Author Message
back2newbelf
Joined: 21 Jun 2005
Posts: 275
PostPosted: Thu Sep 17, 2009 6:11 am Post subject: A new player rating system Reply with quote
Hi
I came up with an idea for a new player rating system.
First of all let me say that I do not believe that a single metric player rating system is that useful to judge a players' talent or contributions to the team. There are obvious flaws with many metrices, and of course
this one has its' flaws too. I do believe that watching the player play is more valuable than looking at a single number.
That said, I had much time this summer and went ahead and implemented this, using the play by play files from basketballgeek.com (Big Thanks)
WARNING:
Some of what's coming up may sound weird to many and there were assumptions made that are not 100% correct (but were somewhat needed to make this work). Most assumptions are not set in stone, if I had more time I'd probably change some of them. Right now the assumptions pretty much favor big players over small ones
GOAL:
the goal was to create a ranking where
(credit of all players of team A - credit of all players of team B) =
team_A_points - team_B_points, at the end of each game.
Every assignment of credit is either split between two players, or one player of team A and five players of team B.
ASSUMPTIONS:
-This rating does not try to split in offensive and defensive ratings, instead I made the assumption that offense and defense are somewhat interweaved, more on that coming up
-the (assumed) 2 states of a basketball game:
regular possession:
A regular possession is the offensive possession that follows a made shot, defensive rebound or the conclusion of an irregular possession.
irregular possession:
an irregular possession is the offensive possession that follows a regular possession where a steal, turnover(including charges), offensive rebound or a block(with the defending team getting the ball afterwards) happened.
This type of possession only ends when the team that turned the ball over (or failed to get the defensive board) regains possession.
-For every action where there's positive credit assigned, there also needs to be negative credit assigned. A made shot (positive credit) leads to negative credits for defenders and so on. The sum of all credits needs to be 0
-it is kind of assumed that if a player would not have grabbed an offensive rebound, the defending team would have grabbed the defensive rebound
ASSIGNMENT OF CREDIT:
regular possession:
the player that shot the ball receives credit for (the number of points he just scored MINUS average*)/2
the defending team receives -(the number of points he just scored MINUS average*)/2
*where average is: league wide average of points scored per possession
and1's and 3's are accounted for.
Let average be 1 for simplicity. If player A shoots and misses, he scored 0 points, so he gets:(0-1)/2=-1/2 credit
The defending team gets -(0-1)/2=1/2 credit, split over all 5 players (so every defender gets 1/10)
If player A hits a 3, he gets (3-1)/2=1 credit, the opponent team gets -1 credit (which gets split again)
made technical free throws are negatively credited to the player who got the technical, if the free throw is missed, he does not get negative credit.
irregular possession:
All points that result from an irregular possession get credited to the player(s) that "created that irregular possession".
If playerB1 steals the ball from playerA1(A1 and B1 are the "creators of the possession" here) and playerB2 hits a 3, playerB1 gets 3/2 credit, playerA1 gets -3/2. PlayerB2 does not get any credit.
I might change this in the future. The way I see it now: without playerB1 playerB2 would have never had the chance to score, although it could be argued that the creator of the possession needs those that effectively run the break as much as they need him
Credit for points scored after an offensive rebound gets split between the rebounder and the team that failed to grab the defensive board. Similar with turnovers that weren't also a steal.
This means that, if the steal etc. doesn't result in points, players get no credit for it. This also means that stealing is more valuable for teams that are good at running the break.
Likewise, offensive rebounds near the basket are more valuable (because they will probably result in more points)
These things do not get any credit:
Assists: with all the talk about assists being awarded pretty arbitrarily by the scorekeepers I decided to only include things that scorekeepers can't really mess with. This is also a reason why guards tend to look bad in this rating. I might change this in future versions
Defensive rebounds: instead, the entire team gets punished for the points that result from the other teams' offensive rebound.
Everybody on the defending team is punished the same in case of an offensive rebound. I guess it can be argued that low post players are more responsible for denying offensive rebounds, so in future versions I might punish players differently according to their position.
Anything that happens during a irregular possession: As already noted above If playerA1 gets an offensive rebound and playerA2 scores, all credit goes to the guy that "created" that offensive possession(playerA1). Subsequent offensive rebounds or offensive rebounds after a steal also don't get any credit (only the initial creator of that irregular possession gets credit)
SPECIAL CASES:
take the fictional 1on1 between A and B.
A misses, B gets defensive rebound: A gets -1/2, B +1/2
B misses, A gets defensive rebound: B gets -1/2, A +1/2
A misses, B gets defensive rebound: A gets -1/2, B +1/2
game over, game score 0-0
B now has +1/2 credit, A has -1/2. The difference between the two does not match the difference in the game score.
That is because A had one possession more in the game. In these cases I credit the coaches of the team that enabled their team to have an additional possession due to clever clock management. Thus the coach of A gets +1/2, coach of B gets -1/2 (for bad clock management) and the difference in the sum of credits for both teams matches the difference in game score.
