APBRmetrics

I like Jabari's season so far. He's looked more athletic than he did in college, might be the weight. I still think his career is going to be closer to Antoine Walker than Pierce and Carmelo but nothing wrong with a say a long career of 20/8 and spacing the floor at PF

Wiggins season is obviously troubling since his 3pt shooting regressed a few weeks ago (As much as in attempts as %, which I assume means they stopped coming out right in practice or something).

In terms of judging the class between the injured guys like Embiid, Randle, Gordon, buried guys like Vonleh and Adams, and Euros like Bogdanovic and Saric, clearly we're not getting the whole board here

Does this rookie class performance seem to anyone else to be abnormally bad so far? I did a quick look using the B-R play index and found that from 05-06 to 13-14, the weighted average PER for all rookies ranged from 11.2 to 13.8 and weighted average BPM -2.3 to -1.4. Yet currently the weighted average of this year's group is a 9.7 PER and -2.9 BPM (!) Last year's rookies already set the low standard in the years I looked at (11.2 PER, -2.3 BPM) and so far this year's class is much worse.

What I'm not sure about is the degree to which we can expect rookies to improve over the course of the season - perhaps by the end of the year, the numbers will look more like normal. But my instinct is that this seems like a pretty big gap to make up (the 2014-15 rookies have already played 11,307 minutes).

Or to look at it another way, here's the lottery rookies so far:

1. Andrew Wiggins - 705 MP, 9.4 PER, -5 BPM
2. Jabari Parker - 724 MP, 14.7 PER, -0.9 BPM
3. Joel Embiid - 0 MP (injured)
4. Aaron Gordon - 165 MP, 15.5 PER, -1 BPM
5. Dante Exum - 418 MP, 7.6 PER, -4 BPM
6. Marcus Smart - 164 MP, 11.8 PER, 0.5 BPM
7. Julius Randle - 14 MP (injured)
8. Nik Stauskas - 338 MP, 5.3 PER, -3.3 BPM
9. Noah Vonleh - 24 MP
10. Elfrid Payton - 639 MP, 10.4 PER, -2.1 BPM
11. Doug McDermott - 198 MP, 4.1 PER, -6.9 BPM
12. Dario Saric - 0 MP (in Europe)
13. Zach LaVine - 509 MP, 8.2 PER, -5.3 BPM
14. T.J. Warren - 104 MP, 10.1 PER, -3.6 BPM

Looks pretty bad, no? Injuries are certainly a factor, but we only have 3 rookies who are currently beating both the 2013-14 average PER and BPM (Parker, Smart and Gordon - and Smart just barely above last year's average PER), which was the worst rookie class in several years.

jmethven wrote:Does this rookie class performance seem to anyone else to be abnormally bad so far? I did a quick look using the B-R play index and found that from 05-06 to 13-14, the weighted average PER for all rookies ranged from 11.2 to 13.8 and weighted average BPM -2.3 to -1.4. Yet currently the weighted average of this year's group is a 9.7 PER and -2.9 BPM (!) Last year's rookies already set the low standard in the years I looked at (11.2 PER, -2.3 BPM) and so far this year's class is much worse.

What I'm not sure about is the degree to which we can expect rookies to improve over the course of the season - perhaps by the end of the year, the numbers will look more like normal. But my instinct is that this seems like a pretty big gap to make up (the 2014-15 rookies have already played 11,307 minutes).

Or to look at it another way, here's the lottery rookies so far:

1. Andrew Wiggins - 705 MP, 9.4 PER, -5 BPM
2. Jabari Parker - 724 MP, 14.7 PER, -0.9 BPM
3. Joel Embiid - 0 MP (injured)
4. Aaron Gordon - 165 MP, 15.5 PER, -1 BPM
5. Dante Exum - 418 MP, 7.6 PER, -4 BPM
6. Marcus Smart - 164 MP, 11.8 PER, 0.5 BPM
7. Julius Randle - 14 MP (injured)
8. Nik Stauskas - 338 MP, 5.3 PER, -3.3 BPM
9. Noah Vonleh - 24 MP
10. Elfrid Payton - 639 MP, 10.4 PER, -2.1 BPM
11. Doug McDermott - 198 MP, 4.1 PER, -6.9 BPM
12. Dario Saric - 0 MP (in Europe)
13. Zach LaVine - 509 MP, 8.2 PER, -5.3 BPM
14. T.J. Warren - 104 MP, 10.1 PER, -3.6 BPM

Looks pretty bad, no? Injuries are certainly a factor, but we only have 3 rookies who are currently beating both the 2013-14 average PER and BPM (Parker, Smart and Gordon - and Smart just barely above last year's average PER), which was the worst rookie class in several years.

