|
|
Post by Martinez on Jun 10, 2011 9:20:41 GMT -5
They are awesome You are just jealous Both of them are basically PER, with some strange weighting....which would be ok, except there's tons of bad assumptions in there, for example: -Possessions don't end on a FGA or FT, they end on a FGA or FT that isn't rebounded by the offense. Actually calculating possessions is quite complicated. That .93 number in no way represents the value of a possession in this league (it's actually about 1.11). -Points and assists are double counted basically. You can't credit 2 points to someone for making an assisted shot. 2 points is 2 points is 2 points, if you're going to give the assist man some credit, some credit has to be taken away from the guy taking the shot. -FTA/2 is a decent estimate of 2 free throws actually being a FGA, but it's closer to FTA*.44. If anyone wants to understand basic advanced stats, they should pick up a copy of Basketball on Paper by Dean Oliver. owned. |
|
|
|
Post by greeme on Jun 10, 2011 9:37:28 GMT -5
Both of them are basically PER, with some strange weighting....which would be ok, except there's tons of bad assumptions in there, for example: -Possessions don't end on a FGA or FT, they end on a FGA or FT that isn't rebounded by the offense. Actually calculating possessions is quite complicated. That .93 number in no way represents the value of a possession in this league (it's actually about 1.11). -Points and assists are double counted basically. You can't credit 2 points to someone for making an assisted shot. 2 points is 2 points is 2 points, if you're going to give the assist man some credit, some credit has to be taken away from the guy taking the shot. -FTA/2 is a decent estimate of 2 free throws actually being a FGA, but it's closer to FTA*.44. If anyone wants to understand basic advanced stats, they should pick up a copy of Basketball on Paper by Dean Oliver. owned. Nah, pretty much the same questions that were raised last time Why don't you make yourself useful and go cut and paste the answers Thanks |
|
|
|
Post by Martinez on Jun 10, 2011 9:41:44 GMT -5
Nah, pretty much the same questions that were raised last time Why don't you make yourself useful and go cut and paste the answers Thanks lol @ arguing with me yet you just got shut the fuck down with your shitty fucking formulas you queer. |
|
|
|
Post by Martinez on Jun 10, 2011 9:42:09 GMT -5
good job of completely ignoring repole's answer because you have absolutely zero clue as to how respond to it
|
|
|
|
Post by greeme on Jun 10, 2011 10:20:50 GMT -5
It is actually pretty simple
Just annoying to write on a phone
I'm pretty sure you can easily copy and paste the answers
Whatever
When I use the word possession I mean it as I defined it
It doesnt matter that it isn't the same as the regular nba use of the term
I weight offensive and defensive rebounds the same so, for my purposes, it works fine to define possession as I did
Also, as a practical consideration, there are no offensive rebound numbers in fbb
That also explains why the worth of a possession is lower in my calculations
More possessions lower the value of each possession
Ft/2 is obviously an estimation. Otherwise the calculation is more complicated and the change would be minimal
What you said about assists is true but impossible to gauge. How do you figure out which baskets were assisted and which were not? And why would a player get more points if he brought the ball up and hit a shot as opposed to a player who took a pass and hit a shot?
You could argue that the worth of an assist should be changed but in the end any number picked would be arbitrary
No formula is perfect and can only be useful if one is aware of its shortcomings
|
|
|
|
Post by greeme on Jun 10, 2011 10:28:58 GMT -5
Also, unlike per, my formula punishes players who shoot a lot at low percentages
|
|
|
|
Post by repole ಠ_ಠ on Jun 10, 2011 11:06:08 GMT -5
It is actually pretty simple Just annoying to write on a phone I'm pretty sure you can easily copy and paste the answers Whatever When I use the word possession I mean it as I defined it It doesnt matter that it isn't the same as the regular nba use of the term I weight offensive and defensive rebounds the same so, for my purposes, it works fine to define possession as I did Also, as a practical consideration, there are no offensive rebound numbers in fbb That also explains why the worth of a possession is lower in my calculations More possessions lower the value of each possession Ft/2 is obviously an estimation. Otherwise the calculation is more complicated and the change would be minimal What you said about assists is true but impossible to gauge. How do you figure out which baskets were assisted and which were not? And why would a player get more points if he brought the ball up and hit a shot as opposed to a player who took a pass and hit a shot? You could argue that the worth of an assist should be changed but in the end any number picked would be arbitrary No formula is perfect and can only be useful if one is aware of its shortcomings -Offensive and defensive rebounds should NEVER be weighted the same. If you do that you're essentially giving the defensive rebounder the credit for everything that went into creating that missed shot. -There aren't any offensive rebounds recorded in FBB, but you can estimate that about 30% of rebounds are offensive. It's rough, but