Things get handled similarly for turnovers or offensive rebounds that end quarters.
what makes you look bad or good:
-having a high TS% makes you look very good, unless all your points come in possessions that were created by others
-having a low TS% obviously will make you look bad
-playing good team defense or playing with players that play good team defense makes you look good
-playing on a good defensive rebounding team will make you look good and vice versa
-unfortunately, playing matador defense will make you look better than you are. Playing with others who play matador defense will make you look worse than you are
-getting steals, blocks, charges, offensive rebounds that lead to scores will make you look good
-creating points with below league wide efficiency will make you look bad even if you shoot better than your entire team.
This punishes high scoring players on bad teams(read:Durant) too much.
-high turnover rate will make you look bad (unless the opponent team doesn't score often after your turnovers)
Unfortunately debugging the code took longer than I imagined and I have more important things to do now. As a result not all errors got fixed and there's ~1 credit point assigned wrongly for every other game. Also, I could only get 1089 of the games from last season to work
expectations:
Before I looked at the numbers I guessed that either Gasol or Dwight Howard would be #1 and Baron Davis would be dead last.
absolute:
http://spreadsheets.google.com/ccc?key= ... 4T1E&hl=en
per "regular possession":
http://spreadsheets.google.com/ccc?key= ... BVnc&hl=en
The numbers certainly need some kind of position adjustment, but i guess it's save to say that a guard is very good when he's in the upper 1/3 and a low post player is bad when he's in the lower 1/3. I find it impressive that Chris Paul is looking that good even without any credit going to assists. One could probably also say that if a player has the best rating of his team, then he's an above average player even if his rating is in the negative(he's probably suffering from getting the best opponent defender assigned to him every game).
Unfortunately there's no adjustment for strength of opponent, players who only play against starters will obviously look worse than players that play more against bench players.
Last edited by back2newbelf on Fri Sep 18, 2009 4:05 am; edited 1 time in total
Back to top
View user's profile Send private message
Crow
Joined: 20 Jan 2009
Posts: 825
PostPosted: Thu Sep 17, 2009 2:56 pm Post subject: Re: A new player rating system Reply with quote
It is good to think outside the box.
In my opinion some of your ideas may be stand better alone than the roll-up.
Points off possession changes you created would be good to see on its own.
The split credit system has some resemblance to the approach I've occasionally and recently mentioned, similar to what protrade once did briefly before succombing to fantasy basketball standards.
Seeing player stats completely without assists is worth seeing.
Back to top
View user's profile Send private message
DJE09
Joined: 05 May 2009
Posts: 148
PostPosted: Thu Sep 17, 2009 8:24 pm Post subject: Reply with quote
I find this interesting, particularly on the posession front. I think there is some value in your method, but it is very hard to get a grasp of what the number is telling me without some (more) context.
As far as I can tell, the standard definition of a Posession goes something like this:
A possession is one continuous section of the game where one team has control, or "posession" of the ball.
In particular, made baskets, turnovers (including off. fouls), defensive rebounds and shot clock violations imply a change in posession. Whilst the official rule book deems a posession to have ended when the ball hits the rim (so an OReb is a 'new' posession) this is not how a posession is treated in most metrics - not that I agree with this, I just haven't come up with a good definition yet that doesn't make things like non-shooting fouls a 'new' posession as well.
I am trying to make sense of your definitions of a "regular" posession and an "irregular" posession in basketball contexts. Because it seems you are using the "usual" definition of a posession, but then adding in OReb as new "irregular posessions" - these will not be captured in current Points per posession metrics.
Anyhow, suppose Team A makes a basket, so the next possession, Team B's, is a regular posession, but player B1 turns the ball over - passing out of bounds. There has been no shot - how do you apportion this? this is clearly a regular posession. The following play is the ball being inbounded, from a turnover, so therefore an irregular posession but there is no player on Team A who will be assigned credit for this - suppose we just carry on, and player A1 is assessed a charge - turnover, ball is now with Team B again - how does the credit / blame for that posession get allocated? Further, Team B has the ball again, but the previous posession was not a regular posession, so what do you call this sort of posession? - or is this all one irregular posession until one of the two teams shoots and scores or there is a defensive rebound - couldn't we then have a situation where the same team scores and initially turned the ball over - so they get the credit and the "blame".
I like what you have as a start, but I think there are a lot of game situations you don't seem to give credit / blame for, especially if you are adjusting to a league average points per possession (or is that the average from your own count of posessions from the PbP?).
Back to top
View user's profile Send private message
magicmerl
Joined: 30 Dec 2007
Posts: 54
PostPosted: Fri Sep 18, 2009 4:30 am Post subject: Reply with quote
Basketball on Paper defines two different things, plays and possessions that I think may be helpful to the discussion here.
A play is a sequence where one team has the ball, ending in a turnover, shot attempt, or time expiring.
A possession is a sequence where one team has control of the ball, ending when the other team gains the ball, or time expiring.
The key distinction between the two is that offensive rebounds cause a team to have two back to back plays, but the two plays are only counted as a single possession.
As far as I understand it, the reason possessions are used instead of plays is to balance up the ledger with both teams. Both team will have the same number of possessions in a game, (give or take a couple for exception teams like the Spurs that always take the 2-for-1 at the end of each quarter if they can get it).
DJE09, you seem to be groping towards defining statistics in terms of plays rather than a possession basis. I don't think that there's any real difference as long as the value of an offensive rebound is factored as being worth the same positive value as the cost of a field goal attempt.
