Discuss TrueHoop's Feature on The Mavs Stats Guy Here
by Gils_Keloids on Oct 2, 2009 12:01 PM PDT
Email
PrintI know a lot of you despise Henry Abbott for not being able to see that Kobe is and has been the best player in the league, but there is a reason Truehoop has a large following: Like it or not, he has good content. Slightly off-kilter opinions and interpretations, but still, good content.
Read here about his latest Q&A with Wayne Winston, the Mavs professorial stat expert, whose expertise is breaking down and analyzing the best and worst lineups for teams. Although I don't agree with some of his conclusions, this is a way of analyzing stats that I can agree with, as opposed to trying to figure out who the best player is by some ranking. Instead, Mr. Winston digs up what the most effective lineups are.
Read first, and then let's discuss. We can talk about anything related to it, but I'd like to focus on this:
The key to the Lakers -- it doesn't take a genius to figure it out, is you close with Lamar Odom, Pau Gasol, and Kobe Bryant. If you do that, you can't really go wrong.
You take out Bryant and leave Odom in, they were good. You take out Odom and leave Bryant in they weren't good. That's why I say Odom was better than Bryant.
OK, this is what they closed with: Trevor Ariza, Kobe Bryant, Derek Fisher, Pau Gasol and Lamar Odom. That's 18 points better than average. And isn't that the one that closed every game? That's a good lineup, and they played that one the most.
Artest will be interesting.
Another good lineup is if you take that same thing, but put in Luke Walton for Fisher. Or Walton for Ariza. That's just as good. Doesn't make any difference.
This lineup is off the charts, and didn't play much: Bryant, Gasol, Odom, Luke Walton and Shannon Brown.
That lineup won by 55 points a game. They played 50 minutes together.
73 comments
|
0 recs |
Do you like this story?
Comments
In the " I say Odom was better than Bryant." link he ends with...
So it looks like the fantastic Kobe Bryant could not make the Lakers outscore playoff opponents with Gasol in unless he had Odom in. On the other hand Odom and Gasol easily dominated their opponents without Kobe. Makes you question the "conventional wisdom"!
Conventional wisdom told me that Lamar Odom played against more 2nd unit’s, and Pau was regularly subbed back in with the 2nd unit. I know this because I actually watch the games.
This is one of those stats lie kind of numbers. I don’t think this because it was Kobe related, but when of these so called “Stat Gurus” can win a Chip based of this “data” then call me. Or field a competitive team cheaply. Which was the whole point of “Moneyball” , which these “gurus” are trying to emulate. Mark Cuban sure spends a lot of money with no Rings to show for someone who keeps a stat guru.
I swear it’s going to ruin basketball.
I do think that some stats can work.
But I don’t think it’s as simple as crunching numbers. Basketball is one of those sports that it just too hard to look at that way. The variables are too much when it matters.
Agreed.
There are waaaay too many assumptions that need to be made in order for stats to be capable of making definitive statements.
For the Lamar vs. Kobe arguement posed above, the assumption is that over time, the level of competition that the player is going against will even out.
Another statement I had a huge problem with from this piece was the going for 3 when down two at the end of a game argument.
But the math is solid there. If I’ve got the ball and I can take the shot with the horn going off, you should go for 3. You only win the overtime half the time. Suppose you have a 50% chance of hitting the two. So you make a shot half the time, and then you win in overtime half the time, you win the game just 25% of the time.
But if you shoot the 3, you’ve got at least a 30% chance. That’s all you need to know.
Because it’s a sure thing that if the two teams play against each other for 5 minutes, each has a 50:50 shot of winning that period? I’m sorry, but if you are a good team, and you had an off night causing you to be put in this position against an otherwise bad team, your chances of winning the overtime period (and the game) are going to be significantly better than 50:50. Likewise, if you are a bad team that has kept it close against a good team, you should definitely be going for the three because your odds of winning the overtime are significantly less than 50:50.
I understand what he was trying to say about going for the 3, down 2....
but I think his logic is also off in his math. A 3-pt. may be a 30% shot, but theoretically, you still have a 50% chance of making it. It either goes in or it doesn’t. It’s only a shot that goes in 30% of the time over a prolonged period. But it’s still 50% on any given play.
Now, what I really want to know is why haven’t the basketball gods that Henry Abbott knows tell him that going for the 3 isn’t the right play. Surely, the basketball gods want the ball passed to big man down low for a 2-pt bucket? As Henry Abbott would say, I find that curious.
“you still have a 50% chance of making it. It either goes in or it doesn’t. It’s only a shot that goes in 30% of the time over a prolonged period. But it’s still 50% on any given play.”
I don’t mean to play math police here, but you should try that math again. Think of this example. If what you are saying is true then if you can make a 50 putt 2% of the time you still have a 50/50 shot of making the next 50 foot putt. It doesn’t work that way. If you are a 30% three point shooter then you have a 30% chance of making the shot every time (statistically speaking of course).