Just wanna mention - my projected rookie WAR rank of the 9 college guys with "real" minutes above before the NBA draft - NONE of which were projected to be better than an average NBA player as a rookie (last four below or well below replacement player):

1. Smart 3.7
2. Warren 3.4
3. Parker 1.6
4. McDermott 0.5
5. Payton 0.3
6. Gordon 0.0
7. Stauskus 0.0
8. Wiggins 0.0
9. Lavine 0.0

I took a bit of heat before the draft by saying this rookie class didn't look very good (on twitter) - ESPECIALLY since the high draft picks (Wiggins, Parker, Gordon, Stauskus, Payton, McDermott, Lavine) who would get minutes merely based on draft position & guaranteed $$ weren't going to match the projected better (closer to league average) rookie per minute production guys (Napier, Stokes, McCrea, Dinwiddie, Adams, Warren, McDaniels, Prather, Smith, Anderson) who were drafted later (or undrafted) & weren't going to be assured of any minutes whatsoever. Many of those latter guys are older, which is partly why their projected per minute production SHOULD be better, while the draft mainly is based on career potential.

Smart & Parker stood out in my projections in terms of career, but Smart was the only guy that projected anywhere near "NBA average" as a rookie AND had a nice career curve (the younger Parker had the career curve). Embiid, if his personal foul rate wasn't horrible in college, would be included with these two. If Embiid starts playing in the NBA & produces a lower pf rate than projections - then his career WAR projections will improve, since all this is fluid whenever I decide to update the projections with new data.

jmethven wrote:Does this rookie class performance seem to anyone else to be abnormally bad so far? I did a quick look using the B-R play index and found that from 05-06 to 13-14, the weighted average PER for all rookies ranged from 11.2 to 13.8 and weighted average BPM -2.3 to -1.4. Yet currently the weighted average of this year's group is a 9.7 PER and -2.9 BPM (!) Last year's rookies already set the low standard in the years I looked at (11.2 PER, -2.3 BPM) and so far this year's class is much worse.

What I'm not sure about is the degree to which we can expect rookies to improve over the course of the season - perhaps by the end of the year, the numbers will look more like normal. But my instinct is that this seems like a pretty big gap to make up (the 2014-15 rookies have already played 11,307 minutes).

Or to look at it another way, here's the lottery rookies so far:

1. Andrew Wiggins - 705 MP, 9.4 PER, -5 BPM
2. Jabari Parker - 724 MP, 14.7 PER, -0.9 BPM
3. Joel Embiid - 0 MP (injured)
4. Aaron Gordon - 165 MP, 15.5 PER, -1 BPM
5. Dante Exum - 418 MP, 7.6 PER, -4 BPM
6. Marcus Smart - 164 MP, 11.8 PER, 0.5 BPM
7. Julius Randle - 14 MP (injured)
8. Nik Stauskas - 338 MP, 5.3 PER, -3.3 BPM
9. Noah Vonleh - 24 MP
10. Elfrid Payton - 639 MP, 10.4 PER, -2.1 BPM
11. Doug McDermott - 198 MP, 4.1 PER, -6.9 BPM
12. Dario Saric - 0 MP (in Europe)
13. Zach LaVine - 509 MP, 8.2 PER, -5.3 BPM
14. T.J. Warren - 104 MP, 10.1 PER, -3.6 BPM

Looks pretty bad, no? Injuries are certainly a factor, but we only have 3 rookies who are currently beating both the 2013-14 average PER and BPM (Parker, Smart and Gordon - and Smart just barely above last year's average PER), which was the worst rookie class in several years.

