said:
not gonna let this thread slide off the front page so quickly—especially without comments from @Handlogten's Heros @Saint Cy of JFC @infection @latin jazz @Ron Mexico etc.

(I'll save future @s for when this needs to be bumped again)
Click to expand...
I like numbers but am not like an advanced stats guru type. I like some of the fancy catch alls but when they explain how the numbers are derived it starts to sound like some magic Harry Potter ****. Just like a McRib I'm cool with the result but I don't necessarily want to know how you got there.

The stuff I understand well and like/believe in are pretty simple. True Shooting percentage is one of my go to stats I love. My favorite stat related thing to do is to look at +/- differential per 100 possessions... I think it uncovers some hidden gems... its obviously noisy on bad teams but even then it can help you find a guy to get excited about that you might not have thought about. Its the kind of thing that put me on to Christian Wood and Alec Caruso. I like to look at multi-year samples because it can get wonky though. Certain players show up multiple times on multiple teams you can get the picture that whatever it is they do... its helpful.

Some fun ones:

Lakers are +53.7 pts per 100 with Wood on the floor this year.

Tre Jones +50 - not sure the Sochan at pg thing is good based on this
AB is +35
Brogdon +21
KO is +18
 
Handlogten's Heros said:
I like numbers but am not like an advanced stats guru type. I like some of the fancy catch alls but when they explain how the numbers are derived it starts to sound like some magic Harry Potter ****. Just like a McRib I'm cool with the result but I don't necessarily want to know how you got there.

The stuff I understand well and like/believe in are pretty simple. True Shooting percentage is one of my go to stats I love. My favorite stat related thing to do is to look at +/- differential per 100 possessions... I think it uncovers some hidden gems... its obviously noisy on bad teams but even then it can help you find a guy to get excited about that you might not have thought about. Its the kind of thing that put me on to Christian Wood and Alec Caruso. I like to look at multi-year samples because it can get wonky though. Certain players show up multiple times on multiple teams you can get the picture that whatever it is they do... its helpful.

Some fun ones:

Lakers are +53.7 pts per 100 with Wood on the floor this year.

Tre Jones +50 - not sure the Sochan at pg thing is good based on this
AB is +35
Brogdon +21
KO is +18
Click to expand...
These ideas also create some conundrums... like what am I supposed to do with FVV who is a **** TS% guy and a high +/- guy?
 
I kinda hate what 'stat's have done to sports, and basketball is a big offender.

I watch games like a math problem now, especially if I am interested in the player doing well. I'm always just mentally ticking misses or makes what kind of shot, etc. it doesn't feel right. Like when the big man puts up 5 shots and they all come out but he keeps grabbing the misses and he finally puts it in. My reaction is usually "Sucks for the shooting percentage, nice rebounds though" Isn't that just a dumb thing to think instead of wow great effort? Surely I'm not the only one who sees the late close out 'contest' on defense, jump to put a hand up , and after the swish "Op, well now that guy was the closest defender, that's gonna hurt their defense stats" I could go on, but I think you get it.

And basketball compounds the problem because it seems like many teams and players are completely coasting through the regular season. Donovan Mitchell goes for 30 a night, plays amazing, is an all NBA level player. And then just gets smothered in the playoffs. Like did any of his VORP or LEBRON, BPM etc stats in the regular season give a single indication that he would **** his pants as soon as the playoffs started? And there are a million variables you could explore, because we basically default to "Well no one was trying in the regular season anyway, especially on defense" so what are we even learning with all these regular season stats?
 
Bawse Dawg said:
I kinda hate what 'stat's have done to sports, and basketball is a big offender.

I watch games like a math problem now, especially if I am interested in the player doing well. I'm always just mentally ticking misses or makes what kind of shot, etc. it doesn't feel right. Like when the big man puts up 5 shots and they all come out but he keeps grabbing the misses and he finally puts it in. My reaction is usually "Sucks for the shooting percentage, nice rebounds though" Isn't that just a dumb thing to think instead of wow great effort? Surely I'm not the only one who sees the late close out 'contest' on defense, jump to put a hand up , and after the swish "Op, well now that guy was the closest defender, that's gonna hurt their defense stats" I could go on, but I think you get it.

And basketball compounds the problem because it seems like many teams and players are completely coasting through the regular season. Donovan Mitchell goes for 30 a night, plays amazing, is an all NBA level player. And then just gets smothered in the playoffs. Like did any of his VORP or LEBRON, BPM etc stats in the regular season give a single indication that he would **** his pants as soon as the playoffs started? And there are a million variables you could explore, because we basically default to "Well no one was trying in the regular season anyway, especially on defense" so what are we even learning with all these regular season stats?
Click to expand...
Good post
 
Bawse Dawg said:
I kinda hate what 'stat's have done to sports, and basketball is a big offender.

I watch games like a math problem now, especially if I am interested in the player doing well. I'm always just mentally ticking misses or makes what kind of shot, etc. it doesn't feel right. Like when the big man puts up 5 shots and they all come out but he keeps grabbing the misses and he finally puts it in. My reaction is usually "Sucks for the shooting percentage, nice rebounds though" Isn't that just a dumb thing to think instead of wow great effort? Surely I'm not the only one who sees the late close out 'contest' on defense, jump to put a hand up , and after the swish "Op, well now that guy was the closest defender, that's gonna hurt their defense stats" I could go on, but I think you get it.