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
I know what you mean.
But in one play? On a shot that can easily be made. A 50 ft. putt is very unlikely (I’m assuming). It’s already highly unlikely that a golfer will make it. But a 3 pt. shot is within any NBA player’s range of ability.
Tim Duncan probably had a 20% chance of hitting that shot in the playoffs to beat the Suns in Game 1 of the 1st Round in 2008 (let me catch my breath after that), but it still had a 50% odds of going in once it went up. It did. He may have very well missed the next 4, but it doesn’t matter, because he hit the one he needed.
I know what the math says, but I’m talking theoretically. For one play. By an NBA player. On a make_able_ shot. Not a half court heave (comparable to a 50 ft. putt). Do you understand what I’m trying to say?
Yeah I understand what you are trying to say, but you are still wrong.
Mathematically speaking or not. Having two choices doesn’t mean 50/50. I simply used an extreme example, but the math is still the same. The odds of one of two thing happening isn’t 50/50 unless it is a coin flip, etc. Duncan only had a 20% chance of making that shot. He made it. So mathematically speaking he has a 0% chance of making the next 4. Of course that doesn’t really mean he will miss all 4, but he will never have a 50/50 chance on any given play. Even a 49% shooter will only have a 49% chance of making any one shot, even if he has missed 5 in a row.
If I walk out of my house and across the street I have two options, get hit by a car or not. That doesn’t mean there is a 50/50 chance I will get hit. One more example. Does Lebron have a 50/50 chance of making every layup or dunk? Of course not.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
Of course we know LBJ is more than 50/50
because we have that knowledge over prolonged time. Just like we know Timmy is about 20% form 3pt land. That’s why you and I know he’s about 20% to make the shot. But what if we didn’t. It’s still 50/50 to go in.
Look at it this way.
Let’s say you walk into a gym. It’s the same situation we’re discussing. A team is down 2, with 5 seconds on the clock. Now, you haven’t watched any of the game. You don;t know who can shoot, who can’t, or what the fg% of 3 pt % is for the team (or players) with possession. The play unfolds and a player goes for the 3. Not knowing the % of the players shooting or the team, and you have to figure it out. What are the odds that that shot goes in?
There is insufficient information to answer your question.
Without knowing how good of a shooter he is, you can’t answer the question. However the idea that it is either going in or not doesn’t make it 50/50 anymore than the chances of you dying in a skydiving accident on your first try are 50/50 simply because you will either die or not. Two choices does not equal 50/50 regardless of the situation.
I will try one more example and then you can simply ask a math professor and apologize later.
Using your examples it would make just as much sense to foul Steve Nash at the end of a game when you are up 2 as it would to foul Shaq. Even though one shoots 50% and the other one shoots 90% according to your theory that on “any one shot” the odds are 50/50 then you should feel equally sure that one free throw will be missed. Of course this isn’t true.
I don’t know what else to say, you are wrong. I am not trying to be a dick or continue/start a flame war, but the truth is the truth, your math isn’t correct. If in your mind the odds are 50/50 because you don’t have enough information on the shooter, that is fine, but in reality the odds are not 50/50 that the shot will go in. Unless of course the shooter is a 50% shooter.
Now, once the ball is in the air, the outcome has been determined already and at that point the odds are either 100% or 0% because whatever needed to be done to make or miss the shot has already happened. Either way it doesn’t mean the average is 50% since the 0% option will occur more often than the 100% option. Trust me I am a math guy, if I wasn’t I wouldn’t even bother trying to convince you, but math, physics, probability, etc. are kinda my thing.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
By definition you are wrong.
“It’s still 50/50 to go in.”
By definition this means it will go in half the time. If it will go in half the time then if you did it 1000 times it would go in 500 times. This would make Tim a 50% shooter on off balance, last second, 25 foot shots. He isn’t, therefor the 50/50 idea can’t be correct either for his career or for any given shot.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
Trust me.
I don’t take it personally. I understand your points, and the math of it.
My only point is the shot either goes in or it doesn’t, and it very well could, so the fact that it might for one play is either yes or no. which I tried to explain as 50/50, but I understand where you say I’m wrong. Theoretically, logically, whatever I still maintain my point. Maybe I should have just left the math out of it.
By the way. I just knew I was going to come back home and read a 0% or 100% answer. I thought about it as I was driving…..
Can you get a ticket for DMWD? (doing math while driving)........LOL.
I do think that I understand your point. I think you are saying that since we don’t know who the player is that is taking that last shot then we can’t guess as to the chances that it will go in. Is that kind of what you mean? I agree with you there, but the percentage is going to be less than 50% since no player in the NBA makes half of his 3 pt. shots. But I do see where you are going.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
See. There's a reason I do not write articles.
