Leading rookies according to xRAPM

[EvanZ]

Looks like Drummond (-0.8), Draymond Green (-1.0), and MKG (-1.1) are leading the rookies.

I checked specifically to see where Draymond is, because he's been a joy to watch really. I don't know anyone who thought he'd have the kind of defensive impact he's having, let alone as a rookie. He's guarded pretty much every position at times.

[talkingpractice]

Can you help me with what to make of the Woyas rookie trio?

We have Draymond the best rookie in the league on defense (with Ezeli not that far behind).

But we also have Draymond at about -3.5 on offense (with Ezeli not much better).

[EvanZ]

Can you help me with what to make of the Woyas rookie trio?

We have Draymond the best rookie in the league on defense (with Ezeli not that far behind).

But we also have Draymond at about -3.5 on offense (with Ezeli not much better).

There's a reason the Warriors have been such a huge surprise defensively (and rebounding), and Draymond/Ezeli are a big part of it. Throw in Barnes and it's one of the best draft classes I can remember in terms of hitting on each player.

[Mike G]

Draymond Green is hitting .301 eFG% on 88 FGA.
Last players to go a season with at least 88 FGA and such a low eFG% -- Ryan Humphrey in 2003 and Rafer Alston, 2000. Before that, you have to go back to 1972.

Looking at recent forwards who have hit as low as .350 in a season, we see a few guys at the very end of their careers, and a bunch who never had much of a career.
http://bkref.com/tiny/JZCod

[Crow]

As lousy as Draymond is on shooting /scoring efficiency his rebounds and steals help put him ahead of rookie Jeff Geen on winshares per 48. And 5th year Jeff Green only has a very tiny / pretty insignificant edge on rookie Draymond on ws/48. Might not keep it by end of season. Draymond already has a fairly large lead on xRAPM estimate over this season Jeff.

After 4 games in 7 with a gamescore over a modest 10 that prompted "he has it going now" talk, Jeff Green has posted 4 straight under 9, 3 under 7.


And brief notes about another Celtic: Jason Terry is getting about 35% fewer mid-range shots per minute than last season (be it off pick n roll or anything else) but making them a lot more. 20% fewer 3 pt attempts per minute with no change in efficiency there. A mixed bag with his new teammates. Lowest PER of his career, 4th lowest ws/48 (but up slightly from last season), offensive rating below career average (but better than last season), negative XRAPM estimate (but not as bad as last season).

[EvanZ]

Draymond Green is hitting .301 eFG% on 88 FGA.
Last players to go a season with at least 88 FGA and such a low eFG% -- Ryan Humphrey in 2003 and Rafer Alston, 2000. Before that, you have to go back to 1972.

Looking at recent forwards who have hit as low as .350 in a season, we see a few guys at the very end of their careers, and a bunch who never had much of a career.
http://bkref.com/tiny/JZCod

His shooting has been astonishingly bad. Nobody would deny that. It's all the more testament to how much impact he's making in other areas that at times he seems like the best player on the floor, and Mark Jackson keeps giving him more and more minutes when the game is on the line.

Having said that, I think the stat you pulled out is a bit unfair, since it assumes an entire season played with so few FGA (of course, there are going to be bad players who didn't get much playing time, couldn't shoot, etc). What would be more relevant is finding a stretch of 88 FGA throughout the season with such a low average among all players.

[Mike G]

Yeah, I gave him the benefit of the doubt to look at eFG > .350
He could heat up and blister the nets at .360 the rest of the way.

For some reason, I speculated that Green sounds like the rookie Renaldo Balkman.
http://bkref.com/tiny/L5P1D

Some similarities, and some major differences.
From a promising rookie season, Balkman faded quickly.

[v-zero]

SEBF has these numbers for the 30 guys who've played more than 150 minutes this season. These numbers are non-prior-informed, just cumulative performance broken down into a per 48 number, hence the 150 minute cutoff.

