2011-12 Rookies

[Mike G]

In case of Beasley for example, you would expect him to be even less effecient, if he would be coming off the bench....

I apologize for imperfect communication, but at first glance, this looks backwards.
Coming off the bench, it's easier to get rebounds, steals, blocks, and shot attempts.
This is due to being on the floor with less skilled players on both teams.

Williams and Beasley are not far apart in what I call 'productivity' -- probably not the same as what you call 'efficient'. Player efficiency does not exist in a vacuum. There are many examples of high-volume, low-efficiency players who have had big roles for successful teams. More important than the efficiency of the possessions they've "used" are their effects on the whole team's offense.

I see Beasley had another bad shooting game last night (though he was the best rebounder in the game), and Williams got almost as many minutes (and hit 4-5 FT).

Can you back that estimation up with any real data?

Well, having refined these measures for a few years, I have found that my 'standardized' player Sco and Reb rates have significantly smaller year to year variations than are found in b-r.com's TRB%, Pts/36, etc.
And yes, eWins/48 have annual variation about half that of WS/48, also less than PER.

This is not a claim that the magnitude of my starter/sub adjustment is precise or fair in every case. Some players really do better as starters, and vice versa.

But if the alternative is to assume that a players' stats are the same vs starters and vs bench players -- that the best-fitting conversion rate is 1.00 -- this stretches credibility, IMO, and misses a great opportunity in player analysis.

Games Started are known since 1981-82.

[Mike G]

With season 1/8 played, rookies project to 56 eWins.

Code: Select all

e/66   per36 rates     tm   Min   Eff%   Sco   Reb   Ast    PF   Stl   TO   Blk   e484
7.1   Irving,Kyrie    Cle   27   .517   23.2   5.8   7.4   3.7    .9   4.2  1.1   1.93
5.3   Rubio,Ricky     Min   28   .564   13.4   4.9  10.5   2.6   1.4   3.9   .3   1.39
4.0   Morris,Markieff Phx   22   .626   16.0   9.2   2.1   4.5    .6    .9  1.4   1.30
3.5   Brooks,Marshon  NJN   24   .536   20.8   5.0    .9   2.4   1.6   1.5   .3   1.21
3.4   Cole,Norris     Mia   23   .500   16.9   2.5   6.3   3.5   2.2   3.7   .0   1.06

2.6   Walker,Kemba    Cha   20   .471   15.9   4.2   3.8   1.3   1.1   1.9   .2    .98
2.5   Knight,Brandon  Det   27   .481   15.2   4.5   3.3   2.4   1.1   3.6   .0    .68
2.5  Thompson,Tristan Cle   18   .500   15.3   8.6    .5   4.2    .2   2.5  2.9    .99
2.4  Williams,Derrick Min   21   .561   15.1   7.5    .9   2.8   1.2   2.0   .4    .86
2.3   Shumpert,Iman   NYK   29   .515   15.6   4.4   5.0   3.3   2.4   2.4   .5   1.16

2.2   Vucevic,Nikola  Phi   12   .650   17.6  11.6   3.8   4.9   2.2   1.6  1.5   1.92
2.1   Leonard,Kawhi   SAS   19   .478   10.3   9.3   1.2   1.5   1.7    .8   .2    .81
1.9   Leuer,Jon       Mil   17   .543   12.9   8.0   1.3   3.4   1.0   1.4  1.1    .84
1.9   Kanter,Enes     Uta   14   .481   11.0  12.8    .6   2.3    .6   1.7  1.1    .99
1.5   Stiemsma,Greg   Bos   15   .647    9.6   9.9   1.9   7.3    .4   1.5  5.2   1.03

e/66   per36 rates     tm   Min   Eff%   Sco   Reb   Ast    PF   Stl   TO   Blk   e484
1.3   Burks,Alec      Uta   12   .467   16.2   4.2   2.9   1.8   1.1   1.8   .3    .91
1.3  Parsons,Chandler Hou   18   .481   10.9  10.6   1.3   4.3   1.5   2.8   .5    .71
1.1   Pargo,Jeremy    Mem   20   .412   10.7   3.0   5.0   2.7    .9   2.5   .0    .43
1.1   Harris,Terrel   Mia   15   .540   12.6  10.0   3.3   3.0   1.3   3.0   .0    .98
1.0   Harrellson,Josh NYK   19   .421    7.0   8.7    .8   1.9   1.2   1.4  1.0    .40

0.9   Thomas,Isaiah   Sac   10   .468   12.7   4.6   2.5    .7   1.8   2.5   .8    .68
0.8   Butler,Jimmy    Chi    5   .889   31.3  10.3    .0    .0    .0    .0  1.3   2.67
0.8   Fredette,Jimmer Sac   24   .450   11.3   2.2   3.2   1.6    .6   2.3   .0    .23
0.5   Faried,Kenneth  Den   10   .500   12.1  14.1    .0   1.7    .0   3.3  7.4   1.58
0.4   Selby,Josh      Mem   15   .391   10.1   2.3   5.4   1.1    .6   2.8   .0    .36

Shumpert leads all rookies at 29.2 mpg, but he's played just 4 of 8 games. This projection assumes he averages 14.6 min over 66 games.
But if he doesn't miss any or many more games, he'll be vying for 2nd or 3rd in eWins.

