Statman's new site....

[DSMok1]

Dr. Positivity--

Do you have a site where you post your analysis/scouting of college prospects? I would be really interested in hearing your insight as I've noticed you have a fairly unique take on things.

His site is http://asubstituteforwar.com/

[Dr Positivity]

Dr. Positivity--

Do you have a site where you post your analysis/scouting of college prospects? I would be really interested in hearing your insight as I've noticed you have a fairly unique take on things.

I made a top 20 most talented prospects list here though I only say a few things about each prospect/ranking http://asubstituteforwar.com/2013/01/31 ... uary-2013/ (Like I said by trying to grade talent instead of guessing production or value, I go in the opposite direction of statguy/draft rater sites, most statguys would probably near puke at the sight of Nash in a top 5, but talent wise I see a really dynamic player)

[CrackersPhinn]

Dr Positivity--

Thanks for the reply. I've read a lot of your stuff now and it's pretty interesting. Question about Nash: I can ignore a lot of stat-stuff about a prospect, and I can almost see where you're coming from on the guy based on the videos I've watched, but what I cannot really get over is: "Why does he turn the ball over so much?" and "How come he commits so many fouls when he doesn't do a lot of the things (rebounding, defense) that get a player into foul trouble?" Are these really one issue (offensive fouling) that you think moving to the more diffuse spacing of the nba might fix? But shouldn't a player of his skill level and feel for the game be able to counter that by year 2 of dealing with college ball?

[Dr Positivity]

Yeah, like I said his production is easily below the Mendoza line where I'd start to worry something is wrong. Very few legitimate prospects are under 20 PER in college (only Goodwin, McAdoo and Nash among name players. Last year IIRC Teague and Rivers were the guys below it, even the much maligned Drummond, Jones III and Lamb were above). I'm much more comfortable saying C.J. Leslie's reputation as a motor-less, underperforming enigma is pre draft Drummond, Monroe style BS, as Leslie has been fine production wise in his career (21-23 PER). With Nash no doubt the concerns are real.

You could also make an argument that even if Nash makes it as a long time starter, he does it on a Gay-like high volume/inefficiency scorer who takes too many midrange shtos, and a lot of members in the apbr/advanced stats community would say "I don't want that guy anyways". With that said in regards to college production, the question is as much "what he can do" as much as what he does. If someone watched one of his on games/clips like this without knowing anything else about the player or any other games, they would guess that he's one of the NCAA's best scorers just based on the tools he shows. You show that clip or ones similar to it, a clip of a college Beasley game, and a clip of college Melo game, to an alien unaware of the rest of their careers, and there's no way they'd guess there's an insane gap in terms of college success

[Statman]

Yeah, like I said his production is easily below the Mendoza line where I'd start to worry something is wrong. Very few legitimate prospects are under 20 PER in college (only Goodwin, McAdoo and Nash among name players. Last year IIRC Teague and Rivers were the guys below it, even the much maligned Drummond, Jones III and Lamb were above). I'm much more comfortable saying C.J. Leslie's reputation as a motor-less, underperforming enigma is pre draft Drummond, Monroe style BS, as Leslie has been fine production wise in his career (21-23 PER). With Nash no doubt the concerns are real.

You could also make an argument that even if Nash makes it as a long time starter, he does it on a Gay-like high volume/inefficiency scorer who takes too many midrange shtos, and a lot of members in the apbr/advanced stats community would say "I don't want that guy anyways". With that said in regards to college production, the question is as much "what he can do" as much as what he does. If someone watched one of his on games/clips like this without knowing anything else about the player or any other games, they would guess that he's one of the NCAA's best scorers just based on the tools he shows. You show that clip or ones similar to it, a clip of a college Beasley game, and a clip of college Melo game, to an alien unaware of the rest of their careers, and there's no way they'd guess there's an insane gap in terms of college success

At this rate, I'm guessing my statistical projections will not Like Nash at all come draft time. You are right about the "Mendoza line" for college - there's a point in general production a guy really needs to be above to show he is able to produce anywhere near his perceived talent level - and although I'm not sure yet where that line is (I'll know much more after I get all the past seasons done) - I'm pretty comfortable in saying Nash is definitely still below it. His teammate however - Marcus Smart - as a frosh looks (statistically) to have a great future.

