What's new

Advanced +/- for draft class

idiot

Well-Known Member
Don’t know if you are aware of this site with college advanced +/- season statistics since 2009-10: https://godismyjudgeok.com/DStats/2013/nba-stats/ncaa-advanced-statistical-plus-minus-released/

As with other +/- stats, it’s dependent on the particular formula the statistician chooses to use, as well as how the player fits into the team's concept; and it should never be taken to be the end-all-be-all stat. But like good +/- stats (as opposed to raw, especially single-game usages), it attempts to value a player’s rate of offensive, defensive, and total contributions to his team compared to an those of average player--after taking teammates/opponents into account, and over a long-term sample size.

I just discovered this site and thought I’d highlight what I found interesting about this year’s possible draft picks. Check it out yourselves because it has worthwhile visualizations that I cannot reproduce here.

Generally, after playing around a bit with previous year’s data, these are the simple rules of thumb I’m following:
a) any number -- offensive or defensive -- close to average (0) gives a red flag. Most successful NBA players showed decently above average numbers both offensively and defensively.
b) I like to see improvement from year to year, where the data is available.
c) generally smalls show better offensively, and bigs defensively

Numbers for draftables (Total, offense, defense):

Trey Burke 11.6, 9.1, 2.5 (very high total, great offense; defense slightly worrisome, but better than I expected; good improvement from freshman)
CJ McCollum 9.9, 6.1, 3.7 (good numbers overall; big improvement from sophomore to junior year; defense somewhat inconsistent from year to year and generally on the low side)
Pierre Jackson 9.0, 7.8, 1.2 (not bad; but defense a clear issue; fair yearly improvement)
Shane Larkin 8.9, 5.4, 3.5 (much better than I expected; defense surprisingly OK; very good improvement from year one to two)
Erick Green 8.7, 9.2, -0.5 (very good offense, terrible defense; strongly regressed on defense over years)
Michael Carter-Williams 8.6, 3.4, 5.2 (great defense for PG, rivaled only by Marcus Smart; can low offense number work for PG?)
Nate Wolters 8.4, 8.2, 0.3 (very good offense; terrible defense; good yearly improvement)
Peyton Siva 7.7, 3.3, 4.4 (moderate yearly improvement; quality defense)
Myck Kabongo 6.6, 3.7, 2.9 (may have chance as showed good improvement from freshman)
Ray McCallum 6.0, 5.6, 0.4 (bad defense; offense OK)
Lorenzo Brown 5.9, 3.4, 2.5 (not much to see here)
BJ Young 5.1, 5.3, -0.2 (terrible, terrible defense; moderate improvement since last year)
Isaiah Canaan 4.9, 5.3, -0.4 (inconsistent from year to year; not terribly impressive; awful defense)
Phil Pressey 4.9, 2.7, 2.2 (regressed last year; nothing impressive)

Victor Oladipo 13.0 6.8, 6.3 (second highest total in database behind Anthony Davis; great balance, great improvement esp. sophomore to junior year)
Kentavious Caldwell-Pope 9.8, 5.6, 4.2 (well balanced; very good [Klay-Thompson like] jump from freshman to sophomore)
Ben McLemore 7.9, 5.4, 2.5 (not bad, some other freshmen in this range have done well)
Jamaal Franklin 7.5, 3.1, 4.1 (almost exactly what I would have expected; decent yearly improvement; a bit, but not terribly, far behind Kawhi’s numbers)
Allen Crabbe 6.1, 4.2, 1.9 (nothing impressive; slow yearly improvement)
Tim Hardaway, Jr. 5.6, 3.8, 1.9 (pretty low numbers across the board; only modest yearly improvement)
Tony Snell 4.7, 3.5, -1.3 (Meh)
Archie Goodwin 2.7, 1.5, 1.3 (very risky, even for freshman)

