After the first few weeks of the season, I posted some stats I called "net possession created" (see here ). I tweaked the formula here after some input from you. I liked the name "blue collar index", which someone suggested, a little more, so I'm going with that here. The concept remains the same. I want to use the information in the boxscore to capture how a player gets his teams extra "possessions" to work with, while at the same time limiting the opposing team's scoring opportunities. If you immediately think this will be biased towards "unskilled" players who just do the dirty work around the basket, you're right! The method could be further improved by considering charges, as some have pointed out, but that information isn't publically available as far as I know (sorry Luis). So, here's the final formula I use: index = 40/Min * [ 0.7*orb + 0.3*drb + stl - tov ] It's pretty simple, but I think it highlights a particular type of contribution that often goes unnoticed when people look at the box score statistics. Below is the top 20 in the league by this index, with at least 250 minutes played, through 12/28. I also list their On/Off +/- which would reflect how much better their teams play with them on the court versus when they're off the court (an important limitation being that it does not adjust for quality of teammates or opponents faced). If this index I concocted is totally worthless with no connection to winning whatsoever, one would expect the On/Off numbers to be all over the map. In fact, you'll see that players with a high blue collar index rating tend to also have a good On/Off rating as well. I think that suggests there is something to this. Code: [B]rank Player Team Min On/Off index[/B] 1 thomas,kurt sea 474 +11.9 6.50 2 foster,jeff ind 701 +3.6 6.37 3 camby,marcus den 964 +12.2 6.05 [I]4 hayes,chuck hou 690 +4.2 5.85[/I] 5 wallace,ben chi 790 -4.2 5.64 6 oberto,fabricio san 624 -1.4 5.16 7 marion,shawn pho 1108 +8.9 5.15 8 varejao,anderson cle 270 +11.5 4.99 9 biedrins,andris gsw 853 +3.7 4.98 10 noah,joakim chi 260 +9.4 4.89 11 davis,glen bos 270 +2.7 4.84 12 millsap,paul uta 666 +3.6 4.82 13 collison,nick sea 708 +0.4 4.76 14 boone,josh njn 417 -0.8 4.71 15 hollins,ryan cha 259 +6.6 4.63 16 lee,david nyk 733 -3.0 4.63 17 balkman,renaldo nyk 272 +9.0 4.57 18 dampier,erick dal 529 +1.6 4.52 19 mohammed,nazr na 429 +3.2 4.50 20 okafor,emeka cha 954 -3.2 4.49 Here's the index numbers for the Rockets thus far (250 minutes cutoff): Code: [B]rank Player Min On/Off index[/B] 1 hayes,chuck 690 +4.2 5.85 2 scola,luis 586 +1.9 3.67 3 wells,bonzi 680 -1.5 3.45 4 battier,shane 1046 +2.9 2.21 5 ming,yao 1118 -5.5 1.63 6 james,mike 489 -6.6 0.72 7 alston,rafer 842 +7.6 0.48 8 head,luther 439 -0.4 0.41 9 mcgrady,tracy 880 -1.6 0.41
He rebounds, and he gets steals, so he'll do well by this index. He also commits too many turnovers and is (on average) an inefficient scorer.
Interesting stats. How about charges a player takes or how many offensive fouls a player creates against opponents?
Even though his shot didn't fall at all, Bonzi cut hard and rebounded. He's an energy guy. He really helped team recently.
Our two best starters (Yao and T-Mc) show negative numbers and the two worst (Hayes and Battier) show positive numbers? No. These stats show they are meaningless.
Very interesting, especially T-Glass. Thank you. My only cavil here is that people get bumped too much for limited minutes (e.g. Balkman, Hollins, Davis). I understand the scaling by 40/min, but it is much easier to play blue collar in 15 minutes per game than 40. This is probably too much math/effort, but a more gentle scaling factor for playing time would be good, I think. Maybe (40/min)^(0.5) or something like that. So if you play 12 minutes per game, your stats are multiplied by 4, but rather by 2. You could even do (40/min)^(2/3)... anything less than 1, ideally. Anyway, bottom line is more cool work from durvusa!
The fact that you are surprised show that you do not understand these stats. Of course they are going to show negative numbers. Tmac and Yao are our offensive forces. They aren't going to spend as much time and effort doing OTHER things. These stats aren't useless. You are just too stupid to understand them. Thats all.
LOL great post. Also, you have to take this stat with a grain of salt because it favors big men. It is no coincidence that the bottom 4 players on the Rockets list are ball handlers.
Well if you look at the formula, its a function of Rebounds + Steals - Turnovers. The OP shouldnt have ranked our perimeter players in the index because perimeter players arent in the post as much to wipe up boards and are more prone to turnovers because they handle the ball.
The index isn't a measure of our best players. It's measure of our most "blue collar" players. So, I think it's quite fair to say our bigs are more "blue collar" than our ball-handlers. They doesn't mean they are better overall. It's putting focus on a particular type of contribution, indicative of players who are often considered "unskilled", but which I think actually brings real value. And if it wasn't clear in the first post, I am not saying that the On/Off rating is a measure of true value or anything like that. By it's nature, it is a very noisy statistic. So, a player can have a poor On/Off rating but still be very good, and vice versa. But ... here's the important point -- the better you are, the more likely you'll have a strong On/Off rating. Yao and McGrady's On/Off is negative so far this year. That's surprising, and uncharacteristic for them considering the past 2 or 3 years. It indicates that we haven't played very well as a team, and that they've both played big minutes for us against the opposing team's strongest units. But, if you look at the best players in the league (say top 20), the large majority of them have very strong On/Off numbers. And vice versa for the bottom 20. And that's just for a couple months of play. So there's a difference between a statistic being noisy, and a statistic being "meaningless". What I wanted to show by also posting the On/Off numbers was a pattern -- if you have a high blue collar index, you also tend to have good On/Off numbers. And if you go through the list of those players, they tend to be guys who are underappreciated by fans, but highly valued by their coaches. I thought it was interesting.
That's an interesting idea. I think what you're looking to capture is what a player could potentially give you if he played full starter's minutes (say, 35 mpg). That's an important question, but I was looking at something a little different -- what players are currently giving their team while they are on the court. Yes, it could favor players who are playing less (how much, I don't know), but it's more descriptive of what's currently happening. Your approach might be better if we were looking to acquire a player to play big minutes in our rotation, and we wanted to project their productivity.
I don't know if the stats are available, but another interesting thing to factor in would be points scored off of offensive rebounds -- tip-ins, putbacks, etc. A big indicator of blue-collar BB work.
Interesting. But your +/- seems to be different from 82games.com's. For example, Tmac had a much worse +/- on that site.
That's strange. I copied the numbers from 82games earlier today, but now it looks to have changed. Perhaps they were old. The Rockets have gotten two good wins without Tracy in the past two games, so that's would explain his On/Off decreasing.
The 82games.com +/- are not kept all that current all the time. This morning, I saw it was updated as of 12/27/07, so that explains the difference. Also the one on the team page and the one on each player's page are different... so I don't know which is the correct one. In any case, this is still early enough in the season that a few good or bad games would cause a guy's +/- to change significantly.
