APBRmetrics

Doesn't seem like his article was caused much of a stir or reaction, yet. Anyone working on response articles?


So who is the eastern GM that he got the anti-analytics comments from? My best guesses would be Stan Van Gundy, Joe Dumars, Rod Higgins, Billy King, or Larry Bird.

crow, would help if you linked the article

might this be it?

http://www.forbes.com/sites/markheisler ... nba-draft/

Yes that article. Sorry, I thought it was probably widely known but should have dug it back up.

What a terrible waste of time and words

Yes, Smart is a good player. No, he’s far from the best prospect in this draft. That would be Jabari Parker

Oh. OK. Thanks for telling us, I guess?

I have one large issue with what he said. He acts as if every "stats" guy comes to the same conclusion in relation to player evaluation. How do we know how the analytics department from every team evaluates a player? I would hope they are doing evaluations on how well a player would fit into their system and with surrounding players. I would be shocked if every team would say that Marcus Smart would be the "best" player. Again, what is a "best" player anyways? Teams aren't spending their time with top 10 lists. I would agree if he would say that most media-driven "stats" people are over-evaluating a particular player. However, to paint the entire community with one large brushstroke isn't accurate.

The great thing about anti-analytics people is that they can freely prove any point using anecdotal evidence, because any more thorough assessment would be veering dangerously close to analytics

Projecting the development of 18 and 19 year olds is tough. Because somebody has potential, doesn't mean they will realize it. Just because somebody plays well over 30 games, doesn't mean they will over the course of 82 games.

strange article. it was somewhat nonsensical, as he hedged back and forth about when to bash analytics. It just strikes me he doesn't really understand them that well, which is fine... he's a writer. he's not paid to understand math, so why should he?

Bobbofitos wrote:strange article. it was somewhat nonsensical, as he hedged back and forth about when to bash analytics. It just strikes me he doesn't really understand them that well, which is fine... he's a writer. he's not paid to understand math, so why should he?

Well, he is presumably paid to write about things he understands...therein lies the problem

Bobbofitos wrote:strange article. it was somewhat nonsensical, as he hedged back and forth about when to bash analytics. It just strikes me he doesn't really understand them that well, which is fine... he's a writer. he's not paid to understand math, so why should he?

I don't think that's a good enough excuse. Allow me to make an example out of myself. I am not good at math, not at all. I am also not an analytics guy. I don't have a model of my own or anything. I am an NBA fan who find analytics extremely helpful and very cool. Even though it's mostly about the thing I am not good at, I try to understand analytics and most of all, I try to understand it what it is for. I've been following the "analytics circles" for some time (probably more than a casual fan) and every time I read some article that's anti-analytics, I cringe because they don't even understand the basic rules of analytics. Mark Heisler (and that Eastern Conf. GM) seems to think that Kevin Pelton is saying Marcus Smart is the best player in the draft. That's not what Pelton's saying. What Pelton is saying is this: I have a model, and when I apply this model to the draft prospects, Marcus Smart comes ahead of everybody else.

The other misconception I observed among the anti-analytics group is that they think analytics is one day going to replace the old school scouting and teams will be run by math geeks. That's not true and likely won't ever happen (although I'd love that to happen because math geeks are way more open to new ideas than old school people), because the human element in basketball matters a lot. The question is not whether analytics can replace the "eye test". Both are complimentary to each other and people need to understand that. And that's going to happen with more open and honest communication.

/rant

The biggest problem with this article, which also exists in NBA front offices, I'm sure, is that analysts generally present draft rankings (i.e. predictions) every year, whether they do it before the draft or retroactively, so people can see "who the model would have picked X years ago". Analysts do this because it's easy for them to do so, and knowing how well it would have worked in the past is obviously valuable information.
When anti analytics people make predictions, it's very often just for one or two players, never for every player in the draft class (at least, I've never seen anything like that).
Predictions from whatever analyst are obviously never perfect, and if you're given their predictions for every player in the last 10 years it's very easy to spot 10-20 "horrible mistakes". Because the anti-analytics person did not publish such a list, the analyst does not have the same ammunition.
Had the anti-analytics person published predictions for every player (e.g. "number of All Star selections", "minutes played", "PER", whatever) before each draft, I'm sure it would be full of "horrible mistakes", too. The anti-analytics people don't publish such lists, though. So, the anti-analytics people get to take a look at the analysts predictions with hindsight, while the analyst never gets to take a look on the anti-analytics' predictions
Thus, the anti-analytics can criticize the analysts model for "missing on Lillard" while the fact that they'd have picked Oden over Durant, or Morrison over Aldridge, always stays hidden