I do not have the patience to properly write out my thoughts and explain them fully (or correctly?).
‘See. There’s a reason I do not write articles."
I thought you do write articles, or am I wrong? Anyway I don’t write articles because I am too cynical, inflamatory, and sarcastic. At least that’s what the rest of the idiots tell me. Kidding, kidding.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
Just say
the shot he is about to take is independent (i.e., in no way related) of all the previous shots he took. Like he is taking the shot for the first time. Hence, the result should be 50/50
Sure, he could say that........
If he still wanted to be wrong. The real answer is that there is not enough information to determine the correct percentage. However the answer is not 50/50 simply because there are two choices (ball goes in, or doesn’t). The shot he is about to take IS related to all of his previous shots because those shots show how good of a shooter he is.
Like I said before if the answer is 50/50 simply because there are two choices then the answer would be 50/50 when given ANY two choices which is ridiculous. Like the example I gave earlier regarding making a 50 foot putt in golf. The situation is EXACTLY the same as the basketball situation except that the shooters/golfers percentage of success in previous attempts are not the same. However if the math works for one, it must work for the other (by definition). Since NO golfer in the world has a 50/50 chance of making a 50 foot putt on any given attempt (simply because there are only TWO outcomes, make it or miss it) then the math doesn’t work for this example or any other example of having TWO choices without any other information.
The only thing you can really say about the outcome of an event with only two options (and no other information) is that the percentages of success or failure will add up to 100%, but the ratio can be anything.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
OK, I've read enough and I have to put in my two cents
Probabilities are relative.
If you are God (supreme being), there is no such thing as a “probability” – you know what’s going to happen. A coin toss is not 50/50, it’s 100% heads or tails.
For us humans, probabilities are a statistically driven phenomenon. If a player has made 40% of his shots in the past, then you say as a shorthand that he has a 40% chance of making a shot, given normal circumstances. But what you’re really saying is that in the past, he’s made his shots 40% of the time, which is very different than saying he has a 40% chance to make a shot.
But what else can you go on but the historical record? So what wondabap is saying is that he doesn’t believe the historical record applies to the future, which is perfectly valid viewpoint. But pslakerfan is not in the minority when he believes that the past record is a good indicator of future success, so he says the %s are lower for someone who has shown to be a poor shooter in previous attempts.
These probabilities are much different than the example of pulling different colored marbles from a bag. If we know how many marbles are in the bag and how many colors of each, then we can say we know the chances of pulling a certain color marble. But in the basketball shooting analogy, we don’t know how many are in the bag and the colors. But if we’ve in the last 100 tries we’ve pulled 40 red marbles and 60 blue marbles, we can take a guess at what the distribution is inside the bag.
You may not realize it, but it’s a philosophical argument you are having as to what it means to have an X chance of something happening.
"This is not a game for boys. This is a game for men." - Phil Jackson
by Gils_Keloids on Oct 6, 2009 12:49 AM PDT up reply actions
ladies and Gentlemen
The incomparable Gils_Keloids.
Thank you put my argument in it’s proper perspective. You summed up what I was trying to say very nicely.
LOL
I am not sure how he just made your argument since he made mine, but I guess it doesn’t really matter.
My point was simply that there is not enough information to determine the odds in this case, but that the odds were definitely not 50/50 simply because you have two choices. I think that is exactly what Gils said. Maybe I am not interpreting it correctly.
Wondahbap, weren’t you saying that the answer is 50/50 simply because there were two choices or am I mis-remembering.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
He didn't say I made his argument
Only that I put it in its proper perspective.
There is no right or wrong here.
No one knows the true probability of a player making a shot.
What you are speaking of, pslakerfan, is the statistical probability, which is based on the thousand of shots the player has taken before. This has shown to be pretty reliable, but it’s still different that the true, or theoretical, probability.
Go back to my marble example. If I knew how many marbles were in the bag, and of what color, I could tell you the true probability by some simple calculations.
But what if I didn’t know how many there were and what colors there were? This is more like real life. Say I pulled a marble and replaced it, and did this a thousand times. If I got 800 red marbles and 200 blue ones, I could say that there’s an 80% chance of pulling a red marble and 20% of pulling a blue marble. But I might be wrong. I could actually pull out a green marble on the 1001th try! I can’t be sure of the true probability, but the best I can do is make an educated guess through many trials.
In fact, some mathematicians (dead now) have devoted their studies to finding the ideal number of trials to have a certain measure of confidence of estimating the true probability.
I believe that what wondahbap is seemingly getting at, is that the statistical probability could possible be a faulty indicator of the true probability.
"This is not a game for boys. This is a game for men." - Phil Jackson
by Gils_Keloids on Oct 6, 2009 11:05 PM PDT up reply actions
Okay, this is my last attempt at this, I swear.
