RWCG


Which Taleb output isn’t a hoax?
December 5, 2012, 7:48 am
Filed under: Uncategorized

I’d say that what looks to have been a weird, unfunny attempt at a Sokal-like hoax has lowered my opinion of Taleb, but that’s not really possible. Not that I’ve ever read Taleb or have substantial basis for a negative opinion of him per se, but as best as I’ve been able to gather, his entire claim to fame and genius consists of continually yelling at the world that Not All Distributions Are Gaussian, thereby bravely correcting the erroneous belief to the contrary that had been held by, approximately, none people.

Feel free to correct me if I’m wrong there, but assuming I’m not, doesn’t that entire schtick pretty much have to be a hoax? Because if not, that’s sad.

About these ads

12 Comments so far
Leave a comment

Can you explain “Not All Distributions Are Gaussian, thereby bravely correcting the erroneous belief to the contrary that had been held by, approximately, none people.” to someone who’s only through Stat 110 so far?

Comment by tangentstyle

Gaussian distributions refer to the bell-curve distributions you learn in your first stats class. This is the most common type of statistical analysis, used by most people in most setting to analyze most things. An alternative method of analyzing statistical probability is to develop Fractal/Madelbrotian distributions (which might also be referred to as geometric distributions, but I can’t remember).

The main difference between the two is largely categorical. You would use Gaussian distributions for things of limited scalability (e.g. human height or feline weight) because you expect there to be reasonable thresholds and ceilings on these things (for example, you don’t generally expect to meet humans that are eighteen inches tall, or cats that weigh three tons). You would use Mandelbrotian distributions for things of infinite, or near infinite scalability, like natural disasters or economic disruptions, since there is no telling the frequency or severity of these things. You may know, in a sense, that there are limits to the damage they can do, but there are too many variables to account for when, for example, trying to estimate the frequency and scale of damage.

Comment by Simon Grey

He asserts (wrongly) that all of finance makes the assumption that events behave roughly Gaussian-like; i.e. 95% of the time you’ll stay within 2 standard deviations, etc. Then he says that Reality Isn’t Really Like That, as if this is news.

Then he collects fat royalty checks from his pop-finance books.

I think, anyway. As I said, I haven’t read him (other than the occasional column). This is what has trickled down to me as Taleb’s insights. But again, if I’m wrong someone should let me know.

Comment by Sonic Charmer

@Sonic Charmer- Taleb’s assertion wasn’t that people weren’t aware of non-Gaussian distributions. For fuck’s sake, his book The Black Swan was partially addressed to quants who probably knew far more math than he did. His complaint was that quants were making category errors in their analysis, which was why the market kept blowing up in their face and they kept being surprised. Their risk analysis simply did not match reality.

Now, he’s a bit of a blowhard about this, and appears to think that he’s discovered this wonderful theory of everything, but his book was well-timed and made some quite valid points, his personal douchiness aside.

Additionally, you’d probably agree with a good number of the criticisms he makes of quants and of scientists, who have apparently been educated beyond their intelligence, for blindly trusting their analytical models without giving much thought to the assumptions governing them. It’s basically the inverse of your recent tirade against neo-cons bitching about Nate Silver’s stats. They didn’t like the outcome so they criticized Silver while never bothering to understand his model and what made it tick. Likewise, a lot of quants and even scientists will implicitly trust their statistical analysis because they like the results, and never bother to question the assumptions upon which their analysis is predicated. Basically, Taleb was criticizing data guys (and especially economists) for being to trusting of their implicit assumptions and the theories predicated thereupon.

Comment by Simon Grey

But ‘risk analysis did not match reality’ is not an insight, it’s a restatement of 2007-08’s news headlines.

I think it’s a misdiagnosis to suggest that quants ‘believed in’ their models. I think there are certainly times where for various reasons an institution can end up *acting as if* it ‘believes in’ a model, without anyone involved – particularly the quant who built it – quite doing so. If that’s his criticism, then ok.

