APBRmetrics

[Mike G]

[kjb]

I understand the limitation of Bob's simulation -- if it is indeed a limitation. I think Bob is right to avoid instituting a "penalty" to efficiency as usage increases absent evidence of an average effect. We can all go back through the record and find examples of guys who increase or decrease efficiency as usage changes, but what's the "normal" effect of a usage change? And how can we predict which guys will maintain (or increase) efficiency as they increase usage and which guys will drop? For example, if we could use statistical data to reliably predict that Hoiberg cannot maintain efficiency at 13 shots per 36 minutes, but that Brent Barry can -- that'd be very valuable information.

It may be that ultimately this is something that exists, but cannot be proven using stats because of how the game is played -- i.e. a guy plays more because he's playing better, a guy shoots less because he's shooting worse. But, I think we could all go through the record and identify players who didn't get playing time despite stats that demonstrate they should be playing, or guys who continue shooting despite consistently poor results.

I personally like Dean's philosophy -- accept the conventional wisdom as valid until we can demonstrate that it's wrong. Most of the coaches I've talked with seem to accept that efficiency tends to decline as usage increases. I haven't discussed these issues specifically -- I'm making an inference from other comments.

One area where I disagree with Bob is that coaches can't influence usage. I think that would surprise a lot of coaches. With some players, that's surely true -- they're going to do what they're going to do regardless. But, well, I'd like to see some evidence to support that claim.

[bchaikin]

One area where I disagree with Bob is that coaches can't influence usage.

it's not that they can't influence it, it's just that they do not define it for each player - touches/min are a function of a player's ability. for example if a coach switches a player from full time SG to full time PG (reggie theus, dennis johnson, sleepy floyd, allen iverson), the vast majority of the time that player's touches/min will increase, and increase significantly. by that coaching decision he has indeed influenced a player's usage. but not all players can be switched to PG and increase their touches/min significantly (jerome james, mark blount) and i doubt very much an SG is switched to PG with the thought that it would increase or decrease the other players touches/min by specific amounts...

or by a coach making the conscious decision to replace a low touches/min starter with a high touches/min starter (or vice versa), that too will affect the touches/min of the other starters, so in that way he has influenced usage...

but the specific example used where i stated touches/min is not defined by coaches was why can't you replace amare stoudemire with steven hunter and expect to have hunter score like stoudemire or handle the ball as often as him. hunter has for 4 seasons demonstrated a consistent touches/min of around 0.50, stoudemire has gone from 0.7 to 0.9 to 1.0 in 3 seasons. touches/min is a function of a player's ability or desire to get open, to get the ball, to do something with the ball (shoot, pass, get fouled, or turn it over) and for whatever reasons stoudamire has shown the ability to increase his while hunter has not...

am i saying hunter will never increase his touches/min? no. but historically speaking most players who have demonstrated a low touches/min like 0.50 for 4 straight seasons have not increased that significantly later on in their careers. if his new frontcourt teammates also have low touches/min, odds are his will increase. if not, odds are it will stay close to the same...

[bchaikin]

And ignoring your simulator for a moment, do you really truly believe that Minnesota would only win 3 less games with Reggie Evans playing Garnett's minutes if 24 minutes of Sprewell were replaced by Hoiberg?

do you really truly believe your player defensive ratings that were posted at 82games.com? do you really truly believe you understand the nba salary cap and all of its ramifications?....

[Dan Rosenbaum]

[NickS]

[Ben F.]

Here's a question on the flipside of the argument - rather than a Fred Hoiberg leading a team, what about a star being forced to a smaller role?

I heard today that Nick Van Exel is going to the Spurs. Now Van Exel has never been a particularly efficient player, but apparently they're planning on bringing him in to play the role Beno Udrih played during the playoffs last year (essentially a backup guard to spell Tony Parker). They also like his three point shooting. "We wanted him in our program," Buford said. "We think he'll be a great complement to the team we have" (Source).

Now, the question - will Van Exel increase his efficiency in a reduced role? If he's allowed to pick his spots, allowed to feed off the defense keying on Manu and Duncan, will he be more efficient?