[Mike G]

Interestingly, with Beasley out for Minnesota, he's replaced in the starting lineup with the weakest player on the team, a guard, Wayne Ellington -- who then puts in a couple of nice outings.
In 2 games, his season Eff% shot up from .360 to .419, and his eW/484 from .02 to .04 -- from nearly nothing to twice that.

Meanwhile, Derrick Williams has stepped up from 13.8 to 15.6 in Sco/36, from 7.7 to 8.1 in Reb, from 2.7 to 2.1 in TO.
And from .65 to .98 in e484 -- leading to a projected 2.9 eWin season, which is much closer to the 4.3 eW a #2 pick ordinarily gets. (He'd projected to 1.6 a few days ago.)
Now looking like the #7 rookie.

[EvanZ]

It's interesting to see how usage gets allocated when a high usage guy misses games.

On a related note, I'm experimenting with adding a "usage credit" to ezPM. Since nobody cares that much, I haven't optimized it yet. I'm starting with the value that Eli Witus found here:

http://www.countthebasket.com/blog/2008 ... fficiency/

To refresh memories for those who may have read it a long time ago (or are reading it for the first time), he found that (roughly speaking) for every 1% increase in "predicted" lineup usage (obviously usage always sums to 100% in a game), there is a 0.25 increase in ORTG. So that's the value I'm incorporating, and I use 20% as the baseline.

To give an example. Say Player A has a 32% USG. The way I'm doing it now, that will increase his ezPM100 rating by 3 points:

0.25*(32-20) = 3

This moves guys like Irving and Kemba way up the list, while Leuer, Leonard, and Markieff move down a little.

http://thecity2.com/roy-watch/

[Mike G]

In a 20-point win over Was, the Wolves shot 52% eFG, and they had no use for a 42% shooter.
In a 10-pt loss to Tor, they shot 38% and could definitely have used a 42% shooter (who could also rebound).

[mystic]

In a 20-point win over Was, the Wolves shot 52% eFG, and they had no use for a 42% shooter.
In a 10-pt loss to Tor, they shot 38% and could definitely have used a 42% shooter (who could also rebound).

Was is the worst team in the league and I had Toronto being better than Minnesota before that game while Toronto played at home. And Beasley has better stats against lottery team than he has against playoff teams (which isn't surprising, it is the same for a lot of players).

[Mike G]

It's probably true that the vast majority of players do worse against playoff teams; and in fact most have 'worse' stats in the playoffs than they have had in regular seasons.
Of course, if you adjust for the level of competition, then equal fractions of players do better/worse against stronger/weaker opponents.

Just as certain highly competitive players will put forth their best efforts against a strong challenge, it may be that coaches feel they can discern when a rookie may be overwhelmed by the NBA's best, and he may be withheld from a starting assignment. Once you're named as a starter, you can't be withdrawn vs elite teams; that would be an insult.
Increasing player confidence is one job of the head coach, and the flip side is shielding young players from competitive humiliation.

[EvanZ]

Do you think rookies are placed against weaker competition to "shield" them? I think it's mainly because coaches use their worst players only when they can get away with it. If they could help it, coaches wouldn't play their replacement-level talent (or rookies) against good opponents, because it doesn't help them win. Obviously there are exceptions and constraints at work, too. I've heard that coaches "protect" young players, but I'm not convinced it's protecting anything other than the coach's job. I would think the most rapid development of players would come when giving them quality opponents. When I play tennis, I only improve playing against better players, not ones that are worse than I am.

[mystic]

So, now I have a Rookie Watch on my blog.

http://bbmetrics.wordpress.com/rookie-watch/

It is sorted by PW, which just means Player Wins. PW gives the amount of wins back a player would generate with his SPM value and his played minutes. PW48 is simple PW devided by minutes and multiplied by 48.

Rubio is my frontrunner for ROY.

Mike, is it possible to list an eWins per 48 minutes?

mystic, when you add all the "wins" from these rates, do teams sum close to their pythagorean wins?

I tested it for all seasons from 1994 to 2011. The average error is 0.67, RMSE is 1.06. If I compare that to the real amount of wins, I get as average error 2.56 and RMSE of 3.18. Linear correlation coefficient to the real wins is 0.973.