J.E. wrote:The biggest problem with this article, which also exists in NBA front offices, I'm sure, is that analysts generally present draft rankings (i.e. predictions) every year, whether they do it before the draft or retroactively, so people can see "who the model would have picked X years ago". Analysts do this because it's easy for them to do so, and knowing how well it would have worked in the past is obviously valuable information.
When anti analytics people make predictions, it's very often just for one or two players, never for every player in the draft class (at least, I've never seen anything like that).
Predictions from whatever analyst are obviously never perfect, and if you're given their predictions for every player in the last 10 years it's very easy to spot 10-20 "horrible mistakes". Because the anti-analytics person did not publish such a list, the analyst does not have the same ammunition.
Had the anti-analytics person published predictions for every player (e.g. "number of All Star selections", "minutes played", "PER", whatever) before each draft, I'm sure it would be full of "horrible mistakes", too. The anti-analytics people don't publish such lists, though. So, the anti-analytics people get to take a look at the analysts predictions with hindsight, while the analyst never gets to take a look on the anti-analytics' predictions
Thus, the anti-analytics can criticize the analysts model for "missing on Lillard" while the fact that they'd have picked Oden over Durant, or Morrison over Aldridge, always stays hidden

Totally agree.

All I can say is - if you take a model and re-apply it to all the past drafts (well, as far back as you can with the model) - and with no bias (not using actual draft position as part of the model) it generally out performs the past picks of supposedly the smartest basketball guys in the world using their best scouts and maybe analysts - well, it probably is a pretty good model. OBVIOUSLY there will be misses, some BAD, but there will be more hits on average. EVERY draft have TONS of misses - so a model that is probably ranking guys from a 200 player list deserves a number of misses. The model doesn't even have the benefit of team bias - ie teams desperately trying to get worth out of guaranteed $$ bad picks while cutting guys that probably REALLY deserve the shot that maybe models liked. This year - guys like Stokes, Dinwiddie, undrafted Javon McRae, etc. will have to be MUCH better than guaranteed contract guys to earn a spot on a team and justify their solid position on many models.

Hindsight is crazy. I've been in MANY debates about Doug McDermott, and about how his historically low (ie, no NBA player has ever had as poor) stl & stl+bk rates are MASSIVE red flags - probably an indicator of 1 or more of a number of issues (lacking athleticism, defensive instincts, lateral movement, D IQ, etc) that very well could manifest themselves on the O end also (being quick enough to dribble drive past NBA defensters, shooting over defenders, etc). Yet, if I mention ANY somewhat similar player that NBA gms drafted solely because of their scoring (ignoring red flags that weren't as blatant mind you)- say like Adam Morrison - ALL you get is "McDermott is 10x the player Morrison was", "McDermott is a much better defender and knows how to play the game", etc. If it was SO OBVIOUS Adam Morrison was going to fail - why did EVERY mock have him top 5, and why did he get drafted #3. Yet in hindsight, every anti-analytic person KNEW Adam Morrison would fail - of course with no documentation to prove this claim.

Just because we now KNOW past players were bad picks doesn't mean it was obvious then - or they WOULDN'T HAVE BEEN PICKED! But - don't ever dare compare a current projected lottery draftee to a past actual lottery draftee who ended up being mediocre or worse - because it's ALWAYS obvious to the anti-analytic people that the current guy is OBVIOUSLY better than that past now-proven-to-be-scrub. Don't tell people that probably half of the guys in the mock drafts will be busts (because half the guys in actual pasts drafts WERE busts) - because in their mind they are in the mocks because they all are good and should succeed. Of course - 3 years later - that person will swear they knew that the busts in the 1st round were going to be busts, and are MUCH worse than ALL the current guys in the CURRENT mocks. Don't tell them that a guy that isn't in 1st round mocks will probably succeed if given a real shot, because "if they were really that good, they'd be lottery in the mocks" - ignoring the fact that there are a number of 2nd rounders (sometimes small #) who actually do end up with long NBA careers in every draft and prove to be better than most of the 1st round picks in that draft.

Sorry about the rant - I've probably discussed the draft too much on twitter. Having people tell me I'm crazy because my model doesn't like Wiggins that much (#17), or McDermott (#37), but likes Adams/Stokes/Dinwiddie - and my trying to explain over & over WHY my model likes the guys it does (based on historical precedent) is tiring.

Loved this:

Nowhere are “advanced stats” more misguided than the draft with analytics guys using college numbers, which have never stood for much because the game is so different (40 minutes instead of 48 with starting players averaging under 30; zone defenses jamming the middle and taking away post play).

Yep, it's impossible for "advanced stats" to adjust to such a different game - that 40 to 48 minute conversion is nigh impossible. Took me years to get over that hump.

That was the most unfortunate type of anti-stats article, written with an old guard bitterness towards the stat movement as a whole and cherry picking whatever examples backed p his belief