Back to top
View user's profile Send private message
back2newbelf
Joined: 21 Jun 2005
Posts: 275
PostPosted: Fri Sep 18, 2009 7:48 am Post subject: Reply with quote
Crow wrote:
Points off possession changes you created would be good to see on its own.
Whenever I have some free time(read: not in the next 2 months) I will probably release numbers for every facet of this metric that make up the end result.
DJE09 wrote:
Anyhow, suppose Team A makes a basket, so the next possession, Team B's, is a regular posession, but player B1 turns the ball over - passing out of bounds. There has been no shot - how do you apportion this? The following play is the ball being inbounded, from a turnover, so therefore an irregular posession but there is no player on Team A who will be assigned credit for this - suppose we just carry on, and player A1 is assessed a charge - turnover, ball is now with Team B again - how does the credit / blame for that posession get allocated?
B1 is the "creator of the possession" (by turning it over).
Since we will have to assign as much positive credit as we assign negative credit we would also (potentially) need to credit* someone from A. I can't really tell which player from team A should receive sole credit, so I would credit* everybody on team A the same.
This is, of course, different from steals or charges where I could give positive credit to one single player from A.
*we're not crediting anyone yet, I'm just trying to make clear who would get credit.
back2newbelf wrote:
All points that result from an irregular possession get credited to the player(s) that "created that irregular possession".
The irregular possession ends(because B gets the ball back, see below) with a turnover, so no points came out of this irregular possession, so there is no credit, neither negative nor positive, for anyone. The game continues for B as if nothing had happened.
Had teamA scored any points, B1 would get (negative)half of it and the entire teamA would get the (positive) other half
back2newbelf wrote:
These things do not get any credit
Anything that happens during a regular possession
Even though A1 was called for an offensive foul, he does not get any negative credit here. (I sure know this seems weird and I know that players that botch fast-breaks will look better than they should, but it also tells us that B1 didn't have a horrible turnover at midcourt leading to a 1on0 fast-break)
The sequence of events is important here. Had team A started the game and A1 was called for an offensive foul in the first possession of the game he would get negative credit for half the points teamB scores following the charge.
Quote:
Further, Team B has the ball again, but the previous posession was not a regular posession, so what do you call this sort of posession?
back2newbelf wrote:
regular possession:
A regular possession is the offensive possession that follows a made shot, defensive rebound or the conclusion of an irregular possession.
irregular possession:
an irregular possession is the offensive possession that follows a regular possession where a steal, turnover(including charges), offensive rebound or a block(with the defending team getting the ball afterwards) happened.
This type of possession only ends when the team that turned the ball over (or failed to get the defensive board) regains possession.
Team B was the team that turned it over. They get the ball back after A1 gets called for an offensive foul, the irregular possession ends, another regular possession begins
Quote:
I like what you have as a start, but I think there are a lot of game situations you don't seem to give credit / blame for, especially if you are adjusting to a league average points per possession
I adjust to average "points per regular possession" instead of points per possession. Only the regular possession shots get adjusted anyway
Back to top
View user's profile Send private message
DJE09
Joined: 05 May 2009
Posts: 148
PostPosted: Fri Sep 18, 2009 4:37 pm Post subject: Reply with quote
Yeah, that is what I thought was your logic, when no points are scored, there is no credit either way.
But ... you seemed to be saying when a shot was taken, and missed, there was credit assigned - oh, only from a regular posession, hence the adjustment factor, and we should view any credit that stems from an irregular posession as "bonus".
Merl, I would have looked up Basketball on Paper, if I had my copy ... Can you pass it on to Dave and Lou - they are coming over this week Smile
Back to top
View user's profile Send private message
back2newbelf
Joined: 21 Jun 2005
Posts: 275
PostPosted: Sat Sep 26, 2009 8:22 am Post subject: Reply with quote
New version is out with following changes:
-Last 24 seconds of a quarter are disregarded. I figured a lot of go-to-guys were unfairly punished for shooting bad percentages at the end of quarters.
-credit for assisted baskets now gets equally split between the scorer and passer.
http://spreadsheets.google.com/ccc?key= ... xX2c&hl=en
Data is now for the last 3 seasons. (sheet 1 is 08-09, sheet 2 is 07-08 etc).
You should be able to sort columns to look at teams.
Some notes:
-Leon Powe finished 23rd, 1st and 3rd.
-Chris Paul finished 27th, 2nd and 1st.
-Balkman, Ariza and Moon(TOR) look very strong.
-Boozer went from looking good in 07 to average in 08 to quite bad in 09.
-Charlie V, Bargnani and Yi look horrible.
-Billups, Maxiell and Amir Johnson were the best rated Pistons for years.
Back to top
View user's profile Send private message
DJE09
Joined: 05 May 2009
Posts: 148
PostPosted: Sat Sep 26, 2009 8:53 am Post subject: Reply with quote
back2newbelf wrote:
DJE09 wrote:
Anyhow, suppose Team A makes a basket, so the next possession, Team B's, is a regular posession, but player B1 turns the ball over - passing out of bounds. There has been no shot - how do you apportion this? .....
B1 is the "creator of the possession" (by turning it over).