[Mike G]

Mike, is it possible to list an eWins per 48 minutes?

Sure, but you can just mentally move the decimal over one spot.
An average game tends to be ~48.4 minutes most years, and that's how long it takes, on avg, to generate a win.
Then 1.0 W per 484 min is an average rate, per player. Fewer decimal places, fewer zeroes, etc. In a list, it's easy to spot who's above and who's below avg.
Per 48, avg is about .0992

If they could help it, coaches wouldn't play their replacement-level talent (or rookies) against good opponents, because it doesn't help them win.

Actually, near-replacement guys are often used against superior talent, in order to wear them down or give the starters a rest, etc.
Rookies aren't necessarily going to be held back from getting "schooled" by great players, and it's not necessarily better to leave your best players in against the best opponents; a mismatch against you now will later be a mismatch in your favor.

But when you name a player a Starter, it shouldn't be a frivolous thing. I assume nobody wants to be demoted from any promotion. You have to want it and deserve it.
Coming off the bench, and playing against a good number of subs, a rookie experiences a medium level of competition, between amateurs (college) and NBA starters.

[Mike G]

Now 17.1% thru the season, rookies are looking at 62 eWins. This jump is largely due to more minutes.
East rookies projecting to 36 eW and West 26.
Per 484 minutes, it's closer: .85 to .71

Code: Select all

e66    per36 rates    tm   Min   Eff%   Sco   Reb   Ast    PF   Stl   TO   Blk   e484
7.4   Irving,Kyrie   Cle   27   .555   26.3   4.6   7.5   3.9   1.1   4.6   .7   2.00
5.9   Rubio,Ricky    Min   30   .545   13.1   4.7  10.8   2.4   2.0   3.6   .2   1.43
3.8   Brooks,Marshon NJN   27   .544   19.3   5.7   1.3   2.6   1.4   1.8   .3   1.12
3.1  Morris,Markieff Phx   21   .594   14.7   8.9   1.7   4.5   1.1   1.6  1.0   1.07
2.9   Knight,Brandon Det   30   .549   16.2   5.4   2.9   2.8    .8   3.7   .2    .72

2.9   Leonard,Kawhi  SAS   23   .519   12.1   8.3   1.1   1.5   1.9    .9   .5    .91
2.8   Shumpert,Iman  NYK   31   .449   15.0   4.4   4.9   3.3   2.8   3.5   .6   1.03
2.7 Thompson,Tristan Cle   18   .500   15.6   9.9    .4   3.9    .5   2.7  2.3   1.09
2.6 Parsons,Chandler Hou   23   .510   13.1   9.2   1.7   3.4   2.0   1.7   .6   1.01
2.5   Leuer,Jon      Mil   19   .554   13.4   8.0   1.8   3.4    .8   1.1  1.1    .96

2.4   Cole,Norris    Mia   22   .478   15.5   2.9   5.5   4.0   2.0   3.6   .0    .80
2.3   Walker,Kemba   Cha   21   .462   15.5   4.3   3.3   1.5   1.2   2.5   .1    .81
2.2 Williams,Derrick Min   21   .508   13.9   7.4   1.0   2.6   1.2   2.0   .6    .76
1.9   Vucevic,Nikola Phi   14   .614   14.0  10.6   2.4   5.8   1.3   1.6  1.2   1.24
1.7   Kanter,Enes    Uta   13   .483   10.8  13.1    .5   2.7    .5   2.0  1.3    .95

e66    per36 rates    tm   Min   Eff%   Sco   Reb   Ast    PF   Stl   TO   Blk   e484
1.6   Thomas,Isaiah  Sac   14   .527   15.0   3.5   3.8   1.2   1.6   2.7   .4    .82
1.5   Stiemsma,Greg  Bos   12   .639    9.8  10.1   1.8   6.9    .4   1.4  5.3   1.20
1.4  Harrellson,Josh NYK   20   .491    8.5   8.1   1.1   2.4   1.1   1.3  1.0    .52
1.2   Burks,Alec     Uta   11   .471   16.5   4.8   2.5   2.3   1.0   2.0   .3    .88
0.7   Harris,Terrel  Mia   14   .519   11.4   9.7   3.0   2.8   1.6   3.2   .0    .80

0.7   Pargo,Jeremy   Mem   17   .385    9.7   3.0   4.8   2.7    .8   2.3   .0    .31
0.7  Singleton,Chris Was   23   .541    8.4   5.7    .3   4.9   2.1   1.0   .7    .22
0.6   Butler,Jimmy   Chi    4   .900   31.3   8.2    .0    .0    .0    .0  1.1   2.43
0.6  Fredette,Jimmer Sac   24   .456   11.2   2.3   3.2   1.6    .7   2.6   .0    .20
0.6   Johnson,Ivan   Atl   12   .461   13.5   8.7   1.1   7.3   1.7   3.8   .6    .56

