Thanks. From that data, I can use standing reach. The vertical seems a little unreliable. That's something that can change dramatically from a player's pre-draft measurements to what he becomes as a pro. Chuck Hayes, for instance, was measured as having a 30 inch no-step vertical.
Have to agree here. After a somewhat promising start as a Rocket, Hill has proven he may just not get it defensively. I still like when he's active on the boards and blocking a few shots. But, man... he cannot move his feet to cut off or attempt to stop dribble penetration. He really is slow at reacting and that alone has been enough for me to want to see Patterson even more. There is so much to be hopeful for with Patterson. His IQ seems to be extremely good. His help defense seems to be excellent. You have to love his mentality. If Morey pulls off something big before the deadline, I hope it doesn't come at the expense of losing this guy or Chuck. What's frustrating is reading Feigen's answers sometimes on his blog regarding certain defensive centers and how they wouldn't "fit" the system. The SYSTEM isn't winning now there HAS to be a way to sacrifice or adjust the system offensively to incorporate someone like Dalembert defensively. I realize Chuck and Dalembert (or a defensive center) on the floor at the same time would compromise the offense, but man... I miss the days of having SOME kind of faith we can get a crucial stop when needed.
Good suggestion. I added some more factors into the formula: Reach: Standing reach from draftexpress (inches). Where not available, I used 1.32*Ht. SpF: What I'm calling "specialist factor". It equals 0 for players with USG%<0, and it is MPG*(20-USG%)/100 otherwise. So, players who get a lot of minutes for their team despite not having a high usage on the offensive end having a high "specialist factor". I'd argue that there's a high correlation between this factor and their defensive importance. There are rare exceptions where players are low-usage offensive specialists, but they are less likely to get a lot of minutes. Somewhat more common are important offensive players who are also very strong on defense (like Dwight Howard or Tim Duncan). For these players, the SpF is 0. PF/m: Simple fouls committed per 36 minutes. Includes offensive fouls unfortunately, but the vast majority fouls are committed on the defensive end. Adding these factors to the mix, the regression gave me this formula: DEF +/- ~= DSPM = -5.61 + 0.04*Reach + 0.52*SpF - 0.11*(PF/min) + 0.15*STL% + 0.06*DRB% + 0.08*BLK% Here's a table for current centers this season showing all the factors used above and