Since we will have to assign as much positive credit as we assign negative credit we would also (potentially) need to credit* someone from A. I can't really tell which player from team A should receive sole credit, so I would credit* everybody on team A the same.
You addressed most of my questions - except the first one - Team B is in a regular posession - when players shoot and make or shoot and miss, credit is assigned (positive / negative respectively). But you have no (explained) methodology for assigning credit when a regular posession is turned over. If you assign zero credit, you are in effect saying that turning the ball over is less harmful to the team than missing a shot ... I can't see that, surely it is at least as harmful, to my thinking it is more so, but I can see how in your methodology it would be the same. Is that the case, do you treat such turnovers as missed shots?
I am not sure it is appropriate to apportion credit for assists on made shots, since you do not apportion negative credit for assists on missed shots to the passer, but surely they are equally to blame if the equally deserve credit. In fact, since a very large percentage of 3s are assisted (leading to the inferrance that almost all 3pt attempts are potentially assisted), it would be reasonable to assume that wether a 3 pt shot is made is almost totally dependant on the ability of the player shooting to make the shot.
I guess I am feeling like a lot of your "decisions" as to what is in / out are arbitrary, and in fact are determined once you have looked at the numbers, and decided if you like them or not (sometimes referred to as the smell test). If you have a systematic way, or a specific methodology you are following to generate these changes in your method, I appologise, as you have already indicated you will reveal more once you have more time Smile
Personally, I think PG feature too highly in your new list. At least your old list was comprehensible in that it rewarded efficency and scoring in irregular posessions, I can't see that now.
Back to top
View user's profile Send private message
back2newbelf
Joined: 21 Jun 2005
Posts: 275
PostPosted: Sat Sep 26, 2009 9:55 am Post subject: Reply with quote
DJE09 wrote:
back2newbelf wrote:
DJE09 wrote:
Anyhow, suppose Team A makes a basket, so the next possession, Team B's, is a regular posession, but player B1 turns the ball over - passing out of bounds. There has been no shot - how do you apportion this? .....
B1 is the "creator of the possession" (by turning it over).
Since we will have to assign as much positive credit as we assign negative credit we would also (potentially) need to credit* someone from A. I can't really tell which player from team A should receive sole credit, so I would credit* everybody on team A the same.
You addressed most of my questions - except the first one - Team B is in a regular posession - when players shoot and make or shoot and miss, credit is assigned (positive / negative respectively). But you have no (explained) methodology for assigning credit when a regular posession is turned over.
back2newbelf wrote:
If playerB1 steals the ball from playerA1(A1 and B1 are the "creators of the possession" here) and playerB2 hits a 3, playerB1 gets 3/2 credit, playerA1 gets -3/2. PlayerB2 does not get any credit
As said before, if B doesn't score A1 would not get negative credit.
Any action (except for non blocked shots) by player Z that gives the other team possession leads to an irregular possession for the other team, with Z being responsible for half the points scored in that irregular possession.
Quote:
I am not sure it is appropriate to apportion credit for assists on made shots, since you do not apportion negative credit for assists on missed shots to the passer, but surely they are equally to blame if the equally deserve credit.
This is not possible with play-by-play-data.
The assists were arbitrarily added because
-I figured I should reward players for handling the ball, bringing the ball up-court and so on (assists are the closest thing to this I guess).
-It was easy to implement and gave me a rough position adjustment
Quote:
I guess I am feeling like a lot of your "decisions" as to what is in / out are arbitrary
Sure. I just wanted some ground work done to get a feel for the whole thing, see what makes sense in a logical way and what doesn't, and go from there.
To get to to final rating system I will probably mess with coefficients for every possible thing players can get credit for, to finally get the coefficients which produce the best retrodicton results, while keeping the two main ideas that
(the sum of credit of team A - sum of credit of team B) = scoreA - scoreB
and that whenever there's negative credit assigned, the same amount of positive credit will be assigned and vice versa.
Quote:
Personally, I think PG feature too highly in your new list. At least your old list was comprehensible in that it rewarded efficency and scoring in irregular posessions, I can't see that now.
There are 3 PGs in the top 20 in '07, 5 in '08, 5 in '09, which make up for 21% (when in a perfect world it would probably be 20%). The top of '09 sure seems dominated by PGs but I'd say that's random chance.
Back to top
View user's profile Send private message
DJE09
Joined: 05 May 2009
Posts: 148
PostPosted: Mon Sep 28, 2009 9:59 pm Post subject: Reply with quote
You seem to be missing my point, or I have misunderstood your logic.
Assume Team A is in a regular posession. If player A1 shoots the ball and misses, you assign Negative credit to this player - right?
But instead, for a REGULAR posession, player A1 turns the ball over, you don't assign negative credit to player A1 despite their usage of a regular posession.
You only assign negative credit to such players from the resultant "Irregular posession", and only if their opponent scores.
I have just had the thought, is this because this has now become an "Irregular posession" ie the whole sequence of play from Team A begining their 'regular posession' becomes part of the irregular posession?
I guess I don't understand why you would penalise a player (Assigning Total blame to that one player too) for missing a shot, but not for turning the ball over. This just doesn't make sense as a method of evaluating basketball skill.
