Reality is merely an illusion, albeit a very persistent one. ~Albert Einstein
A while back, I gave you, my readers, homework with the implicit understanding that there would be a reckoning. A reckoning that takes the form of my next series. Hopefully, it’ll live up to it’s predecessors. So get ready, because here comes the intro.
I aim high with sequels
The genesis of this series came from an article on Carmelo Anthony by Jeremy Wagner of Roundball Mining Company:
The motivation behind this piece is that the author, cannot believe the premise that Carmelo Anthony is inefficient. While he admits that he in most cases believes in statistics, in this case the statistics do not match his conclusions drawn from direct observation. Faced with this quandry, he decided that he was going to take a look at the numbers himself.
We here at Arturo's Silly Little Stats applaud the use of the scientific method (Image courtesy of xkcd.com)
While I admire his efforts I have some issue with the conclusions drawn (and we shall see that I’m not the only one). Carmelo may be one of the leading scorers in the NBA but we as we shall see volume does not equal quantity or efficiency.
His data sample was as follows:
- Looked at Six Players:
- Carmelo Anthony
- LeBron James
- Kevin Durant
- Dwyane Wade
- Kobe Bryant
- Kevin Martin
- Take a representative sample of 10 games consisting of :
- 2 playoff games
- 3 post all-star break games
- 5 pre all-star break games.
He looked at every shot and determined whether it was open or contested and if it was made or not. The results look like this:
So Melo is the worst at getting open looks (although he’s the best at converting them, useful skill if he took more open shots, wait…). He’s also 4th in this group at converting contested shots. The last column I added and it shows Melo at 5th out of the 6 at converting shots into buckets (not a surprise to WP devotes who know that Martin had an of year last year). So given the data set the conclusion that Melo while talented at converting open j is prone to making bad basketball decisions and that dooms him to be an inefficient scorer seems totally inevitable right? Not quite. I quote:
“In conclusion, the opportunity for Carmelo to become a more efficient scorer is there for the taking. I think the fact that he settles for so many contested shots does provide the bridge between the beautiful offensive game we see night in and night out and the harsh reality of statistical analysis.
Whether Melo ever becomes more efficient or not, I think there is room for both the subjective admiration of the beauty of Carmelo’s offensive game, while acknowledging his objective shortcomings. Carmelo may never prove to be able to overcome his score at will alpha dog mindset I believe clouds his decision making and shot selection, however, his skill level is truly that of an elite player and I think it is appropriate to marvel at his abilities at the same time we are frustrated by his willingness to go about his business in an inefficient way.”
The potential argument. Carmelo could be a more efficient scorer. Here’s the problem with this argument: He hasn’t been, he isn’t and the weight of historical evidence suggest he never will be.
I wasn’t the only one who noted this article. WOW network member Nerdnumbers (Andres Alvarez) said very eloquently in response:
“A subject I’ve addressed a few times (see here, here, and here) I can not tell you how much I love this analysis. Essentially it boils down like this
1) Melo is very good at making open shots (the horse argument)
2) Melo doesn’t take a lot of open shots
3) Therefore Melo is not an efficient scorer by “conventional metrics”
4) BUT Melo is still an Elite Player because he has the SKILL to be amazing (e.g. if he took the shots Durant or Lebron took his offense would be off the charts)
Man I was sold until point four. I felt reaffirmed by this post by the way. ….
…..The net effect of Melo is that he is not actually a good shooter. Of course my biggest issue was point four. When you have a rookie making very little (even #1 picks only have to earn around 3 wins to be worth it and of course their contracts are set so that it’s easy to drop them after two years) and they have potential then point four is great. When you have a seven year veteran that is in the middle of a max contract asking for another max contract then this point should be thrown out the window. The issue isn’t that a better coach or team couldn’t make Melo better. The point is that the risk and cost isn’t worth it. Melo will cost 1/3 of your cap space (possible more depending on the new CBA).”
I also loved the research and the method and I decided I was going to take it a bit further (I know shocker right). There is an inneffable calculus that makes a person an efficient generator of offense. The end goal is simple, the player needs to find the best possible look or find the man that has it and get him the ball without turning it over. The details, is he open, is it a three, will he get an and 1, etc., are not so simple. If the player is successful his team scores. The needs of the many do indeed outweigh the needs of the few or the one.
I’m going to attempt to quantify this very simply. My method will be as follows:
- Look at all the data for the 2008, 2009 & 2010 for every player and get:
- Player Position
- Minutes Played (and eliminate all players with less than 1200 Minutes Played which leaves 383 players)
- Pts48 (points per 48 minutes)
- FGA48 (field goal attempts per 48 minutes)
- Pts per Attempt (pts per attempt = pts48/fga48)
- AST48 (assists per 48 minutes)
- TO48 (turnovers per 48 minutes)
- Offensive possesions used per 48 (FGA48+AST48+TO48 this is the possesions spent by the player)
- Offense Generated (Pts +Asst *2.68) per 48. The 2.68 is the average points generated per FGA for 2008 thru 2010
- Offense generated per possesion used. This is the key measure as it reflects how many points the team generates when the player in question gets the rock.