how DSPM approximates their defensive impact. I added Age and MPG to the table as well which I think are relevant factors but did not incorporate into the formula. Where Reach is estimated based on height, I marked it with an asterisk. Code: [SIZE=1][B][U]Player Tm Age mpg Reach SpF PF/m STL% DRB% BLK% DAPM DSPM DASPM[/U][/B] Tyson Chandler DAL 28 28.0 110.0 1.8 4.3 0.9 26.4 3.3 +1.7 +1.4 +1.6 Andrew Bogut MIL 26 35.8 110.5 0.0 3.3 1 27 6.3 +2.1 +0.9 +1.5 Tim Duncan SAS 34 29.3 109.5* 0.0 2.0 1.2 26.5 4.8 +1.9 +0.9 +1.4 Kurt Thomas CHI 38 23.9 106.9* 2.3 5.0 1.6 20.2 3.3 +1.2 +1.2 +1.2 Marcus Camby POR 36 28.7 109.5* 2.2 3.1 1.6 35.4 5.3 -0.2 +2.6 +1.2 DeAndre Jordan LAC 22 25.5 113.5 2.2 4.6 1.2 21.4 5.3 +0.6 +1.6 +1.1 Joel Anthony MIA 28 20.3 106.9* 3.0 4.7 0.4 12.4 5.9 +0.9 +1.1 +1.0 Emeka Okafor NOH 28 32.7 110.5 1.4 3.6 0.6 25.3 4.5 +0.7 +1.3 +1.0 Dwight Howard ORL 25 36.3 111.5 0.0 3.4 1.8 30.2 4.8 +0.8 +1.1 +1.0 Marc Gasol MEM 26 33.2 112.2 1.2 3.6 1.5 18.9 4.3 +0.9 +1.0 +0.9 Anderson Varejao CLE 28 32.1 108.2* 2.1 3.2 1.4 24.8 3 +0.2 +1.6 +0.9 Omer Asik CHI 24 11.6 110.8* 0.9 6.0 1.2 20.8 4.9 +1.2 +0.6 +0.9 Chris Bosh MIA 26 35.5 109.0 0.0 2.3 1.1 20 1.6 +1.6 +0.1 +0.9 Andris Biedrins GSW 24 26.4 109.5* 2.3 5.2 1.6 25 2.6 +0.1 +1.5 +0.8 Andrew Bynum LAL 23 24.5 110.8* 0.1 4.0 0.2 21.4 5.5 +1.1 +0.4 +0.7 Anthony Tolliver MIN 25 18.2 107.5 1.1 4.1 1.2 17.5 2.1 +1.1 +0.3 +0.7 Samuel Dalembert SAC 29 20.2 109.5* 1.0 4.4 0.7 24 6 +0.4 +1.0 +0.7 Pau Gasol LAL 30 37.3 110.8* 0.0 2.3 0.9 20.4 3.8 +1.0 +0.4 +0.7 Tyrus Thomas CHA 24 21.6 108.0 0.0 4.5 2 21.5 6.2 +0.9 +0.5 +0.7 Jason Smith NOH 24 15.4 106.0 0.4 4.6 1.2 17.6 2.5 +1.4 -0.1 +0.6 Chuck Hayes HOU 27 24.0 104.5 1.8 3.8 1.9 19.5 1.9 +0.4 +0.9 +0.6 Roy Hibbert IND 24 27.0 113.5* 0.0 3.9 0.5 22.3 4.9 +0.8 +0.5 +0.6 JaVale McGee WAS 23 27.3 114.5 1.3 3.7 1.1 21.4 7.1 -0.2 +1.4 +0.6 Zydrunas Ilgauskas MIA 35 17.3 114.8* 0.9 5.4 1.1 19.4 4.4 +0.4 +0.7 +0.5 Joakim Noah CHI 25 36.8 106.5 0.8 3.1 1.7 23.2 3.4 +0.2 +0.8 +0.5 Taj Gibson CHI 25 22.6 109.0 0.5 4.3 1.2 18 5.6 +0.6 +0.4 +0.5 Antonio McDyess SAS 36 17.5 106.9* 0.9 4.0 1.6 22.8 2.1 +0.3 +0.7 +0.5 Glen Davis BOS 25 30.2 106.9* 0.0 3.4 1.6 15.8 1.1 +1.2 -0.3 +0.5 Greg Monroe DET 20 23.3 108.5 1.3 3.5 2.3 19.2 1.5 +0.1 +0.8 +0.4 Brendan Haywood DAL 31 17.9 113.5 1.5 4.4 0.5 19.6 4.1 -0.1 +0.9 +0.4 Derrick Favors NJN 19 19.3 110.0 0.7 5.9 0.8 18.1 2.8 +0.7 +0.1 +0.4 Al Horford ATL 24 34.9 107.0 0.0 2.7 1.2 24.2 2.4 +0.3 +0.4 +0.3 Shelden Williams DEN 27 17.7 104.0 0.9 4.6 1.3 23.3 2.4 +0.2 +0.5 +0.3 Ben Wallace DET 36 23.5 106.9* 2.7 2.4 2.5 23 3.6 -1.4 +2.0 +0.3 Kwame Brown CHA 