Back to top
View user's profile Send private message
back2newbelf
Joined: 21 Jun 2005
Posts: 275
PostPosted: Tue Sep 29, 2009 3:35 am Post subject: Reply with quote
DJE09 wrote:
You only assign negative credit to such players from the resultant "Irregular posession", and only if their opponent scores.
Ah.
I do this because I find not all turnovers are the same. There are different factors which influence what happens after a turnover.
the main factors are:
-where the turnover happened (the further away from your own hoop, the better, I suppose)
-if you're playing on a good/bad transition defense team
-if you're playing against a good/bad transition offense team
I want to punish players that turn it over at halfcourt, leading to a 1on0 fast-break.
Players will look better here if they realize and react to the fact that they're playing against a good transition team while having bad transition defense. The reaction would be safer passes
A "very good" turnover would be being stripped under the opponents hoop by a slow and offensively below average team, then running back yourself and give more effort on defense.
A "very bad" turnover would be both of those:
You may also look at it this way:
If the opponent scores an average amount of points during their irregular possession the guy that turned it over gets exactly as much negative credit as if he had shot and missed.
Thus, players who have mostly "good" turnovers will get a slight boost in the rating.
Players who have mostly "bad" turnovers will look slightly worse
Back to top
View user's profile Send private message
DJE09
Joined: 05 May 2009
Posts: 148
PostPosted: Tue Sep 29, 2009 5:15 am Post subject: Reply with quote
I agree about turnovers being different depending on context, what I was confused about was your decision to assign negative credit only for missed shots on regular posessions - to the player who shot.
But in all other contexts where we would usually consider a player to have "used" a posession there is only credit assigned - positive and negative - when points are scored.
I was confused at first, because I thought you assigned credit positive/negative for all Regular posessions used, but you only do that for regular posessions that end in a shot. Or, you probably say the credit assigned is zero for all other situations.
On another note, I also assumed that your metric was about trying to assign credit / blame for the scoring that occurs during a game, if plays don't result in scoring they are not relevant. I guess this is why I am confused about assists becoming part of the measure - this is a radical change, given 3pt shots are almost always assisted it seems that this would signifcantly devalue players that play a principal role of shooting 3s in their offence, eg Rudy Fernandez.
Back to top
View user's profile Send private message
back2newbelf
Joined: 21 Jun 2005
Posts: 275
PostPosted: Tue Sep 29, 2009 6:08 am Post subject: Reply with quote
DJE09 wrote:
I agree about turnovers being different depending on context, what I was confused about was your decision to assign negative credit only for missed shots on regular posessions - to the player who shot.
I see it this way: if A turns it over it creates a break from the regular possession: the regular possession isn't finished. It continues once A gets the ball back.
So there will be credit assigned at the end of *every* regular possession, for me the regular possession just never ends with a steal, but only with a shot (that wasn't blocked)
Quote:
On another note, I also assumed that your metric was about trying to assign credit / blame for the scoring that occurs during a game, if plays don't result in scoring they are not relevant.
misses in a regular possession are relevant
(the fact that botching an irregular possession does not yield negative credit is one of the biggest problems I have with this metric and will be changed sometime)
Quote:
I guess this is why I am confused about assists becoming part of the measure - this is a radical change, given 3pt shots are almost always assisted it seems that this would signifcantly devalue players that play a principal role of shooting 3s in their offence, eg Rudy Fernandez.
I wanted to reward players for bringing the ball close to the opponents' hoop, because they are more likely to have turnovers (and thus look bad in the rating). Unfortunately there's no "X brought the ball up-court" etc. in the pbp so I took the closest thing, assists, instead.
I wanted to follow my original idea to split everything either in half or in fifths, that's why the passer also gets half for 3s.
I think you could even make a bit of a case for this with 3s:
If I'm a guard and I'm running the break 3on1, I can pass to my PF who dunks it or I can pass it to my SG who shoots a 3. My PF has more of a chance to make the shot, but let's say the 3 of the SG has a higher eFG%(he is making the shot with >33%). Shouldn't I get at least some reward for making the right decision?
(I'm not saying half is perfect)
Let's say we always have the same situation, same SG and same PF, only the point guard is changing. The SG hast the higher eFG%.The team whose PG passes to the SG the most will yield the best differential.
If I credit every assist the same, the team with the "best" PG will have the best differential, but it will not show in the PGs rating at all (everything will have come from the SG). That would seem a bit weird to me.
Certainly it would be nice to know who was the passer before a missed shot, but this is not included in pbp's.
Back to top
View user's profile Send private message
back2newbelf
Joined: 21 Jun 2005
Posts: 275
PostPosted: Sun Oct 25, 2009 9:23 am Post subject: Reply with quote
I had the time to make my program spit out all the details, so here they are.
http://docs.google.com/leaf?id=0B31RF2r ... YjIy&hl=en
I grouped alot of things together which thought would make sense.
Average in one category is not always 0, sometimes everyone is negative.
Most things are listed as
|player_name|sum of differential in this category|sum of differential divided by # of regular possessions|# of regular possessions.