it's a much better solution than weighting them all the same. -The change from FT/2 and FT*.44 isn't as minimal as you'd think, especially for guys who get to the line a ton. If you want to start using all sorts of precise numbers like .9323423423423 or whatever it was for the weight of a possession, you should use exact numbers for everything that you can. -It's not impossible to gauge that value for assists and break up the value of a made shot, in fact oRtg does it in a very reasonable manner. It is NOT an arbitrary number (PER does use an arbitrary number however) for weighting assists. It actually weights the value of assists based on FG% (higher FG% tends to imply a higher rate of baskets being assisted....this isn't perfect, but again it's far from arbitrary). If you use 82games data this estimate can be eliminated as they track the percentage of shots that are assisted for each player. -The player who brings the ball up and takes a shot has created those 2 points entirely by himself. Shouldn't he get more credit than the player who cuts to the basket, receives a pin point pass, and finishes it? -The "problem" with your formula is that it doesn't represent anything tangible, just some sort of magical value. PER represents total points produced per minute played essentially, while oRtg represents points produced per possession used. Both of these can be evaluated in a very logical sense. You use the word arbitrary, but your formula is as arbitrary as anything can be. -PER does penalize players for missed shots, in fact in penalizes them in nearly exact same way yours does. Missed field goals count as a negative value of a possession. I want to make it clear that I'm not trying to "show you up" or anything like that, just trying to broaden your (and anyone else's) view of statistics. |
|
|
|
Post by Martinez on Jun 10, 2011 11:52:44 GMT -5
wheres the next article?  |
|
|
|
Post by repole ಠ_ಠ on Jun 10, 2011 11:56:23 GMT -5
On its way....Spencer's email got sent to my spam folder for some reason. Fail.
|
|
|
|
Post by Martinez on Jun 10, 2011 11:59:00 GMT -5
cool wanted to see it before playoffs start might come in handy  |
|
|
|
Post by Biggie on Jun 11, 2011 9:51:03 GMT -5
5
|
|
|
|
Post by greeme on Jun 11, 2011 14:15:16 GMT -5
-Offensive and defensive rebounds should NEVER be weighted the same. If you do that you're essentially giving the defensive rebounder the credit for everything that went into creating that missed shot.
no you are not, not any more than you give the offensive rebounder the credit for everything that went into the missed shot. Weighting them the same is pretty easy to justify - the ball is in the air and whoever gets it, offensive or defensive player, gets possession for his team. From what you wrote it seems that you give offensive rebounds a higher weight? Why? and by how much?
-There aren't any offensive rebounds recorded in FBB, but you can estimate that about 30% of rebounds are offensive. It's rough, but it's a much better solution than weighting them all the same.
You could but that only uses the total % and not each particular players ord/drb ratio. In any case since I weight them the same it makes no difference to me.
-The change from FT/2 and FT*.44 isn't as minimal as you'd think, especially for guys who get to the line a ton. If you want to start using all sorts of precise numbers like .9323423423423 or whatever it was for the weight of a possession, you should use exact numbers for everything that you can.
Maybe, but I don't know if the .44 is an NBA number or a BBS number. If it is, as I assume, an NBA number, then I don't know how useful it is in BBS.
-It's not impossible to gauge that value for assists and break up the value of a made shot, in fact oRtg does it in a very reasonable manner. It is NOT an arbitrary number (PER does use an arbitrary number however) for weighting assists. It actually weights the value of assists based on FG% (higher FG% tends to imply a higher rate of baskets being assisted....this isn't perfect, but again it's far from arbitrary). If you use 82games data this estimate can be eliminated as they track the percentage of shots that are assisted for each player.
Obviously in real life everything can be tracked but any change will, by definition, hurt some players and help others. For the league average I agree that higher FG% tends to imply a higher rate of baskets being assisted but for any specific player it doesn't necessarily hold. In any case logic has to be applied to applying any formula so one can assume that a center shooting 60% probably is the recipient of a bunch of assists but a slashing sf shooting 52% isnt and a three point bombing pg is the recipeint of almost no assists. I would rather keep it simple and let people add in their own logical deduction than make it more complicated.
-The player who brings the ball up and takes a shot has created those 2 points entirely by himself. Shouldn't he get more credit than the player who cuts to the basket, receives a pin point pass, and finishes it?
yes, if it can be tracked. Though you could argue that getting free and cutting to the basket is also a skill which should be rewarded even if it doesn't show up in the box score. In any case since it can't be tracked I don't think it should be bothered with. As I said I don't see any formula or statistic as being perfect and logic has to be applied.