28 22.0 108.0 1.0 4.2 0.6 23.2 2.1 +0.6 +0.3 Shaquille O'Neal BOS 38 21.1 113.0 0.0 5.6 1.2 19.9 4.3 +0.4 +0.2 +0.3 Amir Johnson TOR 23 25.3 106.9* 1.1 5.4 1.4 18.4 3.1 +0.2 +0.4 +0.3 Nene Hilario DEN 28 30.4 109.0 0.5 3.8 1.7 21 1.9 +0.1 +0.4 +0.3 Marcin Gortat TOT 26 19.5 110.0 1.1 3.9 0.9 25.5 3.6 -0.7 +1.0 +0.2 Ed Davis TOR 21 21.5 108.0 1.7 4.3 0.9 21.4 3.8 -0.8 +1.0 +0.1 Channing Frye PHO 27 31.8 110.5 1.2 4.3 1.1 19.2 2.4 -0.6 +0.6 +0.0 Spencer Hawes PHI 22 19.7 110.0 0.2 4.4 0.7 23.3 2.9 -0.3 +0.3 +0.0 Zaza Pachulia ATL 26 14.2 109.5* 0.7 5.9 1.3 19.4 1.5 -0.1 +0.1 -0.0 Brook Lopez NJN 22 34.1 113.0 0.0 3.4 1 11.9 3.3 +0.1 -0.2 -0.1 DeMarcus Cousins SAC 20 26.3 113.0 0.0 5.5 1.6 23.3 2.1 -0.5 +0.3 -0.1 Brad Miller HOU 34 18.8 109.5* 0.5 3.8 1.6 18.4 1.6 -0.5 +0.3 -0.1 DeJuan Blair SAS 21 21.1 106.5 0.1 4.4 2.9 22.7 1.5 -0.7 +0.3 -0.2 Nenad Krstic OKC 27 22.5 110.8* 0.8 3.5 0.9 14.1 1.3 -0.5 +0.1 -0.2 Semih Erden BOS 24 14.8 110.8* 0.9 6.3 1.3 17.6 3.1 -0.7 +0.2 -0.2 Darko Milicic MIN 25 24.7 111.5 0.0 4.8 1.4 15.8 7.2 -0.7 +0.2 -0.3 Nazr Mohammed CHA 33 16.8 108.2* 0.0 4.8 1 23 5 -0.9 +0.3 -0.3 Al Jefferson UTA 26 34.8 110.0 0.0 3.1 0.8 22.2 4.6 -1.2 +0.4 -0.4 Amare Stoudemire NYK 28 37.1 108.5 0.0 3.6 1.3 18.6 4.3 -1.1 +0.1 -0.5 Robin Lopez PHO 22 17.5 110.8* 0.1 4.8 0.6 13.7 2.9 -0.7 -0.4 -0.5 Shawne Williams NYK 24 18.1 104.0 0.8 5.1 1.2 15.1 2.6 -1.2 -0.2 -0.7 Jordan Hill HOU 23 18.0 108.0 0.4 4.8 0.4 16.7 3.1 -1.7 -0.2 -0.9[/SIZE] One player turned out to be at least 1 standard deviation ahead of the average for both DAPM and DSPM rating: Tyson Chandler. How different might our season have been if that trade the Rockets were reportedly close to finalizing went through? Relaxing to players who are half a standard deviation above average by both ratings, the players are Tyson Chandler, Emeka Okafor, Kurt Thomas, Dwight Howard, Joel Anthony, and Marc Gasol. A player like Dwight Howard is probably underrated here, because he's a dominant defensive force despite being a relatively high usage player on the offensive end. That combination is fairly rare these days. Three players ended up at least 1 standard deviation below the rest by both DSPM and DAPM: Shawne Williams (who probably should not have been included anyway, he hardly ever plays center), Robin Williams (surprise, though the Suns are a really bad defensive team), and Jordan Hill (not a surprise).
To correlate, are you basically adding each coefficient*variable, and checking how it compares with the Defensive +/-? And to improve the correlation (I'm guessing R^2), you're adjusting the coefficients of these variables?
