Taleb's "The Black Swan"

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
Topic Author
Chas
Posts: 846
Joined: Sat Mar 24, 2007 8:41 pm
Location: America

Taleb's "The Black Swan"

Post by Chas »

Thanks to a tip from someone here I ordered and am now reading "The Black Swan" by Nassim Nicholas Taleb. I'm only in the early stages of the book, but I am fascinated with the theme he is developing, i.e. that "earth-shaking" events are all that really matter in our world and that we will surely fail to understand that fact if we live our lives guided only by personal observation. Taleb’s example of this is the life of a young turkey in America which is fed and cared for every day by people. Logically, the turkey soon decides that it is to be its life to spend every day being cared for and protected by benevolent humans. That is, until Thanksgiving week, when something unexpected will happen to the turkey and it will experience a revision of belief!

Ah, yes. Are WE turkeys? Or rather, as we surely would prefer to think, soaring eagles. :(

Chas
Chas | | The course of true love never did run smooth. Shakespeare
User avatar
NAVigator
Posts: 2531
Joined: Tue Feb 27, 2007 7:24 am
Location: Iowa

Post by NAVigator »

In case you missed it, this book has been discussed previously in Nassim Nicholas Taleb and Black Swan Book

Jerry
"I was born with nothing and I have most of it left."
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Post by larryswedroe »

I read it and thought it had lots of great insights. Definitely worth reading
User avatar
tetractys
Posts: 5343
Joined: Sat Mar 17, 2007 3:30 pm
Location: Along the Salish Sea

Re: Taleb's "The Black Swan"

Post by tetractys »

Chas wrote:...something unexpected will happen to the turkey and it will experience a revision of belief
Well to tell you the truth, I've never ran into a turkey that was capable of any kind of belief. Nor a chicken. Maybe a pig or a cow, or a cat or a dog or two, or even a tortoise or bear; but never a turkey.

Turkeys are just plain mindless, even wild turkeys. Even if you could say that some viewpoint or sensation they posses is a belief, no matter what happens to them they certainly do not experience any "revision" of it. Chop of their head, they don't even know it.

This might explain the root of Taleb's viewpoint if he "believes" people are turkeys.

I think I'll check out this book though. It sounds interesting. 8)

Tet
yakers
Posts: 316
Joined: Mon Feb 19, 2007 11:52 pm
Location: Pasadena, CA

Post by yakers »

I read a review in The Economist and I was surprised. I thought they would pan it but they actually said some nice things about the book and the idea. Maybe its good to be reminded that we don't know everything.

http://tinyurl.com/35zj5w
User avatar
Topic Author
Chas
Posts: 846
Joined: Sat Mar 24, 2007 8:41 pm
Location: America

Re: Taleb's "The Black Swan"

Post by Chas »

tetractys wrote:
Chas wrote:...something unexpected will happen to the turkey and it will experience a revision of belief
Well to tell you the truth, I've never ran into a turkey that was capable of any kind of belief. Nor a chicken. Maybe a pig or a cow, or a cat or a dog or two, or even a tortoise or bear; but never a turkey.

Turkeys are just plain mindless, even wild turkeys. Even if you could say that some viewpoint or sensation they posses is a belief, no matter what happens to them they certainly do not experience any "revision" of it. Chop of their head, they don't even know it.

This might explain the root of Taleb's viewpoint if he "believes" people are turkeys.

I think I'll check out this book though. It sounds interesting. 8)

Tet
It is only a metaphor Tet.

Chas
Chas | | The course of true love never did run smooth. Shakespeare
User avatar
tetractys
Posts: 5343
Joined: Sat Mar 17, 2007 3:30 pm
Location: Along the Salish Sea

Re: Taleb's "The Black Swan"

Post by tetractys »

Chas wrote:It is only a metaphor Tet.
As is my answer Chas, in a dry way.

Tet
psteinx
Posts: 4696
Joined: Tue Mar 13, 2007 2:24 pm

Post by psteinx »

I just finished it. FWIW, I skimmed a bit in places because it felt sooooo repetitive.

In general, I wasn't very impressed.

Taleb's arguments basically boil down to "Past history is almost useless as a predictor of the future. Almost nothing follows the bell curve, even when it appears to. Always bet on the surprise rather than something in line with history." And he illustrates along the way with a bunch of examples of varying quality (though generally poor, IMO).

There's a nugget of truthiness here. The past is not a PERFECT predictor of the future. Few things follow the bell curve PERFECTLY. The future is likely to show more variation than the past because the past does not contain the full set of possibilities (i.e. in financial terms, there could very well be future outcomes much worse than the depression.)

But he takes it too far - basically stating that because our mental/historical models are imperfect, they are worthless.

Moreover, he supports his arguments, or tries to, with examples of dubious merit in my mind.

For instance, the title of the book, "The Black Swan" is based on the story of the discovery of the first black swan(s) in Australia a few centuries ago, an event that was supposedly wholly unpredictable (because swans in the known/European world were all white). Now I can't know exactly what was in the mind of the average European back then, but I doubt that this event was as unpredictable as Taleb claims. Given some large percentage of the world that was unexplored (and known to be unexplored), and that previous exploration and communication with distant lands had introduced a wide variety of new animal types and variations on old animal types, I rather doubt that the idea that a new variation of swan that was of a different color could be entirely inconceivable.