Adjusting to # of regular possessions is certainly not perfect but it has to do for now.
negative credit from points after opponents' offensive rebounds is adjusted to # of opponents' misses
Shot defense is adjusted to shots taken against this player
"stealing" is points created through steals
"assists to" is credit for assists plus (negative) credit for TOs
"negative from oreb" is negative credit accumulated from opponents' points after opponents' offensive rebounds
"irregular" is points created through irregular possessions(stealing, taking charges, blocking, offrebounds..)
"shot defense" is positive credit from opponents' (non-blocked) misses plus negative credit from opponents' made shots (divided by # of opponents' shots against this player)
"shooting" is positive credit from (assisted) made shots plus negative credit for missing shots.
"self creators" is credit for unassisted baskets divided by credit for assisted baskets.
"overall" is everything put together
Back to top
View user's profile Send private message
back2newbelf
Joined: 21 Jun 2005
Posts: 275
PostPosted: Sat Dec 05, 2009 7:07 am Post subject: Reply with quote
This years' possession adjusted leaders so far:
NOH_Chris Paul
LAL_Pau Gasol
DET_Ben Wallace
POR_Greg Oden
BOS_Rajon Rondo
ATL_Josh Smith
CLE_LeBron James
TOR_Amir Johnson
CHA_Nazr Mohammed
ORL_Jason Williams
HOU_Chuck Hayes
HOU_Kyle Lowry
ORL_Dwight Howard
SAS_Tim Duncan
DEN_Nene Hilario
DAL_Erick Dampier
CLE_Anderson Varejao
PHX_Steve Nash
POR_Joel Przybilla
ORL_Jameer Nelson
Jennings looks pretty average, I would guess he looks average in all ratings considering how he shot the ball in the last 10 games.
Ariza went from being one of the top rated SFs in the last years to one of the worst. I guess he's not comfortable with his role (yet)
bottom:
MIN_Jonny Flynn
PHI_Jason Kapono
BOS_Rasheed Wallace
WAS_Nick Young
DET_Austin Daye
SAC_Andres Nocioni
PHI_Jrue Holiday
LAL_Josh Powell
NOH_Marcus Thornton
GSW_Stephen Jackson
NOH_Julian Wright
MIN_Wayne Ellington
TOR_Antoine Wright
WAS_Randy Foye
MIN_Corey Brewer
CHI_Jannero Pargo
GSW_Vladimir Radmanovic
NJN_Terrence Williams
MIN_Sasha Pavlovic
NOH_Bobby Brown
MIL_Jodie Meeks
in the shot defense-without-blocking-department a whole lot of Toronto and Philly players look bad. Josh Smith and Zaza Pachulia look bad too, I guess Smith is making up for it with his blocking. This stat might suggest that he gambles for a lot of blocks though.
Stephen Graham is 8th worst which is weird considering he is playing for a team that is leading the league in DRTG.
10 Laker players in the top 21
(note: I have no homecourt adjustment implemented here, the Lakers had an incredibly lopsided H-A schedule)
Many Hornets players are getting scored on alot after the opponent already missed the initial shot (and then grabbed the offensive board). Speights near the bottom, just as last year.
Kris Humphries, Shaquille O'Neal, Pau Gasol, Travis Outlaw, Manu Ginobili, Donte Greene are best in that category.
back2newbelf
Joined: 21 Jun 2005
Posts: 276
PostPosted: Sat Apr 24, 2010 5:39 pm Post subject: Reply with quote
Even though this metric will undergo some changes in the (hopefully near) future I still decided to publish the rankings from this regular season, simply because last years' rankings were really helpful to predict team wins for the 2009-2010 regular season, even though it made really bad predictions for Oklahoma and New Jersey.
I feel like it helped the most with
-Boston (not seeing a valuable addition in Rasheed Wallace*)
-Philly (the fact that Andre Miller was importand for them*)
-Milwaukee (didn't see valuable players in Charlie V and Jefferson*)
-Detroit (Charlie V again*)
-New Orleans
*you can obviously argue that those teams did bad for enterely different reason, this is just my take on it
https://docs.google.com/leaf?id=0B31RF2 ... MmE1&hl=en
(let me know if it doesn't work)
most files have the following format:
name|differential created through specific action|sum of possessions|differential created by specific action per possession
Unfurtunately this metric has a crush on low post players and point guards, so that will be the story for most teams.
For some weird reason white Power Forwards look really strong, there are three of them in the top 5 (possession based).
What surprised me the most is that, even though his TS% is really high, this metric still doesn't like Durant. He made a big jump from last year though. A lot of high volume scorers don't look good: Bryant, Carmelo, Ellis, Nowitzki.
This years' all NBA team would be Lebron, Rondo, Ginobili, Ben Wallace and Dwight Howard.
Comments on some of the files:
-assists vs turnovers:
This is credit for assisting added to the negative credit for turning the ball over. If you want to look for why Durant doesn't look strong , this is the main reason. (Possession based) he is surrounded by players like Tyson Chandler, Tyrus Thomas and Amare'. His TOV% is low but his turnovers seem to result in alot of opponent points (through fast breaks).
-creating extra possessions/offensive rebounding:
I had never heard of Jon Brockman before but his OREB% is pretty sick (18.2%)
-negative from opponent offensive rebounds:
Dallas, Cleveland, Orlando and the Spurs shining, with GSW, PHX, DET on the other side of the spectrum. Surprised to see Chris Andersen and J.R.Smith near the bottom.