[kbche]

Hi,

The term efficiency originally applied to an engine. By definition engine efficiency(%) is the actual/theoretical*100. Engine efficiency will increase with usage until a plateau is reached, then efficiency will decrease. This same phenomenon can be applied to basketball. What coaches need to assess is the plateau for each player. This plateau may be based on the position played, the type of shot taken, physical condition, opposing opponents, etc.

KAB

[kjb]

[bchaikin]

this is not a simulation but just one example from real life of how players touches/min can increase significantly (more than one on the same team) and maintain efficiency not so much because the players have improved on their own but because of the acquisition of a new teammate or teammates whose combination of touches/min and how often they shoot and pass end up modifying the touches/min of their other teammates....

here are the denver nuggets of 95-96 and 96-97 (i only listed players that played 500+ minutes):

95-96 (W-L 35-47)
min - player
2713 dikembe mutombo (0.59)
2280 antonio mcdyess (0.72)
2626 dale ellis (0.75)
2810 bryant stith (0.96)
2029 abdul-rauf (1.63)
1107 don maclean (1.15)
1045 tom hammond (0.52)
1269 laphonso ellis (0.82)
817 reggie williams (0.90)
2134 jalen rose (1.76)
607 doug overton (1.37)

96-97 (W-L 21-61)
min - player
2599 ervin johnson (0.43)
2565 antonio mcdyess (0.89)
2002 laphonso ellis (1.06)
1788 bryant stith (0.93)
2001 mark jackson (2.17)
1758 tom hammond (0.57)
2940 dale ellis (0.84)
1047 brooks thompson (1.47)
654 kenny smith (1.35)
612 anthony goldwire (1.55)
600 ricky pierce (0.92)

and here are the differences in their touches/min for a few of them:

l.ellis (0.82-1.06) +23%
mcdyess (0.72-0.89) +19%
d.ellis (0.75-0.84) +11%
stith (0.96-0.93) -3%

mutombo was replaced at C by ervin johnson and johnson got less touches/min, so that meant a few more touches for other players. the combination of abdul-rauf (chris jackson) and jalen rose at PG being replaced with mark jackson and other low minute PGs was significant because despite rauf and rose both having high touches/min jackson passed ball more per touch than did either of those two players. as a matter of fact jackson passed the ball per touch probably as much as any PG ever (as much as stockton anyway)...

jackson left the team before the next season. laphonso ellis never handled the ball as much as he did this season with jackson either earlier or later in his career. dale ellis had handled the ball more often earlier in his career, but his touches/min with jackson was his highest in 5 seasons (he was 35 years old in 96-97). 96-97 was just mcdyess' 2nd season in the league, and although his touches/min did decrease the next season without jackson it did increase later in his career to greater than what it was with jackson....

in this case the touches/min of players on the same team did not increase because of the acquisition of another player with low touches/min, but because of the acquisition of a player who passed the ball with a very high proportion of his touches...

[HoopStudies]

[NickS]

[bchaikin]

If this inverse relationship holds, if it is significant at 95%, will anyone care?

it would be significant information even if it wasn't significant at 95%...

what i had tried to do years ago but never got very far on was to compare players per game averages for a season to their season bests (most) in a single game for things like MIN, FGA, REBs, AST, ST, TO, PTS, etc, to see if any relationship existed between what was "average" for players and what was maximum for players. i could then come up with "...well the average player takes only 33% more FGA in the single game where they do take their most FGA versus their avg game, 40% more REB in their max game versus their avg game..." etc. i also wanted do to this for their career per game averages and career bests in certain stats in a single game...

the problem was when doing this years ago i couldn't get season bests for all players nor career bests for all players in a single game. but i thought this would have been a place to start. what i had found at one time was that the maximum number of FGA a player took in a single game in a season was on average just 33% more than their average FGA/g for that season (without regard to the variance). however i don't know if this still holds, and i never finished doing this for other stat categories...

[HoopStudies]

[NickS]