I'm using the regression function which comes in the Data Analysis package in Excel. I believe its the simplest kind, an OLS linear regression. For how it works, you can read here: http://en.wikipedia.org/wiki/Linear_regression_model
Things I would like to see if analyzing defensive stats: - FG% of opposing player - Defensive Rebounding per defensive rebounding opertunity - Blocked shots per opponents offensive possession - Steals per opponents offensive possession - Number of balls tipped / Forced Opponent into a bad pass etc... per opponents offensive possession - Number of opponents offensive shots altered per opponents offensive possession(if at all possible) - Number of charges taken per opponents offensive possession Other things that are just much more difficult to measure: - being a disruptive force i.e. pushing the opposing offensive player out of the post, redirecting opponents out of good offensive position either with or with out the ball, denying the opponents the ball etc... - all around hustle - Is a player a good help defender vs. man defender or both? - Does he guard post players or perimeter oriented players better or worse or both good? - How often does he need help on his defensive assignments? - How well does he cover his teammates defensive mistakes? - How many defensive positions on the court can he guard effectively? - How far out on the court can he alter shots while still being able to get into back into the post to prevent his own man from having an uncontested shot? - How good is he at going after the block while still getting back into position for the defensive rebound? - Consistency? So many things you have to see in order to realize how effective the defender is. Dream could fill up the defensive stats sheet but so could so many other players like David Robinson. Looking at the stats you would think that they were similar defensive players but in fact if you watched both players you would know that Hakeem was by far the superior defensive player Example: I used to hate it when Edwardo Najera would guard Yao Ming. His effectiveness did not always show up on the stats sheet but he was such a pest that it completely effected Yao Ming to the point that it just bogged down the Rockets offense. Some things you just have to see with your eyes.
Nice thoughts there. Suppose the NBA wanted every team's stats keepers to come up with a "defensive boxscore". They would allow 8 stats to be tracked, and they asked you to design it. In addition to providing useful information for analysis, the other criteria would be that it is relatively easy for statskeepers to track in real time. What would this defensive boxscore look like? Currently, the following defensive stats are tracked: - defensive rebounds - steals - blocks - personal fouls (most of which are defensive)
Charges taken vs blocking fouls There are things that are difficult for boxscore keepers but not necessarily hard to track, just on top of my head at the moment: Shooting efficiency of the opposing player guarded Shot attempts of the opposing player guarded Missed rotation Double teaming
Thanks for the attempt, but I really couldn't buy anything else in the analysis after seeing that Shaq was ranked above Hayes.
If by "buy" you mean treat the results like gospel where everything is right on, then by all means don't buy any of the ratings. They are merely estimates based on a few measurables, and they will miss (perhaps considerably) on some players. The goal of these sort of statistical ratings is just to get it mostly right. If its no better than just randomly ordering the players, then that is of course literally useless. If it tends to order based on defensive impact (which you can more or less judge by a quick "eyeball" test), than there is some use I'd argue. Some people will disagree, I guess.
What if the stats agree with the naked eye. Does that make you more confident in your assessment? And what if the stats disagree sharply with the naked eye. Does that make you somewhat less confident in your assessment? This is the point that sometimes people forget. Stats aren't meant to replace "naked eye" evaluations, but rather increase (or, if appropriate, decrease) our confidence in them.
Somebody's got to put a Sticky/FAQ about the difference between raw +/- and ADJUSTED +/-. Hint: "Adjusted" may be key to the distinction.
BUT WHY WOULD YOU WANT TO.....!!! ..........SOUNDS LIKE SOMEBODY HAS ALOT OF TIME ON THEIR HANDS......IM JUST SAYING
Here's how I would rate centers defensively: "Well, you know, I'm not saying I'm an expert, and you'll probably flame me, and there might be better ways to do this and things I forgot to take into account, but IMHO..."
One would think that Tyson Chandler must be having a significant impact on the Dallas defense, based on these stats. Is that the case compared to the last few seasons of the Mavs? It helps that he's healthy this year. He wasn't the last two seasons, which might have influenced Morey's decisions. We look at his play and see a major plus for the Rocks were he on the team. Did he have a similar (projected) impact with Dallas?
Because a defensive center is a big need for my team right now, and I wanted a relatively simple formula that would help me identify strong defensive centers. Something more than blks+rebs.
Kings big Samuel Dalembert has allowed 76 pts on 100 post-ups. Kings big DeMarcus Cousins has allowed 76 pts on 100 post-ups. (via Synergy)