-shot defense(opponent TS% in regular possessions):
Spurs, Orlando and Cleveland shining again. Antonio McDyess is 3rd best in preventing offensive rebounds and/or preventing baskets after offensive rebounds and he's 2nd best in this category. Quite impressive
Team Comments (the words "efficient" and "productive" are used with regard to this metric):
Atlanta seems to be about Horford and Smith, fortunately with no one to drag them down.
Boston seems to be pretty much all Rondo, with Wallace dragging.
Chicago seems to be Noah and Gibson, with a variety of wings and guards dragging.
Cleveland seems to be Lebron and Varejao, with no one really dragging, Jamison looks bad though, especially in comparison with Varejao.
Dallas managed to ship off super inefficient Howard for the super efficient Haywood, but Butler is negative.
Lawson looks awesome for a rookie in Denver, with Smith being the only big negative.
Detroit has Wallace and Jerebko in the green, but everybody else is negative. 4 players in deep red, two of them being offseason additions and one of them being Rip Hamilton who just barely missed being the worst player in the league (production wise). He has been that bad for the last several years.
Houston with several productive players: Lowry, Hayes, Landry and Battier but unfortunately just as many being quite unproductive, "led" by Ariza and MIP Aaron Brooks.
Earl Watson is awesome for Indiana. Foster has been one of the most efficient for several years but he's never playing much.
Camby was pretty much the only positive guy in LAC, Kaman and Rasual Butler are among the 6 most unproductive players in the league.
Miami has several guys in the positive with Wade leading the way. Unfortunately Beasley is dragging them way down
Minnesota with only Love significantly positive, Brewer being the 4th worst player in the league, Flynn being the absolute worst (in overall production).
The blame in New Jersey is split between a whole bunch of guys, with no one being in the deep red (mostly, though, because of limited playing time)
This metric says Oklahoma improved because everybody on their team improved. Collison, Westbrook and Ibaka looking strong.
Orlando has a gazillion players in the positive, only Lewis seems to be dragging a tiny bit.
Roy, Miller and Camby looking good in Portland, Webster being the only slight negative.
Nocioni drags the Kings down, he's been doing that for a couple of years now(in Chicago too)
Blair looks good for for a rookie on the Spurs, who get their biggest lift from Ginobili and Duncan.
Johnson looks really awesome for Toronto but he's not playing that much (foul trouble?). Instead, Bargnani is playing. He's the 5th worst rated player.
Both Brewer and Maynor looked pretty strong in Utah, both were shipped, Maynor now looks strong in Oklahoma.
Haywood seemed to be the lone productive player in Washington. He was in the top10 in overall production this year. That's quite a loss.
Back to top
View user's profile Send private message
Crow
Joined: 20 Jan 2009
Posts: 825
PostPosted: Sat Apr 24, 2010 6:04 pm Post subject: Reply with quote
What is the difference between "overall" and "rating"? Do both include all of defense consistent with your earlier statement?
"Shot defense" is strictly shot defense and nothing beyond it, right?
Back to top
View user's profile Send private message
back2newbelf
Joined: 21 Jun 2005
Posts: 276
PostPosted: Sat Apr 24, 2010 7:29 pm Post subject: Reply with quote
Crow wrote:
What is the difference between "overall" and "rating"?
None, should have made that clearer, sorry. "Overall" is sorted by production over the entire season, while "rating by possession" is sorted by production/possessions
Quote:
Do both include all of defense consistent with your earlier statement?
Both include defense as was explained in my first post.
Quote:
"Shot defense" is strictly shot defense and nothing beyond it, right?
It's credit for opponents missing shots added to (negative credit) for opponents making shots. Regular possessions only
Back to top
View user's profile Send private message
Mike G
Joined: 14 Jan 2005
Posts: 3625
Location: Hendersonville, NC
PostPosted: Sun Apr 25, 2010 7:01 am Post subject: Reply with quote
Quote:
Kaman and Rasual Butler are among the 6 most unproductive players in the league.
Can you offer a layman's description of how Kaman is "most unproductive"?
He's easily his team's most prolific scorer. He rebounds.
Could an ESPN talking head explain to a viewing audience how it is that Kaman is really "unproductive"?
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
back2newbelf
Joined: 21 Jun 2005
Posts: 276
PostPosted: Mon Apr 26, 2010 12:44 pm Post subject: Reply with quote
Mike G wrote:
Can you offer a layman's description of how Kaman is "most unproductive"?
I would say:
If he's asked to create shots for himself or others the end result is worse than average. Unfortunately, either because he either thinks he's doing a good job at it, or because he's frequently asked to do it , he's trying to create shots alot which probably hurts his team more than if he tried it less often
(he is 12th on the Clippers roster in ORtg, yet has the highest Usg% of everybody)
Back to top
View user's profile Send private message
back2newbelf
Joined: 21 Jun 2005
Posts: 276
PostPosted: Sun Sep 12, 2010 10:59 am Post subject: Reply with quote
finally got around doing some tests with this metric. retrodiction performance was tested on march & april for the last four years, only using data through february from the season we want to retrodict.
x = performance if we had predicted the home team to win by 3.61 points every game
y = performance of my method
z = performance of homecourt-adjusted point differential
all players were included, no matter how many minutes they had played before.