-The "problem" with your formula is that it doesn't represent anything tangible, just some sort of magical value. PER represents total points produced per minute played essentially, while oRtg represents points produced per possession used. Both of these can be evaluated in a very logical sense. You use the word arbitrary, but your formula is as arbitrary as anything can be.
not true. Just bc a formula can show something doesn't make it useful. PPG shows exactly what it says but it is useless without number of shots, shot %s etc. Are the numbers I use 100% exact? No, but that hardly makes them totally arbitrary. In the end, the best way to use a formula is to see if the results match up more or less with how we see players and if so how to explain the players who don't match up with what we think, are they better or worse than we think or should the formula be adjusted.
-PER does penalize players for missed shots, in fact in penalizes them in nearly exact same way yours does. Missed field goals count as a negative value of a possession.
as far as I understand that just isn't true.
"Lastly, PER rewards inefficient shooting. To quote Dave Berri, the author of The Wages of Wins:
"Hollinger argues that each two point field goal made is worth about 1.65 points. A three point field goal made is worth 2.65 points. A missed field goal, though, costs a team 0.72 points. Given these values, with a bit of math we can show that a player will break even on his two point field goal attempts if he hits on 30.4% of these shots. On three pointers the break-even point is 21.4%. If a player exceeds these thresholds, and virtually every NBA player does so with respect to two-point shots, the more he shoots the higher his value in PERs. So a player can be an inefficient scorer and simply inflate his value by taking a large number of shots."
|
|
|
|
Post by brophdog88 on Jun 11, 2011 14:32:46 GMT -5
and a three point bombing pg is the recipeint of almost no assists
not true
just look at Kidd in the playoffs in the NBA
|
|
|
|
Post by repole ಠ_ಠ on Jun 13, 2011 9:09:21 GMT -5
-Offensive rebound+missed FG = 0. The possession has been preserved, no total value is added or lost in missing a shot and then getting an offensive rebound.
Defensive rebound+causing missed FG = value of a possession, should be roughly equal to the value of a steal. Giving the defensive rebounder the entire credit for causing the missed shot AND getting the rebound is faulty.
These two simple concepts are pretty core to any statistical formula within the APBR community (basketball's equivalent to baseball's SABR). No one in basketball analytics will take a formula that counts defensive rebounds and offensive rebounds as being equal seriously.
[/b]
-It's not perfect, it's an estimate. It's still a lot better than counting defensive rebounds to be as valuable as an offensive rebound.
-It's an NBA number, but because And 1s exist we know that FT/2 is an overestimate of that number. .44 is an estimate as well, but it's a better estimate.
-If you want to let people make their own logical deductions, you can't create a formula that proclaims to correlate to a player's value. The whole point of the formula is to put some sort of value on the player, and yet you're saying that the person looking at it is supposed to account in their head that "oh well, that guy gets assisted on his shots a lot so he should probably be lower on this list." Saying you don't want to make your formula more complicated, to be blunt, is a poor excuse for leaving a critical concept out of it.
And yeah, as Broph said, 3 point bombers are the recipients of more assists than anyone (James Jones went half a season without an unassisted FG).
-The guy who's getting free and cutting does get credit, he still gets credit for most of the basket. Not sure what you're trying to say here.....My point is simple: The guy who scores off an assisted shot shouldn't get all of the credit for those points, plain and simple.
-I never said that a stat that shows something tangible automatically is useful. PPG tells you something tangible, but because people misinterpret it as an end all be all stat, it is frequently misused. The problem with something that isn't tangible is that it's pretty much guarenteed to be misunderstood and misinterpreted by most of the people attempting to use it.
-I think my head just exploded. This is so, so wrong, stat heads everywhere would cringe at these sentences.
You never, ever compose a formula based on your (or someone else's) perception of something. If you think LeBron James is the best player in basketball and your formula doesn't agree, you don't just arbitrarily adjust it so that LeBron comes out number 1. This sort of goes back to having a tangible formula, because when you don't, this happens.
-This is pretty much exactly what I said, except I forgot to mention that PER also accounts for the possibility that a missed shot is rebounded. Berri is entirely right that PER rewards volume players, and that's because as awful as it sounds, a guy shooting 30% is still producing points, albeit at a terrible pace. You made it sound as if missing a FG has no impact on PER, when it does.