And basketball compounds the problem because it seems like many teams and players are completely coasting through the regular season. Donovan Mitchell goes for 30 a night, plays amazing, is an all NBA level player. And then just gets smothered in the playoffs. Like did any of his VORP or LEBRON, BPM etc stats in the regular season give a single indication that he would **** his pants as soon as the playoffs started? And there are a million variables you could explore, because we basically default to "Well no one was trying in the regular season anyway, especially on defense" so what are we even learning with all these regular season stats?
Click to expand...
I think it helps and hurts... I remember in HS starting out a game and hitting 2 threes ( I was supposed to throw into the post but the other team was in a zone and had clogged the paint)... third trip up the floor I shot another open one but missed. Time out and substitution... and my *** is on the bench. "We can't shoot threes all night".

Okay coach.

I think they tell part of the story and can be helpful... but they aren't the story itself.

The one thing I truly think is dumb is that hardly any players are willing to do the end of quarter heave. They need to make it so heaves aren't considered FG attempts unless they go in... heaves are sooooooo fun.... make basketball fun again.
 
Bawse Dawg said:
I kinda hate what 'stat's have done to sports, and basketball is a big offender.

I watch games like a math problem now, especially if I am interested in the player doing well. I'm always just mentally ticking misses or makes what kind of shot, etc. it doesn't feel right. Like when the big man puts up 5 shots and they all come out but he keeps grabbing the misses and he finally puts it in. My reaction is usually "Sucks for the shooting percentage, nice rebounds though" Isn't that just a dumb thing to think instead of wow great effort? Surely I'm not the only one who sees the late close out 'contest' on defense, jump to put a hand up , and after the swish "Op, well now that guy was the closest defender, that's gonna hurt their defense stats" I could go on, but I think you get it.

And basketball compounds the problem because it seems like many teams and players are completely coasting through the regular season. Donovan Mitchell goes for 30 a night, plays amazing, is an all NBA level player. And then just gets smothered in the playoffs. Like did any of his VORP or LEBRON, BPM etc stats in the regular season give a single indication that he would **** his pants as soon as the playoffs started? And there are a million variables you could explore, because we basically default to "Well no one was trying in the regular season anyway, especially on defense" so what are we even learning with all these regular season stats?
Click to expand...
Great post. Agreed.
 
Handlogten's Heros said:
I think it helps and hurts... I remember in HS starting out a game and hitting 2 threes ( I was supposed to throw into the post but the other team was in a zone and had clogged the paint)... third trip up the floor I shot another open one but missed. Time out and substitution... and my *** is on the bench. "We can't shoot threes all night".

Okay coach.

I think they tell part of the story and can be helpful... but they aren't the story itself.

The one thing I truly think is dumb is that hardly any players are willing to do the end of quarter heave. They need to make it so heaves aren't considered FG attempts unless they go in... heaves are sooooooo fun.... make basketball fun again.
Click to expand...
Yea it's also a larger scale than my post made it.

How teams plays is entirely informed by these stats now. Yes it does make way more sense for teams to shoot a million threes, but it doesn't make for a better product. This infests the playoffs as well, where every team bombs so many threes that if one team is hot and the other isn't the game is just over immediately, not fun or competitive at all. Three point variance feels like the deciding factor so often, and can lead to awful games, as the team getting blown out just keeps jacking threes on the off chance they will get hot instead of trying to chip away at a big lead.

Analytics just about killed baseball as well, and they are now trying to recover with rule changes. It's why football is forever the best sport. Fantasy football is the most played game in the world, obviously the stats matter, but they don't really change how teams play. It's always trying to outscheme, outcoach, and just beat the guy and the team in front of you. There is so much going on, that advanced stats just have very limited value. It's really about one thing:



8804533.gif
 
Handlogten's Heros said:
These ideas also create some conundrums... like what am I supposed to do with FVV who is a **** TS% guy and a high +/- guy?
Click to expand...

Easy, there are so many ways to impact the game besides TS%....and it's important to understand the context behind the shots that are producing that TS%.
 
Bawse Dawg said:
Yea it's also a larger scale than my post made it.

How teams plays is entirely informed by these stats now. Yes it does make way more sense for teams to shoot a million threes, but it doesn't make for a better product. This infests the playoffs as well, where every team bombs so many threes that if one team is hot and the other isn't the game is just over immediately, not fun or competitive at all. Three point variance feels like the deciding factor so often, and can lead to awful games, as the team getting blown out just keeps jacking threes on the off chance they will get hot instead of trying to chip away at a big lead.

Analytics just about killed baseball as well, and they are now trying to recover with rule changes. It's why football is forever the best sport. Fantasy football is the most played game in the world, obviously the stats matter, but they don't really change how teams play. It's always trying to outscheme, outcoach, and just beat the guy and the team in front of you. There is so much going on, that advanced stats just have very limited value. It's really about one thing:



8804533.gif
Click to expand...
make or miss league in many ways
 
Handlogten's Heros said:
The one thing I truly think is dumb is that hardly any players are willing to do the end of quarter heave. They need to make it so heaves aren't considered FG attempts unless they go in... heaves are sooooooo fun.... make basketball fun again.
Click to expand...
No you just dribble that ball for 3-4 seconds until there’s no more air. Maybe do some slam dribbles. The Donovan special. You don’t wanna look desperate. Keep your integrity.
 
You must log in or register to reply here.

Share this page

Top