- Offense Generated at 30 possesions used. Here I’m just projecting every player at an even number of posessions.
How does this look? Let’s look at the players from the above example:
Given the same number of possessions to spend Melo is tied for the least efficient of the group. Durant who’s tied with him improved dramatically over the course of 2010 (as would be expected from a young player). Melo only ranks 235 amongst the players in the data set. To be fair that has something to do with his position.
And we’ll get into it.
In future installments.
Addemdum (10/29/2010): My readers are fantastic. Two of them recommended an upgrade to the metric. The comments were :
some dude:
you include FT into points in points per possession, but does a shooting foul for 2 or 3 shots equate to a possession in the possession part of the equation?
if not, there’s a piece missing. You’re adding in points for a possession while not counting the poss
Man of Steele:
Arturo, I’m inclined to agree with you in thinking of FTs as the result of good choices made by the player. It seems unavoidable, though, that you may have to adjust this theory in some instances. It is not a smart choice for Shaq to get to the free throw line. In fact, more often than not he misses at least one of two free throws, after which the other teams often regains possession. In effect, it is quite possible that a poor free throw shooter is utilizing resources (possessions) without producing any profit (points).
Of course, this caveat really only applies for small percentage of the population. Most players shoot free throws well enough that it is a good idea to get to the line.
They’re right of course. As I said, my readers equal fantastic. So we are going to add a .44 *FTA48 term (an approximation for possessions used thru Free Throws from Prof. Berri) an call it the SD-MOS hack-a-shaq correction. The numbers now look like:
Carmelo is tied with Durant at the rear of the group (Durant is young and improving however).
And everybody on the list ends up doing worse. Live and learn. Point Guards are coming next.
Part 2 (PG) is here.
Arturo's Silly Little Stats v2.0 
Matthew Yglesias
10/28/2010
I think you need some kind of free throw term in there. A player could register 3 points of 3/4 free throw shooting with zero field goal attempts, and your method would end up with an undefined points per shot of 3/0.
Matthew Yglesias
10/28/2010
I should clarify: Obviously over the course of a season nobody’s going to play 1200 minutes and have no field goal attempts, so if you run the numbers in practice you’ll never get an undefined term. But the thought experiment illustrates the point that if you ignore free throw attempts you’re missing something that matters.
arturogalletti
10/28/2010
Matthew,
Good question.
I’m treating free throws as a bonus for smart decision making by the player. So made free throws are embedded in the point totals (as we’ll as in the assist approximation). So in essence I’m looking for a simple cost benefit analysis where cost is equal to FGA + Assists + Turnovers and benefits is points by the player or teammate (which I’m approximating by the asst48*2.68 term). For the purpose of the exercise, FTA don’t actually represent expended possessions and I’m confortable ignoring them.
Fred Bush
10/28/2010
Except if a guy scores very efficiently, but primarily off of assists, then you don’t necessarily want to get him the ball per se, you want to get the passer the ball, right?
arturogalletti
10/28/2010
We’re going to do this by role. But the point is how the possesion ends.
Fred Bush
10/28/2010
Who ends the possession on an assist?
arturogalletti
10/28/2010
Brilliant question. For the point of view of this exercise I’m giving full credit to the passer. This means I’m looking at it purely from the players perspective. Some buckets are being counted twice. If I was doing it for a team i’d look into splitting credit.
some dude
10/28/2010
you didn’t adjust for pace, did you? I don’t like using PER48 when pace isn’t adjusted.
arturogalletti
10/28/2010
I did by making everything per 48 and making it by possesion. The math works (trust me).
some dude
10/28/2010
Also, one issue i have is the “dump off shot.” What I mean by this is how many of those shots taken, that are open or contested, are shots given to he player in a bad spot with 3 seconds or less in the shot clock?
If Melo is more likely to have to take “dump off shots” than other players, should this be his fault. I don’t know if it’s true for Melo, just using him as an example. The problem is this requires a lot of logging and is time consuming.
Also, how do you treat end of quarter heaves? Is that an open shot? Is it not counted at all?
Did you discount technical foul free throws? These are not FTs created by the player shooting them and in the case of someone like Durant, I’m sure he shoots almost all of them while Melo probably shoots none (cuz of Billups). Durants points per attempt are being inflated for a non-basketball situation.
And with that thinking comes end of game FTs. Does the player take them or someone else on the team get the bump from intentional fouls. Again, Melo has Billups probably taking most of those FTs.