“No one knows the true probability of a player making a shot.”
That is EXACTLY what I have been saying for the last two days!!!
“Go back to my marble example. If I knew how many marbles were in the bag, and of what color, I could tell you the true probability by some simple calculations.”
Is this supposed to be for me. Because I understand probabilities quite well, I thought the marble example was for Wondahbap.
“In fact, some mathematicians (dead now) have devoted their studies to finding the ideal number of trials to have a certain measure of confidence of estimating the true probability.”
I think a players career stats are a reasonable if not “ideal” number of trials. Besides his point was that if we don’t know his career stats we should assume the shot has a 50/50 chance of going in which of course is false. If we want to talk about sample size I would say the entire history of the NBA (in which no one has shot 50% on 3 pointers for a career or even a season) would qualify as a reasonable sample size.
“I believe that what wondahbap is seemingly getting at, is that the statistical probability could possible be a faulty indicator of the true probability.”
This statement is true, however it is not even close to the point that he was trying to make. His point, again, was that in the absence of any other information, you should assume the shot has a 50/50 chance of going in. Which, again, is false, statistically or any other way.
“What you are speaking of, pslakerfan, is the statistical probability, which is based on the thousand of shots the player has taken before. "
Actually I have been speaking of statistical and real probability. For example.
“The real answer is that there is not enough information to determine the correct percentage.”
and…
“If I walk out of my house and across the street I have two options, get hit by a car or not. That doesn’t mean there is a 50/50 chance I will get hit. One more example. Does Lebron have a 50/50 chance of making every layup or dunk? Of course not.”
AGAIN, my point here has always been:
1. You can’t determine the true probability of any one shot going in because there is not enough information and too many variables (basically the opposite of the marble example).
2. You can statistically say what the chances of the shot going in are based on a large sample of previous success. Regardless of the size of the sample however (for instance not knowing the shooter) it is not correct to assume the chances are 50/50 for anybody.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
You are right.
"This is not a game for boys. This is a game for men." - Phil Jackson
by Gils_Keloids on Oct 7, 2009 12:26 AM PDT up reply actions
Here' my final say.
What I meant to say the whole time is:
For one play, the shot will either go in or it won’t. There is no 3rd option. Given the skill level of an NBA player, it very well could. So, forgetting previous history or future probabilities, any one shot should have as much chance to go in as missing.
I should not ave included math in my argument. I shouldn’t have used percentages. They really confused my point.
So, while Steve Nash might be a better shooter than Timmy because we know his stats, or LBJ will most likely convert a dunk or lay-up. For one play, Timmy could hit the 3, Steve Nash could miss it and LBJ blow the dunk. We can assume what will happen because of history. But taking that out of it, only one of 2 things will happen; make or miss. Which I described as 50/50.
Does that help? I started this huge debate unintentionally with improper wording.
ehh...
Maybe, but I still think your basic problem is not one of wording.
You still claim that any one shot should have as much chance to go in as missing. Whether or not you attach a mathematical formula to that, it’s still basically saying that it’s 50/50, and that it’s 50/50 for the reason that there are only two options. Neither of those is true.
thank you, that is what I am trying to say.....
of course it took me about 69087 words to do it….LOL
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
The true probability is unknown.
"This is not a game for boys. This is a game for men." - Phil Jackson
i took stats in college
thanks for reminding me why i hated it, lol
by true_lakerfan on Oct 8, 2009 4:36 PM PDT up reply actions
ack!
pslakerfan wrote: Duncan only had a 20% chance of making that shot. He made it. So mathematically speaking he has a 0% chance of making the next 4.
The mathematics doesn’t actually say that. It says exactly what you say: That the result of future shots doesn’t depend on the result of past shots. Obviously you know that this is the case in reality; I’m just pointing out that the mathematics says this too.
wondahbap wrote: I know what the math says, but I’m talking theoretically.
“Theoretically” is exactly where what you say makes no sense. Theory is exactly where it says if the shot is the same, the probability he makes it is equal to the long-term probability. It can’t be different because the long term is nothing more than the collective sum of all of these individual trials.
WW is right about the three-point shot, I think. Overtime is, historically, pretty close to a 50-50 affair. You’d think that good teams would fare much better in overtime than bad teams, but apparently it doesn’t quite work out that way in the long run.
However, I think his sample size for the so-called “best line-up” is tiny tiny tiny. It’s barely a game’s worth. Given how inconsistently a team can play from game to game, that statistic is barely worth the pixels it’s displayed in.
This entire discussion is insane
Down by 2 with time ticking down
Layup > 3 > Jumper
Why? Because the highest percentage shot is the best shot to take.
Even a contested layup is an 80%ish make. You win 50% of the time in OT and you’re winning 40% of the games.
A last second 3 is harder to hit than a non last second 3, but you will win 30ish% of the time.