For what it's worth, I'm not a huge fan of PER, very few people in the statistical community use PER without citing a myriad of other stats. PER isn't really something I have an expert opinion on, I'm not going to defend it as being much more than basically an MVP meter, nor do I have a full grasp on why it uses certain values in certain places.
|
|
|
|
Post by greeme on Jun 15, 2011 3:18:16 GMT -5
-Offensive rebound+missed FG = 0. The possession has been preserved, no total value is added or lost in missing a shot and then getting an offensive rebound. i agree that as far as the team rankings Offensive rebound+missed FG = 0 but the player that missed should have a negative value bc of the missed shot and the player rebounding should have a positive value.Defensive rebound+causing missed FG = value of a possession, should be roughly equal to the value of a steal. Giving the defensive rebounder the entire credit for causing the missed shot AND getting the rebound is faulty. These two simple concepts are pretty core to any statistical formula within the APBR community (basketball's equivalent to baseball's SABR). No one in basketball analytics will take a formula that counts defensive rebounds and offensive rebounds as being equal seriously. you still didn't write what their relative value is. Do you think an offense rebound has no value because it merely preserves possession? Do you think a defender causing a missed shot has no value merely bc an offensive rebound was grabbed? I agree with you that a steal equals rebound + causing a missed shot. I still think an offensive rebound = defensive rebound. the ball is in the air and whoever gets it claims possession for their team. it doesn't matter who had it before. Even though it isn't technically a change of possession it is treated as such as far as the shot clock being reset. I'm not giving the rebounder any credit for causing the missed shot just for getting the ball that was "up for grabs". If it would be easier conceptually it could be looked up as the offensive team giving up possession and the defensive team gaining possession (or the offensive team regaining possession) as two seperate things. In a steal they are one, the offensive team gives up possession and the defensive team takes possession in the same action. That is why a steal is more valuable than a rebound and equals the rebound + the worth of the defender causing a missed shot-It's an NBA number, but because And 1s exist we know that FT/2 is an overestimate of that number. .44 is an estimate as well, but it's a better estimate. k-If you want to let people make their own logical deductions, you can't create a formula that proclaims to correlate to a player's value. The whole point of the formula is to put some sort of value on the player, and yet you're saying that the person looking at it is supposed to account in their head that "oh well, that guy gets assisted on his shots a lot so he should probably be lower on this list." Saying you don't want to make your formula more complicated, to be blunt, is a poor excuse for leaving a critical concept out of it. And yeah, as Broph said, 3 point bombers are the recipients of more assists than anyone (James Jones went half a season without an unassisted FG). but no formula is all inclusive. It admittedly has no man to man defense rating. and what about the player that set a pick that caused the player to be free to hit an "unassisted" shot? and what about the fact that the defender was sagging on my center bc they don't want him to get the ball so my pg got an open look? All this stuff can't be computed in the game, and maybe not even in real life, so i don't bother. if somebody wants to try to logically deduce which player got more help in getting his points then let him, if not let him just now that it hurts guys that create thier own shots[/b
]-The guy who's getting free and cutting does get credit, he still gets credit for most of the basket. Not sure what you're trying to say here.....My point is simple: The guy who scores off an assisted shot shouldn't get all of the credit for those points, plain and simple.
I agree in principle but an assist is not the only way of making it easier to help a guy scoring and since there is no way to know who had help and who didn't then i would rather igonre it. What about factoring the relative defensive talents of the defending players? if i am at a position in which i have to deal with better defenders then that too has an impact.
-I never said that a stat that shows something tangible automatically is useful. PPG tells you something tangible, but because people misinterpret it as an end all be all stat, it is frequently misused. The problem with something that isn't tangible is that it's pretty much guarenteed to be misunderstood and misinterpreted by most of the people attempting to use it.
-I think my head just exploded. This is so, so wrong, stat heads everywhere would cringe at these sentences.
You never, ever compose a formula based on your (or someone else's) perception of something. If you think LeBron James is the best player in basketball and your formula doesn't agree, you don't just arbitrarily adjust it so that LeBron comes out number 1. This sort of goes back to having a tangible formula, because when you don't, this happens.
that isn't what i meant. I don't think you decide who is best and then write the formula. I meant that if the best player in the league, let's say dean hollis, comes out as the 100th best player then you have to figure out why. Is it a flaw of the formula? of does dean hollis have a secret flaw that the formula bares? If there is something wrong with the formula then it should be adjusted.
-This is pretty much exactly what I said, except I forgot to mention that PER also accounts for the possibility that a missed shot is rebounded. Berri is entirely right that PER rewards volume players, and that's because as awful as it sounds, a guy shooting 30% is still producing points, albeit at a terrible pace. You made it sound as if missing a FG has no impact on PER, when it does.
For what it's worth, I'm not a huge fan of PER, very few people in the statistical community use PER without citing a myriad of other stats. PER isn't really something I have an expert opinion on, I'm not going to defend it as being much more than basically an MVP meter, nor do I have a full grasp on why it uses certain values in certain places. |
|
