eWins leaders 2016-17

[Mike G]

eWins projected to the full season, at 15.6% of the season.

Code: Select all

e82     per36 rates    tm   Min   Eff%   Sco   Reb   Ast   Stl   TO   Blk   e484
18.7 Russell Westbrook OkC   34  .555   35.3  10.5  10.1   1.2   5.5   .4   3.21
17.9   James Harden    Hou   37  .603   29.8   8.3  12.4   1.3   5.1   .2   2.87
15.6   Anthony Davis   NOP   38  .575   31.2  10.1   1.8   1.9   2.4  2.7   2.61
15.2   Kevin Durant    GSW   35  .665   30.3   8.0   4.9   1.9   2.4  1.3   2.57
14.6  DeMarcus Cousins Sac   34  .561   30.5  10.9   3.6   1.3   3.1  1.1   2.55

14.4   Chris Paul      LAC   31  .602   24.8   6.3  11.9   3.6   2.4   .3   2.79
14.2   DeMar DeRozan   Tor   37  .565   33.9   4.8   3.6   1.6   2.8   .0   2.28
14.1   Blake Griffin   LAC   33  .552   26.9  10.3   5.5   1.4   1.9   .6   2.53
13.4   Kemba Walker    Cha   33  .594   30.4   4.4   6.3   2.0   2.0   .1   2.38
13.4   LeBron James    Cle   36  .570   24.9   8.2   9.8   1.0   3.6   .5   2.38

e82     per36 rates    tm   Min   Eff%   Sco   Reb   Ast   Stl   TO   Blk   e484
13.4   Kawhi Leonard   SAS   34  .588   30.1   7.0   3.3   2.6   2.2   .5   2.33
12.7   Jimmy Butler    Chi   35  .594   28.1   7.0   4.1   1.9   2.3   .0   2.15
12.7  Hassan Whiteside Mia   33  .541   21.6  16.9    .9   1.1   2.7  3.0   2.25
12.2 Gia Antetokounmpo Mil   35  .547   23.7   8.6   5.4   1.8   3.4  2.2   2.08
12.0   Stephen Curry   GSW   33  .632   29.4   3.8   6.9   1.4   3.4   .3   2.12

11.7   Isaiah Thomas   Bos   33  .575   29.9   3.0   6.8    .9   2.9   .2   2.08
11.1  Karl-Anth. Towns Min   35  .573   23.8   9.6   2.6    .8   2.2  1.3   1.87
10.9   Damian Lillard  Por   36  .594   28.8   4.4   4.1    .8   3.1   .1   1.79
10.8   Kyrie Irving    Cle   34  .555   27.3   3.5   5.9    .9   2.4   .3   1.88
10.6   Kevin Love      Cle   32  .544   24.0  12.2   1.6    .9   2.5   .7   1.95

e82     per36 rates    tm   Min   Eff%   Sco   Reb   Ast   Stl   TO   Blk   e484
9.9   Dwight Howard    Atl   29  .570   20.3  16.8   1.9   2.0   3.0  2.2   2.22
9.5   Andrew Wiggins   Min   36  .568   27.1   4.9   2.4    .4   3.0   .6   1.55
9.3   Paul Millsap     Atl   33  .521   20.6   9.1   4.4   1.8   3.1   .5   1.67
9.2   John Wall        Was   33  .554   26.9   4.5   8.3   1.9   5.0   .6   1.97
9.1   Mike Conley      Mem   32  .593   23.8   4.3   6.7   1.5   2.9   .8   1.80