Otto Porter, 11.0, 5.8, 5.2 (very good balance between offense and defense; good improvement from a very good freshman year)
Reggie Bullock 7.9, 6.0, 1.9 (decent overall; surprisingly ineffective defense; fair yearly improvement)
Deshaun Thomas 7.3, 6.1, 1.2 (no surprises here; little yearly improvement; may hang on in NBA with offense)
Solomon Hill 6.5, 4.2, 2.3 (slow, steady improvement)
James Ennis 5.1, 3.4, 1.7 (regressed strongly last year on defense)
CJ Leslie 3.4, 1.9, 1.4 (not impressive; regressed from previous year)
Shabazz Muhammad 3.0, 3.8, -0.8 (these numbers scare me for a freshman; I’d want to see capacity for year-to-year improvement before drafting; terrible defense, offense nothing to write home about)

Cody Zeller 11.4, 7.7, 3.7 (very high for sophomore; improved slightly from freshman, though regressed a bit defensively; has always been higher than his brother)
Kelly Olynyk 11.3, 8.6, 2.7 (great offense; can he play defense?; earlier years’ scores unavailable)
Arsalan Kazemi 10.7, 3.9, 6.7 (steady yearly improvement; I want him, despite limitations)
Mike Muscala 10.0, 6.0, 4.0 (better balance than I expected; strong yearly improvement)
Eric Murphy 9.4, 6.1, 3.3 (better than I expected; good improvement over final year)
Mason Plumlee 8.8, 5.1, 3.8 (decent numbers, though perhaps a bit low for age; steady, if not rapid improvement through years; always better than brother)
Anthony Bennett 7.6, 4.7, 2.9 (decent for freshman who has room for improvement)
Andre Roberson 8.1, 1.4, 6.7 (basically the same player as when freshman; great defense for size; no offense)
with defensive focus)
Trevor Mbakwe 6.9, 3.3, 3.7 (not much yearly improvement; energy guy)
Brandon Davies 6.8, 4.0, 2.8 (steady, if not rapid yearly improvement)
Jackie Carmichael 6.1, 3.2, 3.0 (not great; but good balance and steady yearly improvement)
Grant Jerrett 5.5, 2.2, 3.3 (not bad for 2nd-rounder freshman)
Tony Mitchell 1.4, -0.5, 1.9 (regressed as sophomore; good but not great as freshman; I wouldn’t touch him)

Nerlens Noel 11.2, 2.7, 8.6 (defense is best in database, offense is slightly worrisome for a #1 pick but probably OK since he was a freshman)
Jeff Withey 10.2, 3.9, 6.2 (impressive numbers; not much improvement junior to senior)
Gorgui Dieng 10.1, 3.6, 6.5 (very similar to Withey; strong improvement from sophomore offensively, though still on low side there)
Steven Adams 8.8, 3.1, 5.7 (fairly impressive for freshman; can he develop offense?)
Colton Iverson 7.3, 5.3, 2.0 (not much defense; huge offense improvement since Minnesota)
Dewayne Dedmon 6.3, 0.1, 6.2 (all defense, no offense)
Alex Len 6.2, 3.3, 3.0 (quite low overall for a high pick; but good balance and year 1 to year 2 improvement)


For comparison:
Favors 7.4, 3.3, 4.1
Hayward 7.5, 3.7, 3.8
Burks 6.8, 6.9, -0.1 (with modest improvement to sophomore)
Jeremy Evans 6.3, 3.7, 2.5

Kahwi Leonard 8.4, 3.7, 4.7 (with strong improvement to sophomore)
Paul George 4.9, 2.2, 2.8
Chandler Parsons 6.1, 3.6, 2.5
Jimmer Freddette 10.8, 9.8, 1.0
Damian Lillard 8.7, 8.9, -0.2 (good, steady, yearly improvement)
Klay Thompson 8.7, 5.5, 3.2 (big jump to sophomore)
Kemba Walker 11.7, 8.5, 3.2 (with huge last-year jump)
Brandon Knight 5.0, 4.4, 0.6
Harrison Barnes 5.7, 3.5, 2.1 (sophomore and freshman almost identical)
 
Last edited:
Thanks for the post.. good stuff in there. Unfortunately, the flaws are numerous/enormous.