A jumper is also harder to hit, you’ll tie 40% of the time, meaning you win 20%.
40% > 30% > 20%.
The turn around jumper is the worst possible shot to take with the clock winding down, just like it’s the worst shot to take any other time during the game.
Poster for next year? I'm thinking My Little Pony.
I hope this isn't too controversial
But the shot that has the best chance of going in when taking in all the circumstances (game time, opponent, weather, shot clock, player biorhythms) – that is the best shot to take.
"This is not a game for boys. This is a game for men." - Phil Jackson
True.....
and the “circumstances” at the end of the game don’t allow for very many “layups”.
And I don’t mind being too controversial. I can even do it without insults.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
I pretty much have to insult someone
To feel like I won an argument
"This is not a game for boys. This is a game for men." - Phil Jackson
No you don't.......
And I wasn’t talking about you………but you already know that.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
Hey watch the swearing.........
we don’t take too kindly to that kind of language here in Boston……. Oh wait, I’m not in Boston. Never mind.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
Here's the flaw in this argument.
“Even a contested layup is an 80%ish make.”
No way is a contested layup in a last second scenario going to be 80%. There is a reason that most teams shoot a 3 or a mid-range jumper in the last second. It is usually the only thing available. No team is going to allow a layup for the last shot (well at least not very often). Most likely you won’t get within 10 feet of the basket. At best you will get fouled, but that isn’t too likely.
This “40%>30%>20%” is only true if your 80% “guess” is true. Give me some evidence that “80%” is even close to correct and I will agree with you. Otherwise I say it is 50% at best, which kinda ruins your formula……..“30%>25%>20%”.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
Umm
If a team gets a layup look at the last second, it’s 80%ish. I never claimed that teams CAN get a layup look at the last second every time, just like they can’t get a look at a nice 3 every time. Games occasionally do end with layups though, even when opponents guard them.
I’ve only been to 14 Blazer games live in the last 2 years and two ended with a game winning layup. One by Roy and one by Lebron sadly. Now, these were both 1 point games so your equation has to change, since the 3 ball loses it’s value quite a bit. (Unless there is time left on the clock.)
Poster for next year? I'm thinking My Little Pony.
Well since we can't argue hypothetical math.....
I will just say you have been lucky that 14% of the games that you have been to have had such an exciting finish.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
Call me crazy
I still think the best shot to take is the one that has the best chance of going in
"This is not a game for boys. This is a game for men." - Phil Jackson
by Gils_Keloids on Oct 9, 2009 10:30 PM PDT up reply actions
Not really...
If you’re down by 2 going for a slightly lower percentage win is way better than going for a slightly higher percentage tie.
Poster for next year? I'm thinking My Little Pony.
I actually agree with you on this one.
Although I don’t think your personal guess as to the shot percentages are correct.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
Well the point is, I am smarter than everyone in the world.
"This is not a game for boys. This is a game for men." - Phil Jackson
by Gils_Keloids on Oct 10, 2009 3:31 PM PDT up reply actions
Arguing hypothetical math is how you win a nobel prize in science
Poster for next year? I'm thinking My Little Pony.
And being president for 2 weeks is how you get nominated for a Nobel Peace Prize.
What’s your point?
BTW, I think you mean theoretical physics, not hypothetical math.
Well, sir, you are a cowardly son of a bitch! You just shot an unarmed man!.......Well, he should have armed himself if he's going to decorate his saloon with my friend. – Will Munny
I swear
I know a lot of people hate Kobe but none to the point where they’ll make crap up like HA does
"I don't want to be the next Michael Jordan. I only want to be Kobe Bryant,"
-Kobe Bryant
I actually thought the Lakers "best" lineup of Kobe, Gasol, Odom, Walton, and Brown was interesting
And it makes sense. A lot of big players with all kinds of complementary skills. These guys wouldn’t be getting in each other’s way.
"This is not a game for boys. This is a game for men." - Phil Jackson
If you are a “stat guru” with a generous owner such as Mark Cuban, you’d think that with all your conventional wisdom you’d have a ring or two. Also, as was mentioned in the article, “Is Lamar Odom Better than Kobe Bryant?” the sample size is way too small for Lamar Odom’s part. On top of the fact that i’m guessing it’s near impossible to determine the net effect of Bryant’s assertiveness as a leader on both ends of the floor. I would have liked to see the numbers rather than the abridged version, especially if he’s going to come to such a conclusion.
I would too
But my guess is it wouldn’t matter. APM is not a suitable proxy for “is better than.” It’s an argument that wasn’t meant to be subjected to statistical comparison (unless we can all agree on the right statistic).
“Conventional wisdom,” by the way, refers to the popular (in this case, non-statistical) conception that Kobe Bryant is the best player on the Lakers. Winston apparently disagrees, at least as far as the APM goes.
