Jeremy Lin out 2 weeks with grade one knee sprain (UPDATE: Will return vs. Portland)

Glad Jeremy is back. Hopefully he's not too rusty because we'll need all the help we can get.

 

Yeah, thought so too. Morey and his stats are too LOF.

 

Note that this is the argument used to in analyzing JLin's linsanity phase in NY. Despite the small sample size, his production was so outside the norm that Morey felt comfortable giving him an 8M/yr salary. Statistically the odds were very good that Lin would at least turn out to be an above average PG, with the possibility of being significantly above average.

 

Just saw meh's post in another thread about Dwight in response to some people claiming that stats show Rockets have a better defense when Howard sits than when he's on the floor:

You've got to admit the bolded is a bit ironic hahaha.

 

I would echo the earlier post about how difficult it is to get across how to properly analyze and contextualize statistics that are thrown at us almost every day without at least a year or two of study, especially when it comes to things like sports, where people think they know everything just because they understand the basic rules of the game and watch it constantly.

People always want easy answers, but the truth is that there's rarely an easy answer to an interesting question. Anything that's easy is understood intuitively by a child. Everything after that takes hard work and a lot of stretching outside of your normal comfort zone.

I'm a Lin fan, and I don't think that Patrick Beverley is nearly the defender that everyone thinks he is (in fact, I would be surprised if he was all that much better than Lin overall, in terms of points prevented). But do I have definitive proof that this is the case? No, I don't. I have my observations and hypotheses, but that doesn't constitute proof. Due to the nature of defensive statistics and the dearth of data we have on Beverley in particular, there's not a lot of good statistics to go on, so all we have at this point are basic counting stats and what our eyes tell us (and everyone knows how much our lying eyes lie).

 

This is what I have tried to point out. However, you have put it more succinctly and clearer.

 

Very nice.

People are often uncomfortable with the idea of data being inconclusive, or people telling them that "it's too early to say" and "there's nothing really convincing", but that's the real nature of data when you're dealing with small samples like in the NBA.

This is especially true when the differences/variance from player to player or season to season are small and can easily fall within the margin of error.

So when someone says that Beverley is a superior defender to Lin, and the data differential leans towards the small side, it's hard not to look at it as inconclusive and say, "Yah, maybe... but hardly convincing."

There is a value to the "eye test", but the typical observer of the NBA game doesn't see much beyond basic man to man defense, shot blocks and steals. Those who are more savvy can see failures in contests and basic rotations, but they often are unable to see how often a player is forced to cover for another player's failure to be in correct position or failure to rotate on the opposite side of the court.

And then there's the highly savvy observers that can spot smaller level positional errors like being 2 or 3 feet too far back or close, or the choice to funnel a player to the baseline vs center, or to funnel them towards the paint defender versus into another player providing them with an inadvertent screen.

Just like your weekend dinner party cook watching a line cook and a master chef watching a line cook, the judgement of what makes an average versus good versus great performance is often invisible to those of us who just watch the game on a non-professional basis.

And this doesn't even get into a discussion of differing eyes in terms of coaches. Some coaches have an exceptional eye for high level offense (like Thibs) and others not so much (insert coach here).

 

But to give meh credit, you do have to wonder though, at what point are these disclaimers on an internet message board slightly unnecessary? It would be like stating any general statement (taking up a quarter of a page) and then having ten pages worth of discussion on exceptions to that general statement -- even if that statement were accurate for 99% of applicable situations! These aren't legal documents being executed by sophisticated parties. These are sports fans on the interwebz

That being said, the small disclaimers you note in many of your posts really don't take that much time to write, so it probably is better form for more posters to do the same (again, in applicable situations). It is just a bit optimistic (or maybe I am just too cynical) to expect most people to adhere to such higher standards, especially in a casual environment haha. As Chris Rock once said, "[It's] like playing basketball with a r****ded kid and calling him for double dribble. Sometimes you've gotta let some s**t slide!"

 

On most boards I would agree, that noting incorrect statistical usage or qualifying data that's being presented can be a nominal effort/reward endeavor.

However, I believe that Clutchfans is different.

Since Morey was signed as the First Analytics driven GM of the NBA, and has acted upon those analytics to acquire players like Harden, Asik and Lin, I believe there is an unusually high number of statistical gear heads on this board.

This combined with the spread of analytics through team front offices, the NBA wide implementation of the SportsVu product, and the increasing reference to non conventional box-score statistics in the media and online is driving a wider level of fan awareness and understanding of statistics and statistical methodologies.