Kyrie Irving's eff% is tracked in this thread first at .435, then .465, .506, .517, and now .555 .
Of 225 players >180 minutes, I have ranked him behind only Kobe, LeBron, Carmelo, Durant, and Rose as a Scorer.
Also 15th in Assists; way high in TO, though. And 12th in eWins/484

[Crow]

On winshares / 48 minutes for career to date, the top 7 picks as a group for the 2007, 2008 and 2009 drafts are doing 20-25% better than picks 8-14 as a group, as one might expect. For the 2010 draft the the top 7 picks are on average about 20% behind the next 7 picks. For the 2011 draft, picks 8-14 are doing very very poorly so far. You'd be better off with replacement players for this season.

[Mike G]

Top 7 picks and next 7

Code: Select all

D#   e66    per36 rates     tm   Min   Eff%    Sco   Reb   Ast  Stl  Blk   e484
1    7.4   Irving,Kyrie    Cle   298   .555   26.3   4.6   7.5  1.1   .7   2.00
4    2.7  Thompson,Tristan Cle   197   .500   15.6   9.9    .4   .5  2.3   1.09
2    2.2  Williams,Derrick Min   230   .508   13.9   7.4   1.0  1.2   .6    .76
3    1.7   Kanter,Enes     Uta   131   .483   10.8  13.1    .5   .5  1.3    .95
7    0.3   Biyombo,Bismack Cha   139   .360    4.7   7.8    .2   .5  3.5    .18
6   -0.2   Vesely,Jan      Was    52   .429    3.7   4.2    .5  3.1  1.2   -.31
T   14.0                        1047                                       1.07
 
D#   e66    per36 rates     tm   Min   Eff%    Sco   Reb   Ast  Stl  Blk   e484
13   3.1   Morris,Markieff Phx   236   .594   14.7   8.9   1.7  1.1  1.0   1.07
8    2.9   Knight,Brandon  Det   358   .549   16.2   5.4   2.9   .8   .2    .72
9    2.3   Walker,Kemba    Cha   249   .462   15.5   4.3   3.3  1.2   .1    .81
12   1.2   Burks,Alec      Uta    98   .471   16.5   4.8   2.5  1.0   .3    .88
10   0.6   Fredette,Jimmer Sac   285   .456   11.2   2.3   3.2   .7   .0    .20
11   0.4   Thompson,Klay   GSW   184   .525   11.6   3.7   2.2   .5   .2    .15
14  -0.1   Morris,Marcus   Hou    17   .227    6.2   3.9    .0   .0   .0   -.30
T   10.5                        1427                                        .61

If these same players were with better teams, you can be sure their WS/48 would read higher.
You cannot get a Markieff Morris, a Brandon Knight, or a Kemba Walker off the streets or from the D-League.

[Crow]

I was referring to group average and based on winshares per 48 as a conversation point.

Markieff Morris is performing very well. I liked him in the draft and more than his brother.

On the other two, it again comes down to looking at incomplete boxscore stats vs boxscores with a crude attempt at adding shot defense (i.e. winsahres per 48) or a complete metric like RAPM. On early RAPM estimate Brandon Knight has twice the negative impact on defense as he has a positive impact on offense. He isn't replacement level but he is estimated as a net negative on RAPM. Kemba Walker is estimated as having a negative impact on offense and defense (and worse impact on team offense) and would be fairly near to the average level of a replacement player.

Fredette is fairly near to the average level of a replacement player on RAPM and looks bad on boxscore stats too so far. Thompson and Marcus Morris are currently estimated near neutral on RAPM.

I guess winshares / per 48 is the harshest metric for these guys for the reason you noted (on bad teams, usually bad defensive teams).

[Mike G]

I guess what you're calling "negative impact" is what I call "below NBA avg", i.e., <1.00 eW/484
On a very good and deep team, a .70 or .80 player might have some negative impact from time to time, and he probably won't get much time.
On a team with one of the league's 3-4 worst offenses and defenses, players much worse than that will have to get serious minutes.

Or are you saying the Pistons and Bobcats have some really good players that they just aren't using?

Knight looks like Det's 2nd or 3rd best scorer, their 2nd or 3rd best distributor, and a pretty good rebounder. They need these productions. He shoots .545, and the rest of the team shoots .483 .

Walker hasn't shot well, which is to say he's .016 worse than the rest of the team (.480 eff%). He, too, has only 2 to 4 teammates who are as productive overall.