[PD123]

Statman-

I'm curious if you/your model try to do things to account for various "structural oddities" on various teams. For instance, you've got Taylor Smith from Stephen F. Austin ranked 9th overall (and you've been pretty him consistent with that ranking) despite his size and current position (a 6'6" center). I know in your original post you said that history has shown that guys who play well get drafted, physical limitations notwithstanding, but I'm curious if that changes depending on team-makeup/opposition faced; the red flag for me about Smith isn't so much his size (DeJuan Blair wasn't much taller after all), but rather the size of his team. Only one player on the team who plays is taller than him, that player only averages about 7 minutes/game, and he's only 6'7". So how much of Taylor Smith's production is representative of actual skill and how much of it is representative of a lack of competition (both for rebounds and low-post FGA) that might not exist at the NBA level?

Also, do you make any attempts to isolate specific skill-sets for players (e.g. rebounding, ball handling/distribution, three-point shooting, etc.) and say that Player X isn't very good but he does this one aspect of the game very well and that might make him better than some people who are ahead of him and more well-rounded, or is that something that you've found either isn't very helpful/not worth the time it would take.

[Statman]

Statman-

I'm curious if you/your model try to do things to account for various "structural oddities" on various teams. For instance, you've got Taylor Smith from Stephen F. Austin ranked 9th overall (and you've been pretty him consistent with that ranking) despite his size and current position (a 6'6" center). I know in your original post you said that history has shown that guys who play well get drafted, physical limitations notwithstanding, but I'm curious if that changes depending on team-makeup/opposition faced; the red flag for me about Smith isn't so much his size (DeJuan Blair wasn't much taller after all), but rather the size of his team. Only one player on the team who plays is taller than him, that player only averages about 7 minutes/game, and he's only 6'7". So how much of Taylor Smith's production is representative of actual skill and how much of it is representative of a lack of competition (both for rebounds and low-post FGA) that might not exist at the NBA level?

Also, do you make any attempts to isolate specific skill-sets for players (e.g. rebounding, ball handling/distribution, three-point shooting, etc.) and say that Player X isn't very good but he does this one aspect of the game very well and that might make him better than some people who are ahead of him and more well-rounded, or is that something that you've found either isn't very helpful/not worth the time it would take.

Thanks PD for asking these questions - the answer is yes (sorta) and yes.

1. First yes - it's a sorta. The sorta yes is that the VERY good rating for Smith goes well beyond being really the only pseudo big guy on his team - mainly because the rating is SO impressive. There are many players that are their team's only "big" (even at 6'6") - but they don't dominate anywhere to the extent Taylor does. Not close. I STILL haven't gotten to see him play btw. Statistically (despite weak SoS), he appears to possibly be an elite athlete (or at least tons of "bounce" w/ probable very long reach and better lateral movement than opposing "bigs"). He had 4 blocks and 3 steals in a road win Oklahoma. He averages over 3 blocks and one steal a game on a team that doesn't really get blocks and steals (he has 75% of his team's blocks - and he MUST avoid foul trouble or they're screwed), with a better than 1 A/TO ratio - on a very slow paced team. They have been dominating their opponents (albeit generally very weak opponents) - and he is an obvious man amongst boys, no player on his team is near his talent (Pomeroy has them as the 51st ranked team in the nation - being by far the best player on debatebly a Top 20% D1 team means something). He probably can't shoot with any range - I'm guessing he dunks on people all day. He's probably similar Roberson of Colorado - with more finishing ferocity in the paint, less range and less rebounding instincts. I wouldn't be surprised - w/o ever seeing him play - that he'll see time in the NBA as a long shotblocking defender and finisher on the break - if he has anywhere near the lateral movement to cover the bigger wings and face up 4s (I'm guessing he probably does). He could easily be a late 2nd or FA invitee steal (or a guy we see a couple years later), guessing from his skillset ratings.