Statistics isn’t so deterministic as to reliably obtain a ring, even given a suitable amount of time, and I bet even Winston (who obviously thinks very highly of his APM) knows this. Many a slip twixt the cup and the lip, as it were. And even Winston can’t make the Mavericks players into superheroes; all he can do is advise Cuban on how best to deploy them.
Umm...
Because knowing who the best/most efficient/whatever players are doesn’t mean you can get them.
Poster for next year? I'm thinking My Little Pony.
But if you go against such “conventional wisdom” and pay a guy like Lamar Odom (or the best, most efficient, whatever) the 21 million Bryant makes, you will more than likely have the upper hand in negotiations.
Oh they did. They inherited Jason Kidd’s 21 million dollar contract last season. The stat guys love Jason Kidd. My fantasy team loves Jason Kidd. They even justified the 40 something million left on his contract and their trading away future all star Devin Harris by saying Kidd is second to one LeBron James in success relative to their position.
Fantasy Numbers vs Statistics... dumb debate
FANTASY SQUAD:
Baron Davis
Jamal Crawford
Corey Maggette
Zach Randolph
Marcus Camby
vs
STATISTICS SQUAD:
Steve Blake
Shane Battier
Luke Walton
Lamar Odom
Joel Pryzbilla
If these teams go against each other… who should win? That is a stupid question. There is one ball hence fantasy squad cannot work together since all their talents rely primarily on being able to have the ball and score. Marcus Camby is a great defensive player… but… his team defense has always been atrocious… ie… still very fantasy-esque. Statistics squad shown above are all a bunch of +/-, wins earned, wages win… etc.. One cannot sum it all in one jar and claim a know it all.
A winning team has a great manager, a coach who can put it together, great assistant coaches, a great trainer, defensive players, players who perhaps are a little selfish who only want to score because they are damn good at it… players who do the dirty work… the +/-, etc… and even perhaps a player who embodies it all like Kobe or Duncan.
Stats squad!
An entire team of guys shooting 35% won’t win, ever! (Except Canby who will never touch the ball.)
The inefficient chucker is the most overrated player in the league. Hence why the top squad gets such ridiculous contracts, while helping their team lose games.
Poster for next year? I'm thinking My Little Pony.
Great quote that applies to Basketball
“Not everything that can be counted counts, and not everything that counts can be counted.”
Albert Einstein
This is not one of those apocryphal quotes by someone famous, Einstein actually did say this, and even though he’s sort of a clever guy, the truth of the statement is more important than the reputation of the speaker.
"This is not a game for boys. This is a game for men." - Phil Jackson
although I agree with Einstein...
…(heh heh) let’s not throw the baby out with the bathwater. In other words, let’s not disregard statistics just because they’re statistics and seem remote from the actual game of basketball. No, basketball is not played on paper. But the result of the game—the score—is a basic statistic.
Instead, be absolutely clear about what the statistic is measuring. In my opinion (putting on my probability and statistics hat now), WW glosses over a lot when he concludes that Lamar is “better than Kobe” (or whatever his words were). Maybe he doesn’t actually believe that. Maybe to him it’s just shorthand for “has a better APM than Kobe.” But when writing to the interested lay public, he really ought to be more careful.
In fact APM says no such thing. What it says is that after you’ve adjusted (to some extent) for his teammates, after you’ve adjusted (to some extent) for his opponents, the Lakers outscore their opponents by more with Lamar on the court and Kobe off it than vice versa. He chooses to summarize that as “is better than Kobe.” Maybe it’s just because I’m sort of paranoid, but I wouldn’t go that far. And it’s not because I think Kobe is superior to Lamar (although I think he is), it’s because I don’t think the way the numbers are derived support that conclusion. Those numbers do adjust for a lot (including many factors that people here are apparently not aware of), but it still isn’t enough.
Ultimately, there’s no point in using the numbers that way. Individual players don’t win games or titles anyway. APM is derived from line-up numbers, so use the line-up numbers, not individual numbers. The only reason for deriving the individual player APM is to make it possible to estimate what the numbers would be for line-ups that have never played together. It has some value this way, as opposed to settling bar arguments about which player is better than which other player.
Nonetheless, this use still assumes (as most people do) that APM adds linearly, that if you have five players whose APMs are +1, +4, -2, +3, +6 (say), the line-up should outscore their opponents by 12 on average per 48. It’s never been demonstrated satisfactorily to me that APMs (or any kind of +/-) add linearly.
None of these weaknesses, by the way, do I feel are inherent in statistics per se. Just in these particular statistics, and especially in the way that they are haphazardly interpreted. WW is acting as an engineer, not a scientist. That’s fine, that’s probably what Cuban wants him to do. But it also means that his conclusions are not scientifically vetted. Closed formulas, no peer review—it’s an approach that can work, with a skilled engineer, but you could also end up with well-dressed garbage. Probably the truth is somewhere in between.