Comment by Sonic Charmer

The thing is, he wrote the book before the bubble popped, and finished it about the time the bubble popped. Maybe his timing was fortuitous, maybe he had an insider’s account of the matter, maybe no one really believed their risk assessment models. But as you implied, they acted as if they did believe the BS they were peddling. Additionally, his work wasn’t with merely saying X is the case, but also with explaining why X happened to be the case so that everyone could learn from the mess and make completely new set of mistakes the next time around.

Beyond that, Taleb’s work dealt with epistemology and over-reliance on “science” (or, more accurately, scientific consensus). I’d advise you to read his work before criticizing it with by oversimplifying one of the points that he made.

Comment by Simon Grey

Read the work first before criticize = a totally fair suggestion :-)

Comment by Sonic Charmer

The bubble pop of ’08 really had less to do with models than simple greed. Ratings agencies, banks, and hedge funds effectively conspired (although mostly not intentionally) to create a serious of investments that looked too good to be true to the investors backed by mortgages that were too good to be true for the home buyers.

Thus, in a very simple, very human way, the sins of greed and pride combined to create massive failure. Most of those responsibile still retain positions of power in banks, corporations, and the government. Almost none have been held responsible for the fraud that was perpetuated.

Comment by Dave

Sure, but where that links up with risk/modeling is that in theory those models were supposed to inhibit the ability of the ‘greedy’ to do all that. The reason they didn’t has less to do with people believing in models than with people intentionally gaming them.

Comment by Sonic Charmer

Taleb provided the perfect excuse right when a really good excuse was necessary.

What actually happened?

A giant socialist ponzi scheme to pretend that unreliable people (specifically unreliable people of color – that’s important because of what it does to people’s thought processes) were actually credit worthy. Houses were bought and built – making someone money and paid for by loans. Loans were made to a bunch of people with zero hope of ever repaying them and then repackaged and sold as AAA assets. Why were they sold as AAA assets? Because banks need AAA assets. Because it’s illegal to think that a strawberry picker from Mexico isn’t just as good of a credit risk as a white electrician. Because the credit models said that mortgage defaults were uncorrelated events and so the central limit theorem makes the default rate predictable.

Those AAA assets were then sold and being AAA assets banks bought them by the billions. Why? Because banks have an insatiable hunger for AAA assets that provide returns. On top of that a bunch of those assets were AAA because they were backed by Fannie and Freddie.

Banks didn’t want to hear that the assets weren’t AAA. Lenders didn’t want to know that the loans wouldn’t get repaid. Home builders were happy to have the business. Real estate agents, mortgage brokers and house appraisers were happy to make money.

Taleb then comes along and says “those stupid modelers don’t know what’s correlated and what’s not and if everything is really correlated then you get massive instability when a rare event happens”. Meanwhile the only reason the “stupid modelers” ever made those assumptions is because it’s illegal to notice the reasons that those AAA loans really aren’t AAA.

Why does Taleb become famous and popular? His excuse permits people smart enough to figure out his writing (most Smart People) to have a mental reason for why the market collapsed without thinking any forbidden thoughts.

Forbidden thoughts like why that market was always going to collapse. Forbidden thoughts like “if the collapse was actually because the whole market was a socialist ponzi scheme and none of the laws and policies behind it have changed why won’t the same thing happen again?”

Such a relief not having to think about that.

Comment by Steve Johnson

“it’s illegal to think that a strawberry picker from Mexico isn’t just as good of a credit risk as a white electrician”

Where does that come from? Doesn’t a person’s creditworthiness have everything to do with their income?

Doesn’t a bank have the right to notice that, “Oh this strawberry picker makes a lot less than this electrician. Maybe we should think twice about giving him a loan of equal size, or any loan at all”

Comment by tangentstyle

If the quants came up with models that threw a wet blanket on the party, those quants would have been fired and new ones who would be hired.

Or they would have been ignored and marginalized.

Comment by Dave




Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s



Follow

Get every new post delivered to your Inbox.

Join 455 other followers

%d bloggers like this: