RealGM

turk3d wrote:Check the boxscores. Most now give 3 pt shots made and shots attempted. Combine that with fga and fgm (you can do a quick calculation in your head) and you'll get a pretty good idea of what TS% or more importantly how the guy shot overall. TS% is nice if you want to do a comparison and it gives you just the one number to look at instead of having to do the calculations yourself.

Listen to the man.

I would say FG% has just about zero value. I'm quite suprised how many people fail to grasp basic math, and utilize this stat.

EFG% has more value, but is still extremly flawed, and I can't imagine ever using it.

TS% is flawed, but captures the scoring efficeincy well enough, and is the best option we have. It would be best if someone tracked each individual 'and 1', but an estimate atleast returns a reasonably accurate picture of scoring efficiency.

TS% in my opinion is pretty much the same as PPP...the formula may vary depending on whether FT's are weighted .5 or .44, and whether or not turnovers are counted, but generally TS% should be highly correlated with PPP.

I agree FG% is pretty useless. At the very least use eFG% instead.

Hendrix wrote:I would say FG% has just about zero value. I'm quite suprised how many people fail to grasp basic math, and utilize this stat.

EFG% has more value, but is still extremly flawed, and I can't imagine ever using it.

TS% is flawed, but captures the scoring efficeincy well enough, and is the best option we have. It would be best if someone tracked each individual 'and 1', but an estimate atleast returns a reasonably accurate picture of scoring efficiency.

How is EFG% flawed?

fallacy wrote:I just told you in my post. Having a raw, unadjusted, percentage of shots made is valuable. That would be like saying that raw assists and raw rebounds are worthless stats because we have ast% and reb%

That doesn't explain the value FG%. You just described what it measures. And called it valuable.

Raw rebounds are useless.

AST% and raw assists aren't really the same thing so they both have use. If I need to explain I can.

Rapcity_11 wrote:

fallacy wrote:I just told you in my post. Having a raw, unadjusted, percentage of shots made is valuable. That would be like saying that raw assists and raw rebounds are worthless stats because we have ast% and reb%

That doesn't explain the value FG%. You just described what it measures. And called it valuable.

Raw rebounds are useless.

AST% and raw assists aren't really the same thing so they both have use. If I need to explain I can.

I have been thinking about this lately. FG% can be useful as a classifier for players. You can make some inference from FG%.

For example, the average mid-range jumper is about 37% (3-pt is even lower obviously). The average layup is around 60%. The average dunk is around 90%. The average hook shot is around 50%.

If you use eFG%, you can't tell the difference between a guy who gets most of his points off inside shots versus 3-pt shots. For example, say you have a guy with 60% eFG%. Is that guy a good 3-pt shooter or does he get a lot of layups and dunks? Now, if you tell me he has a 40% FG%, I can tell you right away he is most likely a jump shooter.

This is not to say that a higher FG% makes one player better than another one. The eFG% is more useful for determining player value. What I'm saying is that the FG% as a *classifier* of player "type" can be "useful" (depending on how you use it).

Why bother infering that (and not always being right) when the shgot distribution data is readily available and more accurate?

tsherkin wrote:Why bother infering that (and not always being right) when the shgot distribution data is readily available and more accurate?

If that is available, absolutely. I'm just saying that if all you had was eFG% and FG%, they each give different kinds of information, and each can be useful. In theory.

I don't have a dog in this fight. I think eFG% is more important for player evaluation. But at the same time, it's not correct to say FG% doesn't give you anything. Because I can clearly construct a case where it does give you something. It would be intellectually dishonest to say otherwise in my opinion.

EvanZ wrote:If that is available, absolutely. I'm just saying that if all you had was eFG% and FG%, they each give different kinds of information, and each can be useful. In theory.

It's just an inference, though. A 40% shooter may just be bad. Or may be slashing for floaters and mid-range Js, which isn't exactly the same as being a "jump shooter." Or they could be a shooter, bombing lots of threes.

It's just so imprecise that the inference isn't valuable.

tsherkin wrote:

EvanZ wrote:If that is available, absolutely. I'm just saying that if all you had was eFG% and FG%, they each give different kinds of information, and each can be useful. In theory.

It's just an inference, though. A 40% shooter may just be bad. Or may be slashing for floaters and mid-range Js, which isn't exactly the same as being a "jump shooter." Or they could be a shooter, bombing lots of threes.

It's just so imprecise that the inference isn't valuable.

Let me ask you this. Which do you expect to be higher, the correlation between eFG% and rebounding or FG% and rebounding?

If you can honestly answer that question, I think I will have made my point.

As far as being "just an inference"...well, yeah. That's kind of a thing that statisticians do quite a lot of.

And "so imprecise"? You tell me which of these is more predictable in terms of the type of player near the top:

FG%
http://www.basketball-reference.com/pla ... _by=fg_pct

eFG%
http://www.basketball-reference.com/pla ... by=efg_pct

See Nathan's post a while back (point #4):

http://www.thebasketballdistribution.co ... death.html

But what is the value of being able to make a general inference about a player based on their FG%? I don't see any value in that. You could probably make a more accurate classification of a player based on rebound rate and I don't see any value in that either.

tsherk did bring up a good point though, that FG% used in shot distribution data is quite useful.

I'm just making the point that it gives some information that eFG% alone does not. To the extent that any new information can be useful, therein lies the potential for value.

Say you're down by 1 with the last possession, do you give it to the guy with the highest eFG% every time?


Say You have two players with the same eFG%, but one has a higher FG%. Does that inform your choice between the two?

Evan, I completely agree with you on that subject. As all information the statistical information about the shooting rates need proper interpretation. Like your example, for instance:

EvanZ wrote:Say You have two players with the same eFG%, but one has a higher FG%. Does that inform your choice between the two?

Differences in the FG% are pointing to a different player type. Typical a player who plays closer to the basket or scores in a fashion by attacking the basket has a higher FG%, while a better outside shooter can catch up in terms of eFG% via the additional point coming from the 3pt shot. FG%, 3P%, FT%, TS% and eFG% can be used to classify players. The inside big has typical a lower 3p% (most times 0, because 0 3pt shots attempts), a lower FT%, while the TS% and eFG% are close or even equal to the FG% (or TS% lies between the FG% and FT%). A pure 3pt shooter has typical a FG% which is close to his 3P% while his FT% is above league average. TS% and eFG% are commonly close together for such player, because he usually doesn't generate a lot FT attempts. The high efficient scorers have usually FG%, 3PT% and FT% which are closer to the league average, and yet the eFG% and the TS% is higher, while the TS% is typically higher than the eFG%, this player either attacks the rim while creating FT opportunities or is taking the 3pt shot. So, by just looking at the shooting numbers, we can actually get a pretty good idea how the player is acting on the court without watching him all the minutes. And that is only possible, if we take all those information into account.

Rapcity_11 wrote:

Hendrix wrote:I would say FG% has just about zero value. I'm quite suprised how many people fail to grasp basic math, and utilize this stat.

EFG% has more value, but is still extremly flawed, and I can't imagine ever using it.

TS% is flawed, but captures the scoring efficeincy well enough, and is the best option we have. It would be best if someone tracked each individual 'and 1', but an estimate atleast returns a reasonably accurate picture of scoring efficiency.

How is EFG% flawed?

It's flawed in the sense that it doesn't accurately tell you how efficient a player is. Ft's count as possessions. Obviously it makes no attempt to do so, so it is not flawed if used in the proper context; however, I'm not really sure why anyone ever would want to use a context for scoring where you arn't trying to determine how efficient the player is in the totality of their scoring.


Judging how efficient a player is by EFG% is kind of like the reverse of judging how efficient a player is based on their ft%. Obviously EFG% is going to be closer to the truth as players tend to take more fga's than fta's. However both ft% and efg% omit a big portion of game.

Hendrix wrote:I'm not really sure why anyone ever would want to use a context for scoring where you arn't trying to determine how efficient the player is in the totality of their scoring.