Yep yep
If you say “better than”, you have to qualify it. Because “better than” does not stand alone.
If someone tells me Bynum is better than Kobe on PER, I can’t argue that. If someone says Lamar’s APM is better than Kobe’s, I can’t say fairer than that. But if you tell me then that means Bynum or Odom is a better player than Kobe, well then, them’s fighting words.
"This is not a game for boys. This is a game for men." - Phil Jackson
Well
Depends on your opponent :-)
This is the issue with Odom/Kobe. Hopefully I explain this well.
What you need to do is take the APM of lineups AGAINST the APM of who you play against. But since your results vs them skew the overall results, you need to take out the APM of the teams playing. So Odom has a great APM, but maybe it’s because his opponents on the floor have a weaker combined APM. (In theory this should average out, but since Odom comes off the bench maybe not.)
BUT, if Odom is that good he’d lower the APM of ALL his opponents, so you’d have to take their APMs minus time against Odom and them factor that against Odom’s APM. Then do the same with Kobe. THEN compare their new APMs. This becomes a serious pain to do… and obviously it was not done here.
You’d have to find the squads that Odom was against every minute, take the minutes vs Odom out of their APM, and then calculate their APM. THEN you could compare this APM to Odom, to find his true APM.
Then do the same with Kobe… again time consuming. As people have mentioned based on their observations of the game, Kobe plays vs tougher competition for a chunk of the game than Odom. But not for as much of the game as people really think. (IMO) Would Odom be as effective once you factored this in? Probably not… although I think Gasol still would be. (But we all know how I feel about Gasol being underrated and Kobe overrated, so that’s a different issue.)
So Mr. Winston… because I know you are reading this. Recalculate your APM the way I am telling you too. I would, but I don’t have access to your data and don’t get paid crazy amounts of cash. With your recalculated APM, I can assure you that Kobe will be ahead of Odom, but I am curious to see Gasol vs Kobe. Thanks.
Poster for next year? I'm thinking My Little Pony.
A super duper simple example with nice numbers
Odom + 4 guys plays 20 mins vs 5 guys and goes +15
Odom + 4 other guys plays 20 vs 5 other and goes +12
Odom + 4 other guys plays 20 vs 5 other and goes +10
Odom + 4 other guys plays 20 vs 5 other and goes +9
Odom + 4 other guys plays 20 vs 5 other and goes +8
Odom has teh Godly APM of +10.8
Kobe + 4 guys plays 20 mins vs 5 guys and goes +4
Kobe + 4 other guys plays 20 vs 5 other and goes +5
Kobe + 4 other guys plays 20 vs 5 other and goes +5
Kobe + 4 other guys plays 20 vs 5 other and goes +5
Kobe + 4 other guys plays 20 vs 5 other and goes +6
Kobe has a good APM of +5.8
Odom > Kobe?!? This is what Mr Winston is saying, but he isn’t doing this. (I’m pretty positive at least.)
5 guys Odom faced Group A: -10APM
5 guys Odom faced Group B: -8APM
5 guys Odom faced Group C: 0APM
5 guys Odom faced Group D: -10APM
5 guys Odom faced Group E: +4APM
But since Odom is Godly, we have to remove his bias. Fortunately for me all these groups plays 96 minutes together. After taking out their time being beat on by ODOM (who gets all caps now), we have adjusted APMS of -3, -2, +6, -6, and +12. (Keep in mind 2 things. One these minutes are way too low, in reality taking out only Odom’s minutes would adjust these numbers by decimals… two I’m tired and rounding the hell out of these in my mind, but the point still works.)
So ODOM’s opponents have an adjusted APM of +7/5 or +1.6.
ODOM’s opponents should outscore his team, but 1.6, but instead lose by 10.8 so ODOM has an APM of 12.4 RAWR.
But, Kobe plays the pro opponents. His guys have APMs of
+12
+10
+8
+6
+4
After adjusting this for the time played vs Kobe (same criteria as above and same disclaimers) the numbers becomes. +18, +16, +13, +11, +9.
Or an average of +13.4.
This means that although Kobe is only outscoring opponents by 5, he SHOULD be losing by 13.4. This gives Kobe an actual APM of +18.4.
Review:
APM says Odom 10.8 and Kobe 5.8
When you test for error
APM says Odom +12.4 and Kobe + 18.4
Poster for next year? I'm thinking My Little Pony.
Not done yet!
I want to stress that these numbers are extreme and fictional. Players plays in dozens of different groups and against hundreds of different groups. So the numbers won’t be as extreme as mine, but when it comes to bench vs starters, the differences could be severe. I don’t know thanks to a lack of being paid, computer programming skills, and data.
One more (much more simple) explanation for why this could be done.
