Demar - pre-merge post - looking at Usage% and the impact

[Primer]

111 @ 29.2 is phenomenal. Basically, superstar level. I would hazard a guess that the whole team DRTG in March is skewed by Vasquez starting, but DD's offense is far exceeding his defense.

BTW, people really undervalue the importance of USG%. The way to look at USG is like "carrying your share of the load" on offense. If you have 5 guys on the court, then each guy should be able to handle 20% USG @ 105 ORTG to be considered an average NBA player. But guys like Amir Johnson are offensively limited, so DD @ 28% USG allows 2Pat or Amir to be at 12% USG with super high ORTG efficiency (say 120 ORTG). So, even if DD is an average SG, or even slightly below (say 103 to 105), the tandem nets out at 110+ ORTG. That's really significant.

Or another simple way to illustrate the value of usage is to normalize it. DD @ 29.2 % USG & 111 ORTG, is like 29.2/20 x 111 = 162.1 effective ORTG. It's like adding half a player to the offense.

Conversely, 2Pat is at a phenomenal 122 ORTG, but only 13.1 USG, so he isn't carrying his weight, offensively, so to speak. Normalizing it, you get: 13.1/20 x 122 = 79.9 ORTG. Not so good. He really needs guys like DD and Lou to carry that extra USG load, although 2Pat still has a lot of room to drop in ORTG, while increasing his USG%.

The trick is balancing and optimizing the USG load. Very high ORTG guys like Amir, JV, JJ and even Hansbro could take some of the USG away from DD for a likely net increase in team ORTG. Ross is troubling - a wing player at only 18.1% USG yet still below league average ORTG (104).

You can't normalize ORtg with usage like that. The fact you think PPat would have a 79.9 ORtg (nearly worst in the league) at 20% usage demonstrates that.

A more logical way to normalize ORtg to a 20% usage would be to do this. Using PPat and DD as examples with your numbers from above.

PPat
122 ORtg x (100% + (13.1%-20%)) = 113.6 ORtg at 20% Usage

DD
111 ORtg x (100% + (29.2%-20%)) = 121.2 ORtg at 20% Usage

[mcHAPPY]

Assuming JV is still a piece we're building around, then I think Lowry has to be traded this off-season, for a number of reasons:

1) Age. Lowry's skills will be declining when JV is peaking.

2) Trade Value. Lowry's trade value is at an all-time high and can only decline from here, considering his recent all-star starter status and what is now looking like an tremendous value contract that is locked in for 3 more years.

3) Health. Except for the last 2 seasons, Lowry has been injury prone throughout his career. Now that he's a clear starting PG and shouldering a heavier load, the wear and tear is clearly affecting his defense.

4) Passing. Lowry is not a good PnR passer, or drive & kick specialist. He's a score-first guard that doesn't really get shooters into rhythm, or have the timing/feel to hit big men rolling to the rim. Obviously, not good if we're keeping JV around.

5) Coachability. Lowry has butted heads with every coach he's had before Casey. Why not Casey? IMO, it's because Casey has given Lowry Peyton Manning-like authority to call his own audibles on the fly. So, I think Lowry has a big hand in the "ugly offense" we've been seeing. Problem is, if Casey gets fired and another coach comes in and tries to install a more egalitarian ball-movement offense, maybe Lowry reverts to his rebelling ways again. In fact, we did see this when Casey tried to get Lowry to play more under control like Calderon and it hurt his game.

Q: What could we get for Lowry in a trade, and who would want to acquire him?

New York and LA would be nice but I can't see them trading away their top pick for Lowry. On the other hand, considering what it took to get a guy like Kevin Love, maybe a Lowry would be a good value for the pick. Normally I'd say no way the other team does it but when you're talking LA and NY, patience is hardly a virtue historically.

The other team I'd throw out there is Houston. They have a lot of desirable trade assets, a weakness at PG, and an interest in Lowry last year. Hopefully the Pelicans miss the playoffs in this scenario but that would give them the #14 pick (or likely #18 otherwise), Knicks 2015 2nd round pick (#31), Montiejunas, Jones, Capela, and ability to sign and trade Beverly and McDaniels. NO pick, one of Montiejunas/Jones/Capela, and one of Beverly/McDaniels would be a great haul.

Raps are 6-6 in games without Lowry this season which would still put them in first in the Atlantic assuming it was stretched over the season.


I do believe the Raptors need to do something and trading one of Lowry or DeRozan is a good start to add talent through prospects and picks as well as shed significant salary for this upcoming off season and possibly next year.

[Jrice9]

I have my doubts that the guys you listed for Houston other than Capela (who we havn't seen enough of yet) can be the type of star you want.

