There are defensive statistics out there (defensive rating, defensive win shares, etc) - remind me again why WIS is reinventing the wheel? And at least by using an already available statistic, WIS could just point any criticism toward whoever came up with that stat. Seems win-win.

How far back do those fancy modern defensive statistics go?  Do they cover players from the '60s and '50s?

dws pre 73 where it exists at all is highly fudged and I havent seen anyone generate pre78 DRTG- bball ref has their own pre73 dws but who knows what evil lurks in the heart of wis admin when it comes to this kind of decision making

I think they'd prefer to under estimate Bill Russell's ability to block shots and then base his D rating on the low number of blocks they granted him in the first place thus generating a poorly fudged advanced number from poorly thought out base numbers they themselves generated because that makes more sense to them 

Numbers from that early are going to be fudged no matter what, so why not go with the work someone else already did instead of doing it all over again? Also, any stat that has Russell lead the league by this wide a margin (1959) will get widespread approval around these parts:

Defensive Win Shares

1.	Bill Russell*-BOS	8.2
2.	Bob Cousy*-BOS	5.4
3.	Tom Heinsohn*-BOS	4.1
4.	Tom Gola*-PHW	3.6
5.	Dolph Schayes*-SYR	3.6

I think they'd prefer to under estimate Bill Russell's ability to block shots and then base his D rating on the low number of blocks they granted him in the first place thus generating a poorly fudged advanced number from poorly thought out base numbers they themselves generated because that makes more sense to them

Post of the thread.

Posted by tarheel1991 on 5/2/2011 12:12:00 PM (view original):
seems to me like defense has basically no impact on salary...most of the guys who lost significant portions of their DEF rating seem to have lost like a couple hundred thousand at most in salary

Seems pretty accurate for how much it actually does sim-wise.  I confess I was disappointed to see Sir Charles' d go down into the 40s & 50s yet his salary only go down a couple hundred thousand, but it's actually probably a good thing considering how little defense does.

at best what do we really think defense does in the sim? it's some modifying weight on a shot attempt - this guy is covered by this other guy when he shoots so that means that you have to deduct 0.0whatever% off his chance to make the shot

so when we talk about defensive rating what we are talking about in the end is whether 0.0whatever% is appropriately significant in determining shot success/failure - what if 90+ is a  3% modifier? is that enough? if Joe Blow is a 50% shooter should Bruce Bowen turn him into a 47% shooter? and if so how do we expect that to express itself in the boxscore of any particular game? given that even without the modifier there are going to be times when Joe Blow shoots 40% (and others where he shoots 60% or 70% or more because wis loves its outliers) what the heck are we going to know from looking at a particular boxscore? zip

defense is a popcorn fart in the hurricane of random noise that's built into the sim engine in lieu of a rationalized and systematic correcting approach to the game - everything has been stripped down to randomness - you think a 5% modification on a chance event that has a myriad of other positive and negative modifying factors and occurs 80+ times in an individual game is going to pop out at you?

good luck with that

Well said Felonius. 

People complain about defense having too small an effect in the sim, but I've never really seen anyone do any sort of actual study on it on a large sample - just the purely anecdotal evidence from box scores like you're referring to, felonius. I agree that it does seem like people expect a good defender to hold his man to 35% shooting EVERY game (when in actuality, even the best defenders give up 60% opponent FG% every now and then). I'd love for WIS to just explicitly tell us what effect defense has, since they seem fairly open about how most of the other stats factor into the game.

defense has an effect, and for anybody who has spent any time analyzing it, they know this is true.  The issue used to be that it wasn't worth the cost.  It was a very small effect only seen over a long period of time.  Now it's a very small part of the formula and might actually be pretty accurate at this point.  Now the main problems are that minutes are too expensive, usage is too expensive, and rebounds are too cheap (therefore low usage, low minute board hoarders [(c) - felonius 2011] are extremely efficient to buy while average rebounding players with medium/high usage and normal starters' minutes are too expensive.

All defensive teams are out, please update the defensive ratings accordingly:

http://www.nba.com/2011/news/05/09/2010-11-all-defensive-teams/?ls=iref:nbahpt2

Sorry to bump this but I just saw it and I hope I can be helpful regarding the defense ratings. Going back and adding made-up all-defense teams is really just a band-aid to mask ratings that could be better. Effectively, the biggest stat bump comes from being named to a team, and that's not necessarily realistic either. The defensive stat is currently used in the sim as a modifier of the opponent's shooting %. So, since we're basically trying to model a player's contributions to team and individual defense, by far the most important stats in the equation should be opponent's team FG% and, if available, FG% of that player's man (particularly in the years prior to zone being allowed). Steals, blocks, and even rebounds should still be part of the D rating equation, but not a very big part. Steals probably mean that a player is a good on the ball defender, but it might just mean he's an undisciplined gambler. Blocks probably mean he's a good help side defender, but coming from the weak side is a whole lot different than having your man back you down in the paint. Besides, both these stats - like rebounding - are accounted for directly elsewhere in the simulation engine. A good rebounder might just be someone who crashes the boards without discipline or regard (*coughTROYMURPHY*) but is probably someone who has a lot of effort, physicality, and attention to detail - all important characteristics for a defender. To me, fouls are a function of the star power of the player he's guarding relative to his own, but I can see it being a part of the D equation. Probably the smallest part. The bulk of it should be opponent's FG%, individual and team - because that's what we're trying to simulate in the first place. Do it that way, and the results might be a bit surprising in cases, but the sim will probably be much more realistic.

obviously you don't know seble like we do, highlander.  We were trying to give him the easy way out because he won't do any real work to make the sim any better.  He only takes short cuts.

It's funny how the front page of WIS advertises the ability to simulate the 2011 NBA Finals, but then uses the full season rosters for each team instead of current rosters (and some sort of realistic minutes distribution instead of holding guys to their real minutes divided by 82). It's a general disregard of actual basketball and sticking to their own in-house rules for simulating, which are the kind of things that generally hold them back.