Miller/Rudy/Roy/LMA/Oden is -5 let’s say.
What if they played 80% of their minutes against…
Jordon/Bird/Magic/Worthy/Kareem
Kobe/Lebron/Wade/Paul/Howard
Patrick Chewing/Sasquatch/Mighty Mouse/Me
Obviously these 3 lineups are all going to have a +25-35 APM. So The above lineup would actually be a GREAT lineup if they were only getting beat by narrowly by all these groups. Again, an extreme example, but with starters vs bench players it is one that could occur. (On a smaller scale, such as a +7APM being worse than a +4.)
Poster for next year? I'm thinking My Little Pony.
Last one!
This was my super ultra simplified version. Doing this for ONE actual player in the NBA would take thousands of separate calculations. Each player after would have to redo the calculations with his own opponents/scaling. We’re talking hundreds of thousands of calculations to properly do this.
If someone has mad computer skills and can code it, go to town! But for now, I’m guessing Winston won’t bother to correct his oversight.
Poster for next year? I'm thinking My Little Pony.
the real problem
Winston doesn’t post his formulas (as far as I know). So it’s all speculation with regard to what he does or does not normalize for. Same problem with BCS formulas (except for one of them, I believe, which is open).
Does anyone know if/where Winston has posted his APM algorithms?
Easier to deny accountability that way!
Cleveland Cavaliers 0.8769
Denver Nuggets 0.7773
Boston Celtics 0.7436
Los Angeles Lakers 0.7263
Orlando Magic 0.6328
Philadelphia Sixers 0.6187
Dallas Mavericks 0.6028
Utah Jazz 0.5841
Portland Trail Blazers 0.5570
Houston Rockets 0.5385
San Antonio Spurs 0.5312
New Orleans Hornets 0.5131
Such as these “projections” using “advanced stats.” Really? Has this guy actually tested his equation on past years to see if he was remotely close?
Poster for next year? I'm thinking My Little Pony.
Basically
Cuban is paying this guy a ton of money for flawed analysis.
I forget who said it, but someone smarter than all of us said that the bane of mathematicians was logic. Even if your equation works, if the logic isn’t sound, it’s no good.
(This becomes really obvious in older math. The Greeks believed that flipping a coin twice gave a 100% chance of heads because 1/2 = 50. So 50% + 50% = 100%. The math works, but the logic is awful.)
Poster for next year? I'm thinking My Little Pony.
I believe...
…if you read the comments in his Odom > Kobe post, he does say he corrects for these things.
But again, I say that his corrections almost surely assume that APMs work linearly. (Since he uses them that way.) Your corrections also make that assumption. It’s a natural assumption to make, but it cannot be overstated how insidious but how important it is. It makes the corrections computationally tractable. If they aren’t linear, the whole job of correcting, or adjusting (the A in APM), becomes a heck of a lot more difficult.
Frankly, I think that such an assumption, however natural it might be, is simply unwarranted. It might work to the extent that those players already play alongside one another, but the more they don’t, the less reliable that assumption is. Unfortunately, it’s when they do play separately that we need that assumption. (If they always played on the court at the same time, there’d be no need to separate them ever; we’d just assign a single APM to the Kobe-Odom pairing.)
There needs to be a radical re-thinking of statistics before you can do anything like this. My first off-the-cuff guess would be a kind of vector statistic, with the individual elements representing attributes such as, for instance, finishing ability, court vision, top speed, acceleration, long-range shooting accuracy, etc. At least a couple of dozen elements in all. Since these aren’t ground statistics they would likely have to be computed through some kind of regression analysis. The effect of a pairing would depend in some non-linear way on these vectors. For instance, if you pair a guy with good court vision (and accurate passing) with someone who can finish well, that’s likely to be a net positive. But combining two guys who can just finish well but can’t find each other in semi-transition might not help as much as you would expect from a naive addition of APM. It might be possible to divine some generalities through tensor analysis, but that’s a long way down the road.
People with some imagination and advanced mathematics have yet to really take a good look at this. It’s not an easy problem by any means.
Comments For This Post Are Closed
Secondary Sidebar
User Tools
FanPosts
Community blog posts and discussion.
Recommended FanPosts
Recent FanPosts
Featured Stories
FanShots
Quick hits of video, photos, quotes, chats, links and lists that you find around the web.
Recommended FanShots
Recent FanShots
In Association With
Ball Don't Lie - NBA - Yahoo! Sport
- Days of NBA Lives: Wherein Roy Hibbert tap-dances to Maine
- Grizzlies Valentine’s Day cards are the key to love
- Kevin Love would like the NBA to take the MLB’s approach on All-Star games
- Kobe Bryant fires a few hilarious salvos at Jeremy Lin
- Brandon Jennings states the obvious regarding his 2014 free agency, everyone goes nuts
Certain photos copyright © 2012 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
