PER vs All Star Votes
Where Pete attempts to create one of those new-fangled graphs that all the kids are doing these days. In theory, it will show the reader how the top 5 vote-getters at each position in both conferences ranked in their PER numbers this season, as well as the four players who rank in the top 25 in PER this season but failed to make the top 5 in votes.
Phew.
PER increases as you move up the chart, while votes received increases as you move to the right. Astute NBA observers should be able to tell who the abbreviations stand for, and I've even color-coded them for your benefit. Obviously, those residing in the bottom right corner aren't the kind of guys you want to see at the All-Star Game. [update - new typeface added and option to enlarge included].
14 Comments:
Nice use of a chart.
That's a fun graph. Nice work.
It is nice to see that the fans, apparently, are not quite as bad as everybody makes them seem. Theoretically, if the PER score was a perfect measure of the players ability, and the fans voted the most for the best players, then all the points would be on a line going from the bottom left to the top right. We do, in fact, see a general trend along that line. So at least the fans got the general trend correct.
So according to the graph, the "perfect rating" line is that diagonal from the bottom left to the top right. Anyone above that line is "underrated" and anyone below the line is "overrated", and the farther from the line the point is, the more over or underrated they are. So the most overrated player is "YI" is that Yi Jinlian? and the most underrated player is "CP", Chris Paul, right?
Something to note is that AI, in the "Overrated" square is only about as overrated as Yao.
As a Detroit fan, it is nice to see that Sheed is still a bit underrated. He is a good basketball player, but his personality gets in the way of his fame.
Good observations, Peter. One other aspect to consider is that a majority of the guys in the lower left are big men, and the only reason they got as many votes as they did is because of the lack of competition at that position. When Andrew Bogut and Samuel Dalembert are in the top 5 and Brandon Roy is not, well, that tells you something about the quality of competition. If I had to do it over again, I might have just used the top 50 or so vote receivers in total, rather than the top 5 at each position. Oh, well.
Zydrunas Ilgauskas would rock this chart. Hardly any votes and 22nd in the league in PER, after spending the first month and a half of the season in the top 10 in PER. Too bad he's had 2 nagging injuries...
A few quick notes from someone who does graphics like this. First, it'd help to use a font that distinguished the Y from the V (is that YI or VI down there on the bottom?). Helvetica would do it, as would about fifty other san serif faces. Kids use charts all the time, yes, but the good kids pay attention to the simple things in info design. Take a look at Edward Tufte for examples that usually inspire even the layperson.
http://en.wikipedia.org/wiki/Edward_Tufte
It'd also help to have a legend. BROV is BROY, which is Brandon Roy. But BROY isn't much clearer. The color coding isn't buying you anything because of the restrictions of the web palette. I came away frustrated despite following the NBA with a microscope for 25 years.
Not harsh crits, just saying that there's a bigger world out there that includes people who make a living from this stuff, and it's actually kinda fun to do it right--and just as easy.
Nuss, you could also make a chart of votes vs. PER difference over replacement player. So for Dwight, his PER number would be 5.26 (= his PER - Zydrunas's PER, i.e. the center in the east with the next highest PER). This would be a way of approximating the lack of competition. This method would only work for the selected starters. For forwards and guards you could look at the difference with the third place vote getter, or maybe an average of the third and fourth. - JohnK
No offense taken; I was frustrated by it as well, but looked at it as a first effort. Thought about including a spreadsheet of the names/votes/PER to add some help, but decided that that was just too much information. Good tip on the typeface, though, that would have helped a bunch.
Oh, and great chart btw. -JohnK
I can't see Tony Parker in this chart... -- CT
Good catch - he was being blocked slightly by Jameer Nelson, but it's fixed now. YOu'll be glad to know that Parker is clearly in the Dangerfield quadrant (unless you're a Suns fan, that is, in which case you're bugged).
Peter M Lambert - you're not reading the graph correctly. The farther to right you move, the move votes you've received. The higher you go, the better the player you are. CP3 has received a lot of votes, and is a highly rated player (same with all the other players in that quadrant). He is not the most under-rated - he's the most under-rated of the top 3 players in PER. The best, most under-rated guy is Brandon Roy. He's 5th in PER (5th most efficient player), yet he only received the ~25th most all-star votes. Dirk was close, but not quite as efficient as Roy, yet he received more votes. Al Jefferson was not quite as efficient, and received even fewer votes.
TMAC, AI, and YI all got votes on their reputations, and thus are over-rated.
If you're great, you're on the top half of the graph. If you're only good, you're on the bottom half of the graph. If you're highly rated, you're at the right, under-rated, on the left.
PER not a great measure of player skill because it's inflated by players on bad teams, or players in select systems. It's like saying which are the best fantasy players. You should try plotting all-star votes against Roland rating from 82games.com. Player effectiveness on team +/- as difference between him and his substitute.
I hear you. The problem is that no matter which measuring stick I use, it's going to fall short in someone's eyes. You're right, PER isn't the best one, but it's the most well known all-around stat of its kind thanks to ESPN.
Post a Comment