What's even more gross is there's only 2 sophomores from the 2013 lotto over 15 PER this season (Oladipo and Shabazz), so a really weird two year period for the top of the draft. I was wondering if there's a chance that a more "system-orientated" and analytics-driven league are making players fit in slower, but until it keeps happening, it's probably best to assume for now it's a coincidence

My top 10 in 2014 were

Embiid - injured
Stauskas
Randle - injured
Vonleh - bench (cmon Clifford your team sucks)
Adams - D League
Bogdanovic - Europe
Parker (just got carted off season in danger?)
Napier
Saric - Europe
McDermott (recently announced out for months)

So this class really get its legs taken out from underneath it for me, for now

And of course now Jabari is confirmed out for the year. Depressing for us draft junkies!

jmethven wrote:And of course now Jabari is confirmed out for the year. Depressing for us draft junkies!

It's kinda depressing doing draft models anyway - since it takes so many years to see how things pan out.

Statman wrote:

DSMok1 wrote:

Statman wrote:Yep, very much so. My metric makes assumptions that if a guy has miserable production but plays a ton - especially on a good team - his rating goes up significantly (probably a great defender). A guy with great per minute production but plays little - especially on a bad team - his rating goes down significantly (probably a bad defender). All players on their teams are adjusted accordingly, so player ratings still compile correctly at the team level. BPM has minutes played as part of the regression I believe - but I'm not sure the value is tied inversely to production. THAT would make sense to me in terms of solid correlations - if it hasn't been tested, it should be.

That's a clever approach. Sort of a Bayesian prior methodology, using minutes and "box score production" to inform a defensive prior estimate. Good idea.

Thanks, been using that approach for years, and have never doubted that it was the right way to go in terms of how I approach box-score metrics.

When you make such adjustments do you use a standardized adjustment method in every case or do you apply any subjective judgment about whether to apply or how much?

Are there any other adjustments that are not automatic and standardized?

Does avg BPM for lottery picks cross into positive ground by year 3, 4, or 5?

Crow wrote:Does avg BPM for lottery picks cross into positive ground by year 3, 4, or 5?

Oh, I'm sure it does on average - BPM isn't RAPM, it is box score - & player's box score stats tend to improve quite a bit, especially 19 to 22 years of age.

Crow wrote:
When you make such adjustments do you use a standardized adjustment method in every case or do you apply any subjective judgment about whether to apply or how much?

Are there any other adjustments that are not automatic and standardized?

Everything is standardized - SAME approach every team, every player. Absolutely no subjectivity on my part, all formulas. Same template for every NBA season. Same approach for team ratings every season. How the playing time adjustment applied to each player is dependent on team factors, player production, where the playing time is relative to production of teammates, etc. - the formulas adjust accordingly.

Same goes for my college player ratings.

Same goes for my NBA career projections.

I trust the methodology, and run with the results no matter how much they might fly in the face of popular opinion. This doesn't happen with the NBA player ratings, those tend to correlate fairly well with fan views of players - this happens much more with the college player rankings & especially college to pro projections, where an undrafted guy (like Javon McCrea) might be in my top 15 of prospects, while the #1 pick Wiggins sits at #17.

Most college fans think many of the best college players are the top freshmen - because those are the "new" guys that are hyped the most - many 1 & dones, & the guys who'll have the best shot of being drafted high. But, the college player rankings based solely on performance are dominated by upperclassmen every year. I don't make ANY class adjustment at all - its how it works out - the best players on average are the older guys that have developed & been around a while. This bothers some fans - they want to see Wiggins, Jabari Parker, Embiid, & Randall ALL in the top 10 overall - but they were ranked by HnI 149th, 17th, 23rd, & 21st nationally respectively. Their rankings based on NBA career projections for the draft class were #17, #2, #9, & #8.