9.0   Kyle Lowry       Tor   39  .501   17.7   4.8   7.2   1.7   2.8   .5   1.37
9.0   Andre Drummond   Det   30  .515   19.5  15.7   1.4   1.4   2.7  1.3   1.91
9.0   Draymond Green   GSW   34  .507   11.2   9.4   7.9   2.4   2.1  1.8   1.57
8.9 Kristaps Porzingis NYK   33  .594   23.2   7.6   1.3    .9   1.5  1.3   1.61
8.8   Carmelo Anthony  NYK   33  .557   24.7   6.1   2.8    .9   2.2   .3   1.59

e82     per36 rates    tm   Min   Eff%   Sco   Reb   Ast   Stl   TO   Blk   e484
8.3   Jabari Parker    Mil   33  .529   22.2   7.0   2.4   1.2   2.0   .4   1.48
8.2   Dwyane Wade      Chi   30  .521   23.4   5.3   3.7   1.9   2.5   .8   1.60
8.2   Marc Gasol       Mem   33  .496   21.4   7.3   4.1    .8   2.4  1.5   1.60
8.1   Avery Bradley    Bos   36  .550   18.5   8.0   3.3   1.0   2.0   .4   1.33
8.1   Rudy Gay         Sac   34  .554   22.1   6.7   2.7   1.4   2.3   .8   1.50

8.0  LaMarcus Aldridge SAS   33  .548   22.5   8.1   1.6    .9   1.9  1.2   1.57
8.0   Julius Randle    LAL   28  .585   18.0  11.6   5.1   1.1   3.6   .8   1.66
7.9   Harrison Barnes  Dal   38  .531   22.5   5.5    .9    .6   1.4   .2   1.24
7.7   C.J. McCollum    Por   35  .571   22.8   3.9   3.3   1.4   2.6   .6   1.32
7.7   Nicolas Batum    Cha   34  .537   16.5   6.7   5.6   1.4   2.3   .5   1.33

e82     per36 rates    tm   Min   Eff%   Sco   Reb   Ast   Stl   TO   Blk   e484
7.7   Paul George      Ind   35  .570   22.6   6.9   3.5   1.9   3.3   .6   1.54
7.6   Rudy Gobert      Uta   31  .633   14.4  13.2   1.4    .9   2.1  2.7   1.45
7.6   Pau Gasol        SAS   26  .533   18.4  10.4   4.1    .4   1.5  2.0   1.75
7.5   Zach Randolph    Mem   22  .528   24.8  13.6   1.6    .7   2.5   .0   2.02
7.4   Myles Turner     Ind   28  .598   20.6   8.7   1.2   1.2   2.2  3.2   1.58

7.4   Otto Porter      Was   35  .577   15.8   8.7   1.5   1.7    .8   .8   1.23
7.3   Marcin Gortat    Was   35  .597   12.6  13.5   1.6    .4   1.7  1.3   1.24
7.2   Nikola Vucevic   Orl   27  .424   14.6  14.0   3.3    .7   1.6   .7   1.57
7.2   Rodney Hood      Uta   33  .574   22.6   5.2   2.3    .9   1.4   .3   1.39
7.1   DeAndre Jordan   LAC   31  .560   14.0  14.5    .8    .6   2.1  1.9   1.38

[bchaikin]

harden and westbrook are on pace for some 420 turnovers each for the season, westbrook is at some 15.2 zero point team possessions per game - not good...

chris paul is currently generating wins at a higher rate (per min) than westbrook - although he is scoring just 18 pts/g (to westbrook's 32 pts/g), he is far more efficient on offense (2.46 to 2.02 pts/0ptposs), and is on pace for 246 steals with just 164 turnovers. westbrook is on pace for 95 steals but 423 turnovers...

anthony davis is on pace for some just 200 turnovers for the season (yet some 31-32 pts/g), durant just 190 (yet 28 pts/g), and durant has offensive efficiency of 3.26 pts/0ptposs (dantley's best over a season was 2.95, bird 2.85) - extremely good...

[bchaikin]

right on cue...

westbrook tonight in thunder 115-111 OT lose to indiana had a triple double - 31 pts, 11 reb, 15 ast...

but he missed 21 shots (shot 13/34), 16 were rebounded defensively by okc, and had 7 turnovers...

that's 23 zero point team possessions by westbrook himself - in just one game...

[Mike G]

He also assisted on 73% of his teammates' FG while he was on the floor?
http://www.basketball-reference.com/box ... 00OKC.html
If his 15 assists each receive 1/2 of the scoring credit, he may have created (31+15) 46 points on his 36 'used' possessions, or 1.28 per.
The team had 111 points on 109 poss, or 1.02 per poss.

