• Welcome to OGBoards 10.0, keep in mind that we will be making LOTS of changes to smooth out the experience here and make it as close as possible functionally to the old software, but feel free to drop suggestions or requests in the Tech Support subforum!

KenPom 2015-2016: Back on Top: #1 in Luck (1/11)

Sort of. Wakes still top 100 without our best player. we weren't top 100 for most of the unwiped ass era
 
RedSoxFaithful said:
like, time on the shot clock? yes. time in the game? no



no



yes of course
Click to expand...

Yeah that's just wrong. A player's current shooting percentage is a good predictor of what his shooting percentage will be over a large sample in the immediate future.

It's a much less reliable predictor for a player's chances of making any particular shot. There are times during the game where the expected value of a Steph Curry 3 is much greater than (.472*3) and other times where it is much less.

Shooting a basketball is not like flipping a coin. When you flip a coin any extended deviations from the expected percentage (50%) are easily attributed to pure chance. When shooting a basketball any extended deviations from a player's expected percentage are likely attributable to a number of factors.
 
Deacsfan27 said:
I mean people think Kobe is clutch and he is one of the worst shooters of all-time in the final seconds. For every guy that exceeds his baseline at the end of a game there is a guy who falls short of his baseline.
Click to expand...

Just because people suck at identifying "clutch" players doesn't mean they don't exist. Your second sentence implies this. If "clutchness" were really just "luck (or pure chance)" you would expect that over a player's career his end of game performance would mirror his overall performance.

To borrow Spring Chicken's coin example, if you flipped a coin 1,000 times and ended up with 55% tails you wouldn't call yourself "lucky" you would think the coin was rigged.
 
SpringChicken said:
He is shooting an averagely-defended shot from the exact top of the arc each time, with his team down 88-90 and 1 second left each time. His percentage, X, may not be 54% but it is a number even if you don't know what it is. Or if it helps you, think of a free throw so the location and "guardedness" are controlled. But I doubt it'll help you based on what you just said.
Click to expand...

Well if he makes 15/15 his percentage is 100%. The fact that this greatly exceeds his overall shooting percentage (54%) or his overall shooting percentage on averagely defended shots from the top of the arc (whatever that is) is most likely attributed, at least in part, to something other than luck.

If the sample is large enough, and the deviation from the expected value is large enough, there is reason to believe that factors other than pure chance are involved. Kenpom's "luck" measurement is an attempt to capture all of these factors (which are not measured in offensive/defensive efficiency), including luck, in one place.
 
SpringChicken said:
Let's focus away from the team as a whole and look at the example of a good three point shooter. The best single season NBA 3 point percentage I found is Kyle Korver in 09-10 at 54%. If he had a chance to hit the buzzer-beating game-winner in 15 games, you would expect him to make it 8 times and the team to go 8-7. If he happens to make his 54% shot every time and the team goes 15-0 over that stretch, you would say they are "lucky" under the KP model. If he happens to miss his 54% shot every time and the team goes 0-15 over that stretch, you would say they are "unlucky" under the KP model. It's not a matter of whether each individual shot is lucky or not, because it's close to a 50-50 shot each time. It's just that over time you expect the results to roughly match the probability. Flip a coin enough times and you will get streaks of mostly heads or mostly tails; if you were gambling on heads or tails in that stretch you would be "lucky" or "unlucky," as it went. Or, clutch.
Click to expand...

That shot wasn't the only thing that impacted the result of that game. So your analogy to a coin is a false one.
 
RChildress107 said:
Yeah that's just wrong. A player's current shooting percentage is a good predictor of what his shooting percentage will be over a large sample in the immediate future.

It's a much less reliable predictor for a player's chances of making any particular shot. There are times during the game where the expected value of a Steph Curry 3 is much greater than (.472*3) and other times where it is much less.

Shooting a basketball is not like flipping a coin. When you flip a coin any extended deviations from the expected percentage (50%) are easily attributed to pure chance. When shooting a basketball any extended deviations from a player's expected percentage are likely attributable to a number of factors.
Click to expand...

like the hot hand?
 
deacvision7 said:
That shot wasn't the only thing that impacted the result of that game. So your analogy to a coin is a false one.
Click to expand...

No, because I've granted that it is a buzzer-beating game-winner each time. In other words, the outcome of that shot determines the outcome of the game. But I should know better than to get into this. I beat a hasty retreat to my lair.
 
RChildress107 said:
Just because people suck at identifying "clutch" players doesn't mean they don't exist. Your second sentence implies this. If "clutchness" were really just "luck (or pure chance)" you would expect that over a player's career his end of game performance would mirror his overall performance.

To borrow Spring Chicken's coin example, if you flipped a coin 1,000 times and ended up with 55% tails you wouldn't call yourself "lucky" you would think the coin was rigged.
Click to expand...

This is just sample size. I don't have any idea what a valid statistical number of shots it would take to be significant, but I'm nearly certain that nobody has attempted/made enough game winning shots that would exceed this threshold (whatever it is).
 
ReplacementDeac said:
this is wrong
Click to expand...

It's not a perfect analogy to "error" in a political poll but I think it's apt for the level of discussion we're having on this thread. The metric has been dumbed down and explained in ten different ways that provide better substance to how the number is calculated and some people here still seem to have no idea what it's representing or are still just blatantly misstating what it represents
 
Deacsfan27 said:
This is just sample size. I don't have any idea what a valid statistical number of shots it would take to be significant, but I'm nearly certain that nobody has attempted/made enough game winning shots that would exceed this threshold (whatever it is).
Click to expand...