These things all skew th analysis. I like the work done, but there’s too many holes for me to really draw a conclusion.
arturogalletti
10/28/2010
No worries. I’m going to divvy this up by role as I go along. I’m shooting for a quick and dirty measure for offense cost/benefit. We can use 82 games for particular situations in the future. It’s has great possibilities.
nerdnumbers
10/28/2010
Arturo,
You win an award. I don’t have a name for it, but as you mentioned this line is getting a lot of constructive criticism already. Anyway award for most constructive criticism on unfinished yet awesome article 🙂
EntityAbyss
10/28/2010
I notice when talking about scoring, you and dberri (and others) tend to use points per shot (efg%). Why don’t you just use points per possession shooting (ts%)?
arturogalletti
10/29/2010
EA,
It’s the economics,practical mentality. I don’t really care about shooting percentage by itself. As a gm I would care about getting value (points) from my assets (possessions). I’m looking to measure the profitability of investing possessions in a player. So I start with points per shot add points from assists and subtract turnovers.
FTA are free.
some dude
10/29/2010
you include FT into points in points per possession, but does a shooting foul for 2 or 3 shots equate to a possession in the possession part of the equation?
if not, there’s a piece missing. You’re adding in points for a possession while not counting the possession.
arturogalletti
10/29/2010
SD,
You are right of course. I’ll add in a .44FTA attempt term in the next post (and at the bottom of this one). It’s the SD hack-a-shaq correction!!
Raspu10
10/29/2010
My only problem with this is the assumption that the indirect contribution added via assist can be held constant. Both the true value of an assist and their frequency depend as much on the team as the player. The same is true, of course, for blocks.
So while standardizing the values across teams makes some sense to me, measuring direct contribution makes more sense to me within a team context.
arturogalletti
10/29/2010
Raspu10,
I’m trying to quantify the offensive internal rate of return from an individual player. So it’s fair to hold the value of the assist constant. That said you have a point and I would apply it if looking at it from a team perspective.
Man of Steele
10/29/2010
Arturo, I’m inclined to agree with you in thinking of FTs as the result of good choices made by the player. It seems unavoidable, though, that you may have to adjust this theory in some instances. It is not a smart choice for Shaq to get to the free throw line. In fact, more often than not he misses at least one of two free throws, after which the other teams often regains possession. In effect, it is quite possible that a poor free throw shooter is utilizing resources (possessions) without producing any profit (points).
Of course, this caveat really only applies for small percentage of the population. Most players shoot free throws well enough that it is a good idea to get to the line.
arturogalletti
10/29/2010
MOS,
As noted above with SD, You make a sound economic point and I’m going to correct for this. The SD-MOS hack-a-shaq correction !!!
Ilan
10/29/2010
This may be a silly question, but in your 10-game representative sample Melo shot 43.5%, while over the course of the 82 game 09/10 season, he shot 45.8%. In 07/08 he almost shot .500. Over the past five seasons, he has had four good FG% seasons, indicating that he is a good scorer.
Please can you clarify the discrepancy between your sample and his season-over-season FG% in relation to his efficiency.
Thanks,
arturogalletti
10/29/2010
Ilan,
The sample is not mine rather Jeremy Wagner of Roundball Mining Company. I actually correct for that by increasing the sample to a full three seasons at the end.
Guy
10/29/2010
“BUT Melo is still an Elite Player because he has the SKILL to be amazing (e.g. if he took the shots Durant or Lebron took his offense would be off the charts)
Man I was sold until point four.”
Agreed that having the skill doesn’t make him an elite player. But this breakdown seems to open up an important research question: to what extent is a player’s failure to take as many open shots as other shooters HIS failing, and to what extent is it a TEAM failing (teammates and/or coach)? Isn’t it possible that some players are taking fewer open shots because the rest of the team is failing to facilitate that? That may or may not be the case for Melo, but it seems like an important possibility to explore. (In contrast, it’s less likely that anyone but the player is responsible for his success rate on open shots.)
arturogalletti
10/29/2010
Guy,
That’s the twenty million dollar question. Melo is a particularly frustrating case because in other situations (olympics, pre-season) he has played better. I think young players can be molded to appropriate behavior by the right system but veterans are who they are for the most part. What’s really interesting is that once you’ve identified the right behavior you can start incentivizing it. Ty at Courtside put up a great piece where he talks about Sh*t Shots to quote :
“Now let’s look at the inverse, what I call the “Shit Shot” percentage — the team’s percentage of shots taken from 16 to 23 feet away from the hoop. These are the absolute lowest value shots and should be avoided. Besides being stupid, they are also a sign of (a) weakness — you lack the strength to go to the hoop, or (b) laziness – you are settling for an available “bad” shot instead of working to get a good shot. Shit shots are available all night long. That’s because they don’t return very well on the investment.”
So we want the team to take Shots (1) at the rim; (2) behind the three point arc; and (3) at the foul line and open as often as possible and the better job we do at that the better of we’re going to be. Throw in a rebounding edge and good team defense and you build a champion . If that sounds like the Lakers,Celtics or Spurs? I’m pretty sure it’s not coincidence.