On the season, the Thunder have ORtg of 108.5 with Westy on the floor, 90.5 without him.
Defensively they're 9.5 pts/100 better with him, so a total (O+D) of +27.5.

Team eFG% is .508 with him running things vs .441 without.
Their rebounding% margin is +6.2% with and -6.2% without -- one in 8 boards they don't get.
Turnover margin worsens by 7 per 100 when he sits.
http://www.basketball-reference.com/tea ... 17/on-off/

EDIT: Harden, too, seems to have a profound effect on his team's turnover rate: 15.5 when he's on the floor, 24.3 when he's off.
Can anyone explain why the Rockets then have a team TO% of 14.3 ?

[bchaikin]

If his 15 assists each receive 1/2 of the scoring credit

do you then take away 1/2 of the scoring credit from the players who scored?...

[Mike G]

You could, if you wanted a non-Westbrook subtotal for the team.
It may also be true that a couple of his FG were assisted. But his assisted% is at a career low -- just 11% of his 2FG and 39% of his 3's, so around 20% of all FG.

The difference in this game's OKC scoring efficiency with/without Westbrook is similar to that for the whole season. They don't seem to have either many shot creators or distributors on the team, after RW.

He went 40 min. last night, several beyond his usual mpg. I'll guess that his late game shooting wasn't as good as usual.

[bchaikin]

You could, if you wanted a non-Westbrook subtotal for the team.

let me rephrase - do you then take away 1/2 of the scoring credit from the players who scored?...

They don't seem to have either many shot creators or distributors

or? is there a difference between a shot creator and a distributor?...

[Mike G]

I estimate a player's unassisted scoring, based on other stats. For Westbrook, I estimate 76% of his points are unassisted. For Andre Roberson, it's guessed to be 20%

For calculation purposes, call them .76 and .20 . These are plugged into the term:
(unAst%)^e*k
... where e and k are exponent and constant, and this a factor in the Scoring term.
These vary with the sample, especially variable in a given playoff series. The upshot is that shot creators get more credit for their scoring than those who rely on passes to score.

is there a difference between a shot creator and a distributor?...

One is a player who creates shots for himself, without needing to be fed the ball in favorite spots; the other sets up teammate scoring attempts via passing.
Some players are both: Westbrook, Paul, LeBron, Curry ...

[bchaikin]

shot creators get more credit for their scoring than those who rely on passes to score.

so in your rating system if two players have identical raw stats (min, fgm/fga, ftm/fta, reb, ast, st, to, bs), but one gets assisted on just 30% of his FGM and the other on 70% of his FGM, you rate the former higher?...

One is a player who creates shots for himself, without needing to be fed the ball in favorite spots; the other sets up teammate scoring attempts via passing.

are there teams with neither? or do you simply say players on a team of low assists has shot creators and a team with high assists has distributors?...

in 1988-89 in the 25 team league charlotte was 20-62, new jersey was 26-56, two of the worst five teams in the league that season. they were also two of the five worst teams in offensive efficiency, a similar 102.3 pts/100poss and 101.9 pts/100poss scored...

but where charlotte lead the league in ast/100poss (27.7) and ast/fgm (.678), new jersey was dead last in ast/100poss (21.5) and ast/fgm (.538). either way, that's 26%-29% more assists but similar shooting (47.0% vs. 47.4% eFG%) and similar offensive efficiency...

what kind of players were on those two teams?...

[Crow]

Any system that gives credit for assists without subtracting that amount (whatever it is) from the basket's value (2 or 3 pts) to yield the scorer's credit would be over-crediting.

[Mike G]

I don't worry about double-counting stats, because it all gets normalized in the end.
Turnovers may also be steals or fouls, FG may be assisted or not, FT are never counted as assisted, etc. There's no practical way to ferret out all the double-counts.