As I've said before, Larkin was pulled off a good offensive player and moved to a poor defensive player (because he's a defensive liability). Stats don't tell that story.
Anyway, there's nice data here, but obviously data can only take us so far.
 
Thanks.

Agree that there's a lot of problems in the data. That's partly why I included the comparisons to current NBA players -- to show that it's anything but infallible.

I do tend to think that the extremes tell us something, however. (And since +/- does not depend at all on individual match ups or individual statistics, it should not be very sensitive to someone like Larkin being hid on defense -- though it might be a bit.) It's yet another point in the data (both visual and statistical) that ought to be taken into account.
 
Interesting post. There are the normal problems of there being so few games in a college season. 30 or 60 games isn't enough statistical significance.
 
Interesting. Without going through all of them, Paul George's numbers look about the worst on the whole list across the board. That guy's gonna suck.
 
Interesting. Without going through all of them, Paul George's numbers look about the worst on the whole list across the board. That guy's gonna suck.

Yes, just like any other stat, this can mislead on individual players, and certainly can't predict future development. (And maybe this is partly why several teams "missed" on Paul George.)

I'm not suggesting this should provide a proper draft order. But just like TS%, WP40, rebounding, 3pt%, TO%, scoring, or anything else, it can alert us to certain things that may not be easy to see in other ways.
 
Interesting post. There are the normal problems of there being so few games in a college season. 30 or 60 games isn't enough statistical significance.

Yep. That's one of the key problems with +/-. We can't be as statistically sure as we'd like. (And even if we had better statistical significance, it's still subject to the problem of being a snapshot of a single year and fairly dependent on how well a player fits in with a team's system.)

Of course I still think that it's better to add this statistic to the puzzle than to ignore it.
 
Analytics are nice and all, but I don't bother much with them they are misleading, And frankly I don't trust them. This is partially why wanting Locke is an idiot.
 
I would never say that analytics themselves are dumb or useless. It completely depends on how you interpret them.

This +/- table is interesting to look at, to see if we can confirm suspicions regarding certain players-- but it would be dumb to justify not drafting a given player due to a poor overall rating.


PS: one of the main varying Factors in these statistics comes down to how good a teams bench is
 
I would never say that analytics themselves are dumb or useless. It completely depends on how you interpret them.

This +/- table is interesting to look at, to see if we can confirm suspicions regarding certain players-- but it would be dumb to justify not drafting a given player due to a poor overall rating.


PS: one of the main varying Factors in these statistics comes down to how good a teams bench is
The problem I have with analytics Is people tend to rely only on the numbers and not their eyes. Analytics is just a reason to be lazy, unless used correctly, and even then they can mislead. I don't trust them.
 
I don't understand what the hell you said.

But I repped it anyway.

Looks like really good ****.

Like, way informative good stuff.

You put lots of time into it.

You made this site better with your post!

!!!!!!!!!!!!!!!!!!!! to you sir Idoit!!!!!!!!!!!!!!!!!!!!!!!!!



*Edit*
I read the Oladipo line.

Like, no kidding.

This is why he will go #2.

Cleveland is to vagination to take him number one.
 
It's worth noting that this is an estimate of +/- based on box score stats (points, rebounds, assists, etc.). That is, it's more like PER or win shares than actual +/-.
 
Yes, thanks GVC. You caught what I hadn't on how this stat is derived. Repped.

This means it's probably closer to an efficiency stat than a true adjusted +/- stat. Useful still, but not telling us quite what I thought it was telling us. Closer to info we already have available elsewhere than something we haven't seen before.

(Also means PKM's point about this stat not noticing that Larkin was hidden defensively may be truer than I realized.)
 
Back
Top