Code: Select all

1.  Andre Drummond , 1.9
2.  Jared Sullinger , 1.3
3.  Nando De Colo , 0.5
4.  Pablo Prigioni , 0.5
5.  Jonas Valanciunas , 0.2
6.  Jae Crowder , 0.0
7.  Harrison Barnes , -0.0
8.  Maurice Harkless , -0.0
9.  Damian Lillard , -0.1
10.  Festus Ezeli , -0.1
11.  Draymond Green , -0.1
12.  Anthony Davis , -0.2
13.  Meyers Leonard , -0.3
14.  Alexey Shved , -0.4
15.  Michael Kidd-Gilchrist , -0.5
16.  Bernard James , -0.7
17.  Kyle Singler , -1.0
18.  Tyler Zeller , -1.0
19.  Darius Miller , -1.8
20.  Thomas Robinson , -1.8
21.  Brian Roberts , -2.0
22.  Bradley Beal , -2.0
23.  Andrew Nicholson , -2.3
24.  Terrence Ross , -2.4
25.  Dion Waiters , -2.4
26.  Austin Rivers , -2.5
27.  Jeff Taylor , -3.0
28.  John Henson , -3.2
29.  Will Barton , -3.9
30.  Doron Lamb , -4.8

It appears Draymond is working miracles on D, because as has been pointed out he's been horrible shooting the ball, and hence most of the box score metrics aren't very interested in him. No shock that the Spurs pick up another good looking foreign talent in Nando De Colo - his shooting hasn't been great (but TS% is fine, and good for a rookie) but his impact has been good, he had a fantastic game against the Heat in Miami when he played a full 34 minutes.

[talkingpractice]

Is your metric of the type that you can say 'why' it appears that some of these guys (Sullinger, De Colo, Prigioni) are valued highly in your model? Our rookie values (in the other thread) are very very different than what you have, so comparing them may be useful for each of us.

[v-zero]

I haven't done the analysis myself (yet) but for some insight here is a regression of box-score stats against the full rating. It explains about 70% of the variation in the rating, the rest comes from being tied to team performance and team mate performance.

Rating = -0.05*MINS - 0.23*AST + 0.59*OFFVERS + 0.34*BLK +0.24*DRB + 0.44*DEFVERS - 0.43*FGA - 0.25*FTA + 0.46*ORB - 0.04*PF + 0.41*STL -0.69*TOV + 0.28*TSPTS

Here everything should be obvious, except OFFVERS, DEFVERS and TSPTS, these are:

OFFVERS = SQRT(PTS*AST)
DEFVERS = SQRT(BLK*STL)
TSPTS = SQRT(PTS*TSA) (which with some rearrangement can also be written TSA*SQRT(2*TS%)

Both OFFVERS and TSPTS have an effect on the value of usage - high-usage high-assist guys (like Westbrook) whose shooting isn't all that great percentage wise get a big boost from this interaction, since the implication appears to be that high-usage guys create better assists for their team mates. The TSPTS part penalizes usage less at the high end than straight points to account for further usage effects.
DEFVERS attempts to capture dynamic defenders whose ability to go from inside to perimeter means they get both steals and blocks.

As I said this can only tell you so much, but it may help you see where differences lie. I should say that the above formula will give you a player's rating over whatever period of time the stats you are using were recorded, hence the MINS part. To get a per-48 rating you can just pro-rate it.

EDIT: Just remembered that for this model I used TSA = FGA + TOV + 0.44*FTA rather than the usual non-TOV version - this is pretty crucial.

A few per-48 values using this equation for this season:

Kobe: +0.9
King James: +4.4
KG: +2.2

[talkingpractice]

I think your OFFVERS, DEFVERS, and TSPTS variables are interesting. They definitely seem to 'credit' a lot of the high usage pattern stuff, which happens in our IPV model too. But then we differ on Kobe sooooooooooooooooo much. I guess the Kobe situation could be an interesting thread, separate from this one.

[v-zero]

I do find it interesting that it differs on Kobe so hugely compared to so many other models, even those based on the box score. My pure instinct on Kobe says he's now closer to average than all-star, but the whole reason we're here is that 'gut' isn't anything like the whole story.

[EvanZ]

Draymond went 2 for 4 tonight and hit 2 free throws. He was feeling it!

[J.E.]

Coaching and age adjusted 12 year RAPM ranks the rookies like this

Drummond +3
Waiters
MKG
Lillard
Draymon Green
Singler
Crowder
Prigioni
..
14. Shved
..
17. Barnes
..
29. Anthony Davis
..
43. Robinson
..
56. Rivers
63. Valanciunas
69. (out of 69) Bradley Beal -3

For reference. LBJ scores a +9.3 in this metric

I'd say age adjusted RAPM judges talent, not actual current impact; to get an estimate of current impact you'd have to add influence of current player age. What I want to say is that a 18y old player on this list might be currently performing worse than a 20 year old player, but it's still possible the 12y RAPM ranks the 18y old higher (=as a better basketball player) because he is expected to improve his game in those two years.

Data through January 2nd. Things would look a little different if I ran it now

[EvanZ]

Where's Ezeli on that list?