For comparison, Doug McDermott ranked #3 nationally in HnI - but factoring in his age & weird statistical breakdowns (horrible st/bk rates particularly), his prospect rank for the draft was 37th. I took a TON of heat on twitter for that - I probably explained a million times that my model "sees" a player (evidenced by historically low bk/stl % against middling comp) who just wasn't going to be athletic enough to defend in the NBA, as well as create his own shot. The model saw him as a decently productive 3pt specialist who would rarely see the court due to all his other limitations.

Hindsight is always 20/20 - when I mentioned Adam Morrison was a college scoring machine who flopped - EVERYONE told me that it's obvious that McDermott is much better than Adam Morrison ever was. Adam Morrison being picked top 3 didn't matter to them in hindsight. Well, Wiggins being picked #1 or McDermott being picked #11 or Lavine being picked #13 or Stauskas being picked #8 doesn't matter to me - these picks my model sees the biggest possibilities of busts relative to pick position. Foresight is MUCH harder than hindsight.

Wow, I rambled off topic there. I'm tired.

Statman wrote: Well, Wiggins being picked #1 or McDermott being picked #11 or Lavine being picked #13 or Stauskas being picked #8 doesn't matter to me - these picks my model sees the biggest possibilities of busts relative to pick position.

I like Layne Vashro's HUMBLE projections that factor in the draft consensus, if not the actual draft order. Sort of like what Nate Silver mentions in his book - the scouts get more right than just a statistical model, but the models definitely give insight in terms of who may be undervalued. Which makes sense - good scouts can use the stats + other info, so should beat a pure quant method.

willguo wrote:

Statman wrote: Well, Wiggins being picked #1 or McDermott being picked #11 or Lavine being picked #13 or Stauskas being picked #8 doesn't matter to me - these picks my model sees the biggest possibilities of busts relative to pick position.

I like Layne Vashro's HUMBLE projections that factor in the draft consensus, if not the actual draft order. Sort of like what Nate Silver mentions in his book - the scouts get more right than just a statistical model, but the models definitely give insight in terms of who may be undervalued. Which makes sense - good scouts can use the stats + other info, so should beat a pure quant method.

I agree, the type of work I am doing that does not include projected draft position or height is intentional - I want the model to perform well on its own without the other stuff. We don't have good draft position projections early (they often change wildly, often correlating to my college player rankings). Height data is questionable until the combines. Plus, I just wanted to prove that the model in general filters out the vast majority of guys on its own - without the draft position consensus doing it for me & making my work "look" more like what the scouts foresee.

If I were working for a team, I would use my methodology as a tool to help pinpoint EVERY possible overlooked prospect in the world - helping scout resources. Obviously I'd be including world tourney info I could get my hands on at this point (EYBL, etc.) & past player league to league change in ratings to eventually come up with as viable translation I could get from every major world league to the NBA.

But, in a perfect scenario - the work I do would go hand in hand with a good scouting department. I am not naive to that - while I can make a model that might better past actual gm drafts without actual draft position bias - scouting would be very important in help gain separation between the guys that are fairly closely projected. Scouts also can look into work ethic & such (Beasley) that might very adversely affect progression compared to age norms.

As I said a few years ago - the Timberwolves would have never drafted some 26 year old guy on a whim because some scout saw him for a half - that stone would have long before been discovered & turned well before the draft if I or others with comparable comprehensive models based on box score stats (since that's ALL that we have in almost all international leagues) had been employed by that team.

Folks focused on the numbers were pretty down on this class relative to the hype. I think within FOs and among more obsessive draftniks (i.e. DX) there was a similar sense that the 2014 class just didn't live up to expectations by the time the draft rolled around.

Still... these guys have been even more disappointing so far on the NBA court. Aaron Gordon is the one guy who was starting to get me a little excited, but then he went down to injury.

BTW... the 2015 class is looking way better than the 2014 class did at this point. Outside of Oubre and Montrezl all of the players considered serious prospects are actual producing like they are.

vjl110 wrote: BTW... the 2015 class is looking way better than the 2014 class did at this point. Outside of Oubre and Montrezl all of the players considered serious prospects are actual producing like they are.

I agree, I expect the early projections (coming January) with this possible draft class will look more promising than the last.

I don't get why you aren't including height in your model. Your model is supposed to be predictive, so adding height should add prediction.