As such, I feel that those of us who believe and care about statistical analysis have some responsibility to help increase awareness of the various statistical methods, their limitations and usage. Ideally this is in as non-confrontational a method as possible, since historically revolutionary perspective shifts will typically lead to misunderstandings, inadvertent misuse, and outright rejection of or abuse of those methods.

I see it as similar to the advance of Sabermetrics in baseball. It was a slow and sometimes bumpy process, however over the years it was the work and efforts of not only writers, GM's and statisticians, but the wider fan base as well who eventually turned what was initially a fringe pursuit to part of the mainstream conversation.

To note that a sample size is insufficient or that there are limitations to a statistic in terms of interpretation provides a service to the community. It also has the benefit of creating an environment where intellectually stimulating and challenging debate can grow and thrive, where the perspectives of both those who are mathematically inclined and those who are not can benefit from each other's insights.

And let's be honest, it's not like boards like APBR are hopping with lively conversation. You see someone post a model and data updates about once ever X weeks/months. Interesting for what it is, but hardly sufficient to supply a steady diet of basketball addiction and conversation. :grin:

 

I agree with you for the most part, and I think it does provide a service. I enjoy learning about new things, but again, maybe due to my cynical nature, I just hope you are forewarned that it could be an exercise in futility. If people could all learn so easily (and be that willing to learn), then the world would be a smarter place. I mean, why learn when 80% of the population already know they are smarter than average? :wink:

But again, if you're willing to expend the effort, and it sticks for even one out of one hundred...it could be worth it (depending on how much you value your effort and how much you value the edification of others lol).

 

Wow....this thread is becoming like Maths specialist class, lucky I was good at maths so i could follow most of the postings.

I think at the end of the day, the statistics is there just to predict the future, but no statistics will ever get it 100% right when there is just too many variables and there is a lot of human element involved.

One good example was Linsanity period....no statistics would have been able to predict that. There was just so many random variables (i.e. injuries to key players in NY, NY in a losing skid, Lin getting triple double in Dleague few weeks before, NY lacking in the PG departement etc..) that happened at certain times that led to Linsanity.
And there would have been no one, not even with best minds with the best data (even if they had every single data the moment of Lin the moment he was born) would have been able to predict 2012 Feb in the NBA.

 

Slowly but surely, basketball statistics is becoming less foreign to me.
Really appreciate your posts, Torocan.

 

The stats available at the time clearly showed that Lin was NBA ready. The only thing so pervasive that caused everyone to miss him is obvious: stereotyping. The Knicks poor circumstances opened the door for him. Some things are just written in the stars.

 

What stats are you talking about? He wasn't that good at golden state. Lin got an opportunity and he did very well with it.

 

Brooks opposing eFG should be .355

 

I'm A BOH

LOH

MOH

 

great analysis there but one thing I would like to add is that the eye test is very tricky and not just on a casual observer vs. seasoned technical expert of the game basis. there are many who know about positional defense, heck even eat and sleep Xs and Os, but by some quirk of their gene, would still prefer guys like Harold Miner over Luke Walton or Jamal Crawford over Gordon Hayward, even if stats, results, and Ws all say otherwise.

one person said stereotyping. early in Lin's career, most likely it posed a lot of obstacles. but he is in the NBA now and media has shown what he can do. and i doubt when people say he's a scrub, or he is an average player, that they think of his ethnicity. i believe majority just base it on what they see - a player who does not seem to be as smooth as others when he moves, dribbles, shoots, defends - and thus has less upside or "is at his ceiling".

it takes a very objective person, whether coach, GM or casual observer to take statistics and hard numbers over the eye test

 

Torocan wins the re-match on TKO, by going technical on statistics.

Meh did get a good jab in by pointing out the difference between 7 days and two weeks, thereby (in his opinion) discrediting Torocan's otherwise sound statistics.

 

Sure thing.

Why do you only use 19 data points based on the number of games? There are multiple shots within each game. In an extreme scenario, suppose a game lasts 1000 minutes long and Chris Paul takes 200 shots. Yet by your methodology, he would still count as 1 sample size.

Do you believe the Rockets, when they look at opponent shooting, uses entire games as 1 data size? That is, instead of looking say hypohtetically Chris Paul going 10-20 one game, Blake going 1-5 the other game, and determine that they should count that as 2 games, FG% of .500 and .200?

I would think that would be a very stupid way of analyzing FG%. And a convenient way of limiting sample size.

 

What is that? A game lasts 1000 minutes? The qaulity of this debate is wayyyy down to the toilet.