Probably not, but that just means we don't have enough data to definitively say a player is clutch or not. It doesn't suggest the inverse, that a player is just lucky.

I give you two coins and tell you one of them is weighted towards tails while the other is normal and then ask you to pick one and flip it ten times. If it comes up tails 6 times are you going to have much of an idea whether you are flipping the normal coin or the weighted coin? Not really.
 
RChildress107 said:
Probably not, but that just means we don't have enough data to definitively say a player is clutch or not. It doesn't suggest the inverse, that a player is just lucky.

I give you two coins and tell you one of them is weighted towards tails while the other is normal and then ask you to pick one and flip it ten times. If it comes up tails 6 times are you going to have much of an idea whether you are flipping the normal coin or the weighted coin? Not really.
Click to expand...

Once again I don't think that anybody "increases their level of play" to the point of being clutch (i.e. a number so far above their normal baseline that no other factors can be attributed to it other than he is just better when it comes down to the wire), it's just that we tend to remember the big plays more than we remember the misses. Kobe has missed something like 13 straight game tying or winning shots in the 4th quarter/OT and everybody still thinks he is a killer.

So to your first point, I don't think we can ever have enough data to show that an individual player is "clutch", but instead some are willing to use a small sample size to fit a narrative that somebody really elevates their play when the going gets tough.
 
Deacsfan27 said:
Once again I don't think that anybody "increases their level of play" to the point of being clutch (i.e. a number so far above their normal baseline that no other factors can be attributed to it other than he is just better when it comes down to the wire), it's just that we tend to remember the big plays more than we remember the misses. Kobe has missed something like 13 straight game tying or winning shots in the 4th quarter/OT and everybody still thinks he is a killer.

So to your first point, I don't think we can ever have enough data to show that an individual player is "clutch", but instead some are willing to use a small sample size to fit a narrative that somebody really elevates their play when the going gets tough.
Click to expand...

As to the first bold I don't think anyone is arguing this. It's doubtful that a player shooting a higher percentage than expected at the end of a game is solely attributable to any one factor.

As to the second some are willing to use it to fit a narrative that nobody really elevates their play when the going gets tough.
 
Wakeforest22890 said:
It's not a perfect analogy to "error" in a political poll but I think it's apt for the level of discussion we're having on this thread. The metric has been dumbed down and explained in ten different ways that provide better substance to how the number is calculated and some people here still seem to have no idea what it's representing or are still just blatantly misstating what it represents
Click to expand...

describing it as similar to margin of errors suggests that you still seem to have no idea what it's representing or are still just blatantly misstating what it represents
 
Deacsfan27 said:
Once again I don't think that anybody "increases their level of play" to the point of being clutch (i.e. a number so far above their normal baseline that no other factors can be attributed to it other than he is just better when it comes down to the wire), it's just that we tend to remember the big plays more than we remember the misses. Kobe has missed something like 13 straight game tying or winning shots in the 4th quarter/OT and everybody still thinks he is a killer.

So to your first point, I don't think we can ever have enough data to show that an individual player is "clutch", but instead some are willing to use a small sample size to fit a narrative that somebody really elevates their play when the going gets tough.
Click to expand...

Anyone who thinks Kobe is clutch is an idiot. In fact anyone who thinks Kobe has been good at basketball the last three years or great at basketball the last 6 years is also an idiot.
 
Sure I should have stopped trying to explain it after people with wake degrees didn't follow the actual provided methodology. So for that I apologize - figured if people weren't gonna read the methodology after coming onto a thread solely devoted to a discussion of metrics the next best thing was to analogize to something people might have a better grasp on. I'll just stop posting and let you guys jack each other off about what luck could mean instead of what it actually is.
 
RChildress107 said:
As to the first bold I don't think anyone is arguing this. It's doubtful that a player shooting a higher percentage than expected at the end of a game is solely attributable to any one factor.

As to the second some are willing to use it to fit a narrative that nobody really elevates their play when the going gets tough.
Click to expand...

I would be willing to concede that "clutch" exists in the sense that people do make big time plays at big time moments.

I will not concede that it is predictable, and if you think that somebody is "clutch" and can be relied upon to consistently exceed their baseline, then they are likely about to regress to their mean.
 
Wakeforest22890 said:
Sure I should have stopped trying to explain it after people with wake degrees didn't follow the actual provided methodology. So for that I apologize - figured if people weren't gonna read the methodology after coming onto a thread solely devoted to a discussion of metrics the next best thing was to analogize to something people might have a better grasp on. I'll just stop posting and let you guys jack each other off about what luck could mean instead of what it actually is.
Click to expand...

Ken takes no position on what luck "means" or really what any of his ratings "mean." Luck is simply "measure of the deviation between a team’s actual winning percentage and what one would expect from its game-by-game efficiencies."

It is not an explanation of what causes that deviation which is what people are arguing about on this thread
 
It's a real wonder that we are having trouble quantifying intangible metrics.
 
Deacsfan27 said:
I would be willing to concede that "clutch" exists in the sense that people do make big time plays at big time moments.

I will not concede that it is predictable, and if you think that somebody is "clutch" and can be relied upon to consistently exceed their baseline, then they are likely about to regress to the mean.
Click to expand...

That's not much of a concession. If "clutch" players will eventually regress to the mean then "clutch" doesn't really exist. I argue that "clutch" does indeed exist (i.e. there are players who do not regress to the mean and would not even if their sample size were increased) but we do not have a large enough sample size to verify it or predict it.

It may or may not be predictable, but there would be no way of verifying that prediction and in fact the small amount of evidence we do have would suggest we suck at identifying it (i.e. Kobe).
 
You must log in or register to reply here.
Back
Top