Most of them are young and rather inexperienced which gives them values in terms of having the potential to be making leaps but its not like these are guys who are screaming stardom. Montiejunas and Jones are interesting but its year 4 for both of them next year and neither have ever been even close to a primary option (or even to be honest the #2) option in the offense. These are both guys with under 19 usage rates.

I'm good with taking a gamble on these types of guys but not at the expense of Lowry or DeMar for those guys I want someone with more pedigree. (A top 5 pick in this draft would be much more interesting)

[JawsGT]

Loads of talent on this team, which I believe would be clearly evident if we actually had a decent system. Shit, we don't even PnR at all anymore from what I've seen recently. Lots of screens get set, some get utilized but almost never for a PnR. Big sets screen, guard uses it to get closer to rim, jacks shot or occasionally passes back out to other guard/wing for shot. The type of game we play exposes the weaknesses of the players moreso than it exemplifies the strengths, and that's the biggest problem, outside of the terrible D.

What I think this team needs the most, outside of a coach, is a really good wing defender that can shoot the 3, and a PG that can play D as well, and maybe another big with some length. We have a wing that can shoot the 3, we have one that can defend, but not one that can do both. Amir's health is the only thing that concerns me about PF, but he may never get back into top form so that may need to be addressed as well.

Address those concerns, get a new coach that can implement a system, and then re-evaluate and get rid of the low IQ players, whomever they may be. Right now, our system is so simplistic and crap, that I wonder if it is even possible to determine if any of these guys could actually learn and execute a more complicated scheme?

[golden]

You can't normalize ORtg with usage like that. The fact you think PPat would have a 79.9 ORtg (nearly worst in the league) at 20% usage demonstrates that.

A more logical way to normalize ORtg to a 20% usage would be to do this. Using PPat and DD as examples with your numbers from above.

PPat
122 ORtg x (100% + (13.1%-20%)) = 113.6 ORtg at 20% Usage

DD
111 ORtg x (100% + (29.2%-20%)) = 121.2 ORtg at 20% Usage

Appreciate the feedback. At first glance I thought the way you’re suggesting is the correct way to normalize. So, I checked both methods with an extreme example and now I’m back to thinking the original way (multiplying) is the right way.

Here’s the example:

In 100 possessions:

Player A (a.k.a. Kobe Lou Anthony) takes 96 shots & makes 48 of them. All contested turnaround long 2’s

Players B, C, D & E (actual names that Kobe calls them) each take 1 shot apiece and make that shot. All layups.

No FTs were taken and no OREBs were had and no passes were ever made. Straight up Casey end of quarter ISO.

So Team ORTG = points per 100 possessions = (48+1+1+1+1) x 2 = 104

Player A uses 96 possessions out of 100. USG = 96/100 = 96%
Players B,C,D & E use 1 possession out of 100. USG = 1% each

Player A ORTG = 96pts/96 possessions = 100 pts per 100 poss = 100
Player B-E ORTG = 2 pts/1 possession = 200 pts per 100 poss. = 200

Using your suggested formula (subtracting 20%):

Player A (impact) = 100 x (100 + 96-20)% = 100 x 176% = 176 effective ORTG
Players B-E (impact) = 200 x (100+ 1-20)% = 200 x 81% = 162 effective ORTG

All 5 players ORTGs normalized to a 20% USG are higher than the team ORTG, so something is wrong here. You should be able to add them up and divide by 5 to get the team ORTG.

Using the previous formula (dividing by 20):

Player A (impact) = 100 x 96/20 = 480 effective ORTG
Players B-E (impact) = 200 x 1/20 = 10 effective ORTG

Add them up (480+10+10+10+10) & divide by 5 and you get the team ORTG (104).

This does seems to make more sense relative to their actual contributions. Player A is truly carrying the offensive load of almost 5 players (4.8 exactly), so it is reflected in his effective (impact) ORTG.

I could be missing something here, but my conclusion would be that USG really does make a big difference in how you should view a player’s offensive “impact”. IMO, it’s probably one of the first stats you need to look at when evaluating or critiquing a player’s offense.

[Primer]

Appreciate the feedback. At first glance I thought the way you’re suggesting is the correct way to normalize. So, I checked both methods with an extreme example and now I’m back to thinking the original way (multiplying) is the right way.

Here’s the example:

In 100 possessions:

Player A (a.k.a. Kobe Lou Anthony) takes 96 shots & makes 48 of them. All contested turnaround long 2’s

Players B, C, D & E (actual names that Kobe calls them) each take 1 shot apiece and make that shot. All layups.

No FTs were taken and no OREBs were had and no passes were ever made. Straight up Casey end of quarter ISO.