2. Definite yes. I debated breaking down the college ratings into skillsets this week for the readers - just didn't have the time because of the Chinese New Year (important holiday in my household). I really believe the skillset breakdowns (AFTER pace, SoS, etc adjustments mind you) will yield some great similarity score comparisons (AFTER I get all the past college season ratings done) - with obvious skillset threshholds that players need to show well enough in to even be considered a legit NBA possibility. I'm guessing Shabazz Muhammad ranking #2388 in the nation in his defensive stops skillset with a rating of 5 (steals, blocks, and PF are part of that rating) and Doug McDermott's #2897 ranking (out of 3030 mind you) with a rating of -1 will prove to be MAJOR red flags in terms of their NBA prospects. I'm guessing it's quite possible no single player in the NBA will have a defensive skillset rating that low from college. Nerlens Noel btw is #1 with a 59 rating JUST from his defensive numbers. Now - being GREAT in this skillset doesn't mean future NBA success - BUT being BAD in this rating PROBABLY means some serious future NBA struggles (lacking translatable athleticism). Throw in their mediocre/poor handles/court awareness/passing/whatever rating (assists and TOs after adjustments) of -5 for McDermott (#1823) and -9 for Muhammad (#2213, Drew of UCLA and Burke of UM are top 2 with a 66 and 64 rating) - that seems to be a recipe for NBA disaster for a future NBA wing (mediocre ball skills make it tough to create shots for others or yourself with the obvious step up in talent, while mediocre defensive rating indicate the lack of athleticism to guard the position adequately, blow by defenders, or elevate over defenders). I'm about 100% certain NO NBA player had a worse combination of defensive and handles skillsets in college than either of those two have shown this season thus far.

Shabazz is still projected as the #1 OVERALL pick by Draft Express in the next draft, McDermott a mid 2nd rounder.

While the overall ratings have some correlation for future success (much like minor league, say, OPS+ can give definite insight without ANY scouting on whether a guy MIGHT succeed at the next level - and who really has no shot) - and I, sadly, kinda trust them better than some GMs (Austin Rivers - REALLY?) - the skillset breakdowns should be the real enlightening factors when it's all said and done. That'll be fun.

BTW - I'll be using the NBA skillset breakdowns (I actually posted those in one of the PDFs for this season in the NBA) combined with age to project career trajectories of NBA players. That'll be cool too (at least to me). Having all the past seasons for retrodiction (at least since the three ball) will be nice.

I wish I had much more time than a couple hours a week..... one thing at a time.

[Statman]

You could also make an argument that even if Nash makes it as a long time starter, he does it on a Gay-like high volume/inefficiency scorer who takes too many midrange shtos, and a lot of members in the apbr/advanced stats community would say "I don't want that guy anyways". With that said in regards to college production, the question is as much "what he can do" as much as what he does. If someone watched one of his on games/clips like this without knowing anything else about the player or any other games, they would guess that he's one of the NCAA's best scorers just based on the tools he shows. You show that clip or ones similar to it, a clip of a college Beasley game, and a clip of college Melo game, to an alien unaware of the rest of their careers, and there's no way they'd guess there's an insane gap in terms of college success

Dr. Positivity - note my point #2 in the post above about defensive and handles ratings - Nash's are even worse than Shabazz. Right now that's a serious red flag - his perceived athleticism just isn't translating well to production (with TWO years of data!) - and it doesn't tend to get better when the competition gets better. Combined with that turnover rate - trouble.

[Statman]

You have been busy. I see lots of different cuts of NBA data there now. I have downloaded some and will look into them a bit later when I have the time.

The NBA player rating/ranking updates for the All Star weekend are available:

http://www.hoopsnerd.com/2013_NBA_PDFs.html

[v-zero]

How are the ratings interpreted? I understand bigger=better, but in a plus-minus sense how do they translate?

[Statman]

How are the ratings interpreted? I understand bigger=better, but in a plus-minus sense how do they translate?

100 is league average. It's a ratio - one could mess with it. LeBron, with a 167, could carry a lineup of average players and make them better ((167+4*100)/5=113.4) than a lineup of 5 quite solid players who are all at 110.

I set replacement level at 80.

Great teams tend to be around 110 (pts/OPPpts). San Antonio & OKC are 109 right now. Charlotte is 91.

Miami without LeBron's production is 91. The Lakers without Kobe is 93. OKC without Durant is 96. Take away San Antonio's best player - Parker - and they still are a 102.

College ratings are more extreme - since the average is an average D1 player (347 teams) - great teams tend toward the 135 to 140 mark - great players head toward (and occasionally above) 200.