As Evan wanted to point out, the information about FG% or eFG% can be used to gain information about the player type. Such things can be very useful in order to determine the overall value a player can have to a team success. It should be part of the player evaluation to understand how a player acts on the court, and the different shooting stats can help you by that.

mysticbb wrote:

Hendrix wrote:I'm not really sure why anyone ever would want to use a context for scoring where you arn't trying to determine how efficient the player is in the totality of their scoring.

As Evan wanted to point out, the information about FG% or eFG% can be used to gain information about the player type. Such things can be very useful in order to determine the overall value a player can have to a team success. It should be part of the player evaluation to understand how a player acts on the court, and the different shooting stats can help you by that.

Thanks, mystic. I didn't think people would have such trouble with my point, but that's how these things go sometimes, as you well know.

Hendrix wrote:It's flawed in the sense that it doesn't accurately tell you how efficient a player is. Ft's count as possessions. Obviously it makes no attempt to do so, so it is not flawed if used in the proper context; however, I'm not really sure why anyone ever would want to use a context for scoring where you arn't trying to determine how efficient the player is in the totality of their scoring.


Judging how efficient a player is by EFG% is kind of like the reverse of judging how efficient a player is based on their ft%. Obviously EFG% is going to be closer to the truth as players tend to take more fga's than fta's. However both ft% and efg% omit a big portion of game.

No, all this means is your interpretation of it is wrong. It's not meant to be an all encompassing measure of efficiency. It's only supposed to make up for the added value of the 3 point shot. It's just one of many tools that should be used in player evaluation.

EvanZ wrote:
Thanks, mystic. I didn't think people would have such trouble with my point, but that's how these things go sometimes, as you well know.

I don't think anybody is having troubling following your point. What you're saying makes complete sense. However I still fail to see any value in it because it only makes sense if we didn't have access to any game film or shot distribution charts. But we do so I'm not seeing any kind of real application.

Say you're down by 1 with the last possession, do you give it to the guy with the highest eFG% every time?

Say You have two players with the same eFG%, but one has a higher FG%. Does that inform your choice between the two?

These types of scenarios depend entirely on the context of the game and can never come down to only which guy has the higher %'s. Things like how players on your team are playing, who is defending on the other team, how they are defending, how much time is left, etc, matter. It's not baseball.

I think you're just grasping at straws in order to find a loose correlation between FG% and types of players. Like I mentioned earlier, looking at rebound rate offers a similar result.

Rapcity_11 wrote:
I don't think anybody is having troubling following your point. What you're saying makes complete sense. However I still fail to see any value in it because it only makes sense if we didn't have access to any game film or shot distribution charts. But we do so I'm not seeing any kind of real application.

So the target would seem to be moving. I thought we were discussing eFG% (or TS%) vs. FG%, not eFG% + shot distributions + game film + rebounding rate vs. FG%. That would be tough for any one stat to take on, don't you think? If you need all that to beat FG%, that should tell you something.

And by the way, until a few years ago, all people had to go by was box scores. There wasn't Synergy, hoopdata, etc. FG% is less useful now than it was, to be sure. But I feel like you would be better telling people the whole truth if you're trying to convince them to stop using it. I think people will be more likely to want to hear it from someone who isn't trying to condescend or patronize them and tell them how worthless their currently favorite stat is.

EvanZ wrote:
So the target would seem to be moving. I thought we were discussing eFG% (or TS%) vs. FG%, not eFG% + shot distributions + game film vs. FG%.

No moving target. This is about the use of eFG%, FG% and TS% as metrics in basketball analysis. Under the umbrella of any other analysis methods. So what is the value of FG% in the entirety of basketball analysis.

And by the way, until a few years ago, all people had to go by was box scores. There was Synergy, hoopdata, etc. FG% is less useful now than it was, to be sure.

Right, and we're in the now currently. So not sure what you're point is?

But I feel like you would be better telling people the whole truth if you're trying to convince them to stop using it. I think people will be more likely to want to heart it from someone who isn't trying to condescend or patronize them.

I fail to see how I'm being condescending or patronizing.