So Team ORTG = points per 100 possessions = (48+1+1+1+1) x 2 = 104

Player A uses 96 possessions out of 100. USG = 96/100 = 96%
Players B,C,D & E use 1 possession out of 100. USG = 1% each

Player A ORTG = 96pts/96 possessions = 100 pts per 100 poss = 100
Player B-E ORTG = 2 pts/1 possession = 200 pts per 100 poss. = 200

Using your suggested formula (subtracting 20%):

Player A (impact) = 100 x (100 + 96-20)% = 100 x 176% = 176 effective ORTG
Players B-E (impact) = 200 x (100+ 1-20)% = 200 x 81% = 162 effective ORTG

All 5 players ORTGs normalized to a 20% USG are higher than the team ORTG, so something is wrong here. You should be able to add them up and divide by 5 to get the team ORTG.

Using the previous formula (dividing by 20):

Player A (impact) = 100 x 96/20 = 480 effective ORTG
Players B-E (impact) = 200 x 1/20 = 10 effective ORTG

Add them up (480+10+10+10+10) & divide by 5 and you get the team ORTG (104).

This does seems to make more sense relative to their actual contributions. Player A is truly carrying the offensive load of almost 5 players (4.8 exactly), so it is reflected in his effective (impact) ORTG.

I could be missing something here, but my conclusion would be that USG really does make a big difference in how you should view a player’s offensive “impact”. IMO, it’s probably one of the first stats you need to look at when evaluating or critiquing a player’s offense.

For those reading who don't know, Usage is the Percentage of team plays used by that player while he's on the floor.

For comparison, I did the whole team with both methods. Your method is Norm2. Just going by the smell test, I'm not sure if I trust your method. For instance, Bruno is an average NBA offensive player at 20% usage? Doesn't seem right. Hansbrough and Patterson would be some of the worst offensive players in the league at 20% usage, and much much worse than Bruno? Again, doesn't seem right. Anyways, here are the numbers so people can make of them what they will.

[Axel]

Congrats gents. Not too often that I feel woefully undereducated, but you guys have taken this thread to a whole new level.

Really interesting stuff.

[Axel]

Looking at that chart, I'd have to throw my support behind Norm1.

[enlightenment]

For those reading who don't know, Usage is the Percentage of team plays used by that player while he's on the floor.

For comparison, I did the whole team with both methods. Your method is Norm2. Just going by the smell test, I'm not sure if I trust your method. For instance, Bruno is an average NBA offensive player at 20% usage? Doesn't seem right. Hansbrough and Patterson would be some of the worst offensive players in the league at 20% usage, and much much worse than Bruno? Again, doesn't seem right. Anyways, here are the numbers so people can make of them what they will.

Norm2 for Lou.... Just give Lou the ball on every possession please.

[Axel]

Norm2 for Lou.... Just give Lou the ball on every possession please.

Don't let Casey see this stuff!! Otherwise every post game presser will start with "that's a shot he normally makes"

[mcHAPPY]

Don't let Casey see this stuff!! Otherwise every post game presser will start with "that's a shot he normally makes"

I am fairly certain that is how every post game presser starts now.

[Axel]

I am fairly certain that is how every post game presser starts now.

Sadly true.

[planetmars]

It's difficult to take stats and normalize them for guys that don't get many minutes (like Bruno). Hence the term "small sample size". Amir, 2Pat, Hans, and JJ are your typical role players and role players usage numbers not normalizing properly makes sense to me too.. the numbers are really low but these are your typical bangers or spot up shooter (in 2Pat's case). Problem with 2Pat is he doesn't take many free throws, and his FGA is pretty low on the team as well, so his usage will be really low. If these type of guys got more shots and carried the offensive load I could see their numbers crumbling.

I like Norm2 personally for that reason. Plus it appears to be more mathematically correct.

[Primer]

It's difficult to take stats and normalize them for guys that don't get many minutes (like Bruno). Hence the term "small sample size". Amir, 2Pat, Hans, and JJ are your typical role players and role players usage numbers not normalizing properly makes sense to me too.. the numbers are really low but these are your typical bangers or spot up shooter (in 2Pat's case). Problem with 2Pat is he doesn't take many free throws, and his FGA is pretty low on the team as well, so his usage will be really low. If these type of guys got more shots and carried the offensive load I could see their numbers crumbling.

I like Norm2 personally for that reason. Plus it appears to be more mathematically correct.

Usage isn't terribly well defined in my opinion, so you have some wiggle room when attempting to use it to normalize a rather well defined stat like ORtg. Another issue is that ORtg is inherently normalized, it's how much each player scores over 100 possessions, so every player is already on an even playing field. So we're using an estimate (Usage) to try to normalize a stat that is already normalized, because we know from watching basketball that guys who are part of the majority of the plays will have a tougher time scoring than guys that only get a few plays under the basket. We inherently know that you can't double the shot attempts of a guy with 60% FG and expect him to keep 60% FG, so I like using a few different methods to attempt to normalize ORtg and see which passes the smell test.