Then there's the issue of home vs away assists (and steals and blocks). Using away ast/FG rates, I dispense with home court bias, and that goes a long way toward rectifying things.

When I pull up the Finals of 2016 -- then arbitrarily add 6 Ast/G to Kevin Love's (and the Cavs') totals -- all Cavs but Love see their Scoring column diminished a bit. The formulae assume more of their points were assisted, etc. The net eWins per team is unchanged.

so in your rating system if two players have identical raw stats (min, fgm/fga, ftm/fta, reb, ast, st, to, bs), but one gets assisted on just 30% of his FGM and the other on 70% of his FGM, you rate the former higher?..

Yes.
If I poke Russell Westbrook's uAst% down to .30 (from .76), his Sco column drops from 35.3 to 32.8 (per 36 min), and his eW/484 from 3.21 to 2.99

[bchaikin]

Then there's the issue of home vs away assists (and steals and blocks)

whether an assist is awarded on a FGM is superfluous, the basket is still good...

as for steals and blocks, a steal is awarded when there is a change in team possession, so in that change of team possession whether the steal is awarded or not it's still a good defensive play (remember a steal is also awarded when a defender merely picks up a loose ball), and as far as i know a block is never awarded on a FGM, so whether a block is assigned or not on a missed FGA it's still a good defensive play. i think Vantage has data showing how much lower an opponent's FG% is when a defender has his hand up...

If I poke Russell Westbrook's uAst% down to .30 (from .76), his Sco column drops from 35.3 to 32.8 (per 36 min)

i thought his scoring looked high...

as for my example above of the 88-89 hornets versus nets, where charlotte had far more assists than did new jersey, if you look at the player stats you'll notice that on both teams the PGs (muggsy bogues/michael holton versus lester conner/john bagley) had a similar number of total assists (1044 vs. 995. 12.7 vs. 12.1 ast/g) but the difference in the assists of each team's other players was quite large (1279 vs. 798, 15.6 vs. 9.7 ast/g)...

other than the PGs the players to play the most minutes on the hornets were SF kelly tripucka, PF kurt rambis, SG rex chapman, and SF/SG robert reid, and on the nets PF/C roy hinson, PF buck williams, SF chris morris, and SG/SF mike mcgee. but just because the nets had far fewer assists would not have me come to the conclusion that these four nets players were better "shot creators" than those four hornets players...

btw - did you know that last season in 2015-16, according to the SportVu data, the team that threw the most passes (utah, 354.8 passes/g) threw for the fewest number of assists (19.0 ast/g):

http://stats.nba.com/tracking/#!/team/p ... MADE&dir=1

and the team that threw the least number of passes (okc, 241.8 passes/g) was 10th in ast/g (with 23.0)...

so far this 2016-17 season:

http://stats.nba.com/tracking/#!/team/p ... AST&dir=-1

the team throwing the fewest passes (hou, 271.4 passes/g) is throwing for more assists (24.2 ast/g) than the 10 teams who throw the most passes per game (311-339 passes/g), and are throwing for the 4th most ast/g...

[Mike G]

nets PF/C roy hinson, PF buck williams, SF chris morris, and SG/SF mike mcgee. but just because the nets had far fewer assists would not have me come to the conclusion that these four nets players were better "shot creators"

I don't think I made up the term, but I do believe that's the definition of 'shot creator' -- a player who creates his own scoring opps, not relying on others to get him the ball in a favorite location.

Among the top 100 in minutes this season, the top and bottom 10 estimated unassisted% scorers:

Code: Select all

.77   James Harden     PG   HOU
.76   Dennis Schroder  PG   ATL
.76  Russell Westbrook PG   OKC
.71   Chris Paul       PG   LAC
.71   Sergio Rodriguez PG   PHI
.71   Tim Frazier      PG   NOP
.70   Jeff Teague      PG   IND
.69   Stephen Curry    PG   GSW
.69   Isaiah Thomas    PG   BOS
.68   LeBron James     SF   CLE
  . . .        
.21   Harrison Barnes  SF   DAL
.20   Trevor Ariza     SF   HOU
.20   Serge Ibaka      PF   ORL
.20   Andre Roberson   SF   OKC
.19  Patrick Patterson PF   TOR
.19   Otto Porter      SF   WAS
.16   Robin Lopez       C   CHI
.15   Maurice Harkless SF   POR
.11   Arron Afflalo    SG   SAC
.10   Ryan Anderson    PF   HOU

Some of these guys at the bottom are high efficiency shooters; but a lineup of 5 such players would see their shooting% drop without some passers on the floor. Those at the top would be largely unaffected.

i thought his scoring looked high...