Code:
2007:
games 342
x y z
RMSE 13.8 13.52 13.05
r-squared 0 0.048 0.11
2008:
games 350
x y z
RMSE 14.4 13.41 11.81
r-squared 0 0.133 0.33
2009:
games 339
x y z
RMSE 12.7 12.05* 11.56
r-squared 0 0.098* 0.18
2010:
games 337
x y z
RMSE 13.63 12.24 11.92
r-squared 0 0.22 0.24
*Joe Sill reported an RMSE of 11.47 (R-squared 0.169) for the same time period for regularized adjusted +/-.
He also reported an RMSE of 12.76 for non-regularized adj +/- with a single season of data, and 12.01 (R-squared 0.088) with multiple seasons of data
Unfortunately it seems I don't have all the games he had. I know this because the RMSE I get for the simple method(x) differs from his(my 12.7 vs his 12.58 ).
Still it seems that my method seems worse than regularized adj. +/- and better than non-regularized adj. +/-. It never outperforms point differential, which is ultimately what I want. Nevertheless, I find the results quite promising since I haven't messed with the weights at all so far
Last edited by back2newbelf on Mon Sep 13, 2010 2:29 pm; edited 2 times in total
Back to top
View user's profile Send private message
jsill
Joined: 19 Aug 2009
Posts: 73
PostPosted: Sun Sep 12, 2010 3:45 pm Post subject: Reply with quote
Quote:
Unfortunately it seems I don't have all the games he had. I know this because the RMSE I get for the simple method(x) differs from his(my 12.0 vs his 12.58 ).
How many games do you have in your dataset for March/April of 2009? You should have 348 games.
By the way, I had extracted the final score from play-by-play logs, and there was a small error in the log for one of the games,. With that error fixed, the RMSE that I get if you simply predict the mean home margin of 3.61 points is actually 12.565.
Back to top
View user's profile Send private message
back2newbelf
Joined: 21 Jun 2005
Posts: 276
PostPosted: Sun Sep 12, 2010 6:26 pm Post subject: Reply with quote
jsill wrote:
How many games do you have in your dataset for March/April of 2009? You should have 348 games.
As listed in my earlier post I had 314 games. I found some more game logs and now have:
- 339 games (so still some games missing)
- a base RMSE (method X) of 12.70
- R-Squared of 0.125 for my method
Updating my earlier post..
jsill, do you have the R-Square numbers for 2007/2008/2010 for regularized adjusted +/-?
Back to top
View user's profile Send private message
jsill
Joined: 19 Aug 2009
Posts: 73
PostPosted: Sun Sep 12, 2010 9:07 pm Post subject: Reply with quote
Quote:
jsill, do you have the R-Square numbers for 2007/2008/2010 for regularized adjusted +/-?
First, let me check something. Are you generating your ratings using data from the entire season (including March and April) and then testing the predictions on March and April, or are you generating your ratings only using data through February and then testing on March and April? The question also applies for your method z (homecourt-adjusted point differential).
Back to top
View user's profile Send private message
back2newbelf
Joined: 21 Jun 2005
Posts: 276
PostPosted: Mon Sep 13, 2010 4:16 am Post subject: Reply with quote
jsill wrote:
First, let me check something. Are you generating your ratings using data from the entire season (including March and April) and then testing the predictions on March and April, or are you generating your ratings only using data through February and then testing on March and April? The question also applies for your method z (homecourt-adjusted point differential).
unless there is a bug in my code I'm generating through February to "forecast" March 1st, then generate through March 1st to "forecast" March 2nd and so on
Back to top
View user's profile Send private message
jsill
Joined: 19 Aug 2009
Posts: 73
PostPosted: Mon Sep 13, 2010 9:35 am Post subject: Reply with quote
OK. Well, it's not an apples-to-apples comparison to what I did. I obtained the player ratings one time, based on data through February, and then forecast March and April.
Back to top
View user's profile Send private message
back2newbelf
Joined: 21 Jun 2005
Posts: 276
PostPosted: Mon Sep 13, 2010 2:27 pm Post subject: Reply with quote
you're right. I updated my earlier post with the correct numbers. here they are again
Code:
2007:
games 342
x y z
RMSE 13.8 13.52 13.05
r-squared 0 0.048 0.11
2008:
games 350
x y z
RMSE 14.4 13.41 11.81
r-squared 0 0.133 0.33
2009:
games 339
x y z
RMSE 12.7 12.05 11.56
r-squared 0 0.098 0.18
2010:
games 337
x y z
RMSE 13.63 12.24 11.92
r-squared 0 0.22 0.24
Back to top
View user's profile Send private message
gabefarkas
Joined: 31 Dec 2004
Posts: 1313
Location: Durham, NC
PostPosted: Sat Sep 18, 2010 9:47 pm Post subject: Reply with quote
I was reading something else recently, and it reminded me of this discussion and the whole notion of model validation. So, I wanted to share it with the group:
http://www.stat.columbia.edu/~cook/mova ... els_m.html
Gelman's blog is usually pretty good for some stimulating tidbits. And I'm not just saying that because I'm a Columbia alum...