Westbrook's raw 33.2 pts/36 is bumped up by his above-avg .555 effective shooting%; by his above avg (.76) competition vs starters; his above-avg (.76) unassisted%; and he's penalized a bit by his team's above-avg (102.1) pts/48 allowed.

The unassisted% factor applies to all points, not just FG. That is, I assume FT are assisted as much as FG, even though no assist is awarded.

the team throwing the fewest passes (hou, 271.4 passes/g) is throwing for more assists (24.2 ast/g) than the 10 teams who throw the most passes per game (311-339 passes/g), and are throwing for the 4th most ast/g...

I don't know what to make of this.

[bchaikin]

Some of these guys at the bottom are high efficiency shooters; but a lineup of 5 such players would see their shooting% drop without some passers on the floor.

and just what evidence might you have showing this statement to be true?...

i see evidence to the contrary - some of the very worst shooting teams of the past 4 decades had high rates of assists per FGM, like the 98-99 chicago bulls that shot a 42.4% eFG% - the worst team eFG% since 1977-78 - with a .661 ast/fgm ratio, and the 02-03 denver nuggets that shot a 42.8% eFG% (3rd worst since 77-78) with a 64.6% ast/fgm ratio...

plus the 47-19 thunder of 2011-12 shot a 51.6% eFG% but with just a .497 ast/fgm ratio (9th lowest since 77-78), and the 48-34 suns of 2013-14 shot an even better 51.9% eFG% with an even lower .493 ast/fgm ratio (7th lowest since 77-78)...

I don't know what to make of this.

there is no relation between the number of passes thrown and the assists a team gets, i.e. you don't get more assists by passing the ball more...

[Mike G]

Hmm, I'd suppose teams that have great shot creators don't need as many great passes to score; and they may be better offenses than those who tend to score when they've combined a good pass with an effective finish.

And if ...

there is no relation between the number of passes thrown and the assists a team gets, i.e. you don't get more assists by passing the ball more...

... maybe an effective pass is one that isn't just tossed around the perimeter, but rather a dribbler has broken down the defense and dishes to an open shooter/dunker?

Fewer passes may also correlate to fewer turnovers.

Those low-unassisted% guys without their point guards might get more assists by themselves and still shoot a lower %. Which kind of corroborates your inverse relation argument.

the 98-99 chicago bulls that shot a 42.4% eFG% - the worst team eFG% since 1977-78 - with a .661 ast/fgm ratio

Those Bulls in home games are recorded to have assisted .743 of their FG; on the road, however, the other 28 scorekeepers in the league assigned Ast on just .580 of their FG -- quite an average ratio. That's the number I'd use, and adjust each player's Ast rate (however measured) by a factor of (580/743) .781 .
This alleges that 28% of their home assists would not have been assists on the road, and shouldn't be counted as such.

the 48-34 suns of 2013-14 shot an even better 51.9% eFG% with an even lower .493 ast/fgm ratio (7th lowest since 77-78)...

The Suns franchise has for years been among the stingiest at awarding assists, for their own guys and for opponents. This includes the Steve Nash years.
In 2014, they got assists on .516 of FG on the road, vs only .471 at home. That .516 is still quite low, but certainly not 7th lowest in 38 seasons.

Recently, glaring home assist bias has gone subsided quite a bit. Last season's most egregious were the Cavs, with 1.14 home/away Ast/FG -- that's just half of the bias shown in Chicago'99.
League avg was 1.032, with 9 teams <1.00