It seems a bit like saying that the election of Barack Obama is inconceivable, because we've never had a black (or at least partially black) president before.

He belittles a fictitious individual for responding to this experimenter's question (paraphrased), "I have a fair coin (50/50), and flip it 99 times, all heads. What are the odds of tails on the next flip?" The foolish respondent answers 50/50, instead of a very low number (because the odds of a fair coin coming up heads 99 times in a row are far lower than the odds that the experimenter is lying, mistaken, etc.) I agree with the logic, and frankly think he is setting up a straw man (I doubt very many would give the foolish man's response.)

But then Taleb commits roughly the same mistake himself, taking the limited conditions of his examples a bit too strongly to describe a [weak] proof of his arguments. He tries to illustrate survival bias thusly: Conventional wisdom posits that Russian mobsters are particularly tough, because they have survived the Gulag, which toughened them. But Taleb says the logic is reversed - only the tough survived the Gulag. Even if the Gulag weakened each individual who went through it, the average person emerging from the Gulag is tougher than the population as a whole, because the weak ones who went into the Gulag did not survive.

I find his argument weak - it is possible that there is some aspect of a survivor bias at play with the Gulag, but that does not invalidate the conventional wisdom that being in the Gulag, surrounded by violence and cruelty, toughens one and makes one more suited for the kind of behavior that mobsters exhibit.

In essence, Taleb is quick to belittle conventional wisdom by suggesting alternate theories, but the alternate theories do not disprove conventional wisdom - in all likelihood they supplement it (i.e. Gulag survivors are tough because the weak ones die AND because being in the Gulag hardened them). If he was stating this (conventional models are too simple and other factors are at play), he'd be ok. But he is suggesting that the main premise of the conventional model is flat out wrong, not merely incomplete.

Finally, Taleb name drops heavily, apparently a Who's Who of top philosophers, French intellectuals and the like. (I say 'apparently' because my knowledge in this arena is limited.) Sometimes this is interesting, but on the whole, I find it mostly distracting.

The best part of the book was the only 'hard science' part of the book - a brief review of experiments by psychologists and others of how people make incorrect forecasts and the like, backed by reasonably hard data. Unfortunately, this section is rather short.

The book did make me think about some issues that I do not usually consider - the limits of knowledge and so forth. But overall, I was less impressed than others were by the book as a whole. It felt like a ~300 page rant on a topic better confined to a 20 page magazine article.
Last edited by psteinx on Mon Jun 18, 2007 1:12 pm, edited 5 times in total.
User avatar
ddb
Posts: 5511
Joined: Mon Feb 26, 2007 12:37 pm
Location: American Gardens Building, West 81st St.

Post by ddb »

psteinx:

Excellent review, and I fully agree with everything that you said, though I could have never written it so eloquently and thoroughly!

I also read/skimmed the book at a local bookstore one afternoon. The whole time I was with the book, I kept thinking to myself, "When is he going to get to some new topics?" Definitely more suitable for a magazine article than a 300-page book, as psteinx points out.

- DDB
User avatar
tetractys
Posts: 5343
Joined: Sat Mar 17, 2007 3:30 pm
Location: Along the Salish Sea

Post by tetractys »

Good, bad, or ugly, Teleb's book is pretty popular right now. Just put it on hold yesterday evening and I'm 175th in line at my local library to read it. Won't be long though, because it looks like they have about 25 copies.

Tet
User avatar
nick22
Posts: 859
Joined: Sun Mar 04, 2007 11:00 am
Location: Ohio

Black Swan

Post by nick22 »

It is on my book shelf to still read, but at least Taleb's ideas and stories get me thinking about my risk strategies to see if I have areas where I could better manage downside risk. Larry does this too, but with a more accessible style of writing and prose.
Nick22
User avatar
Topic Author
Chas
Posts: 846
Joined: Sat Mar 24, 2007 8:41 pm
Location: America

Post by Chas »

psteinx wrote:I just finished it. FWIW, I skimmed a bit in places because it felt sooooo repetitive.

In general, I wasn't very impressed.

Taleb's arguments basically boil down to "Past history is almost useless as a predictor of the future. Almost nothing follows the bell curve, even when it appears to. Always bet on the surprise rather than something in line with history." And he illustrates along the way with a bunch of examples of varying quality (though generally poor, IMO).

There's a nugget of truthiness here. The past is not a PERFECT predictor of the future. Few things follow the bell curve PERFECTLY. The future is likely to show more variation than the past because the past does not contain the full set of possibilities (i.e. in financial terms, there could very well be future outcomes much worse than the depression.)

But he takes it too far - basically stating that because our mental/historical models are imperfect, they are worthless.

Moreover, he supports his arguments, or tries to, with examples of dubious merit in my mind.

For instance, the title of the book, "The Black Swan" is based on the story of the discovery of the first black swan(s) in Australia a few centuries ago, an event that was supposedly wholly unpredictable (because swans in the known/European world were all white). Now I can't know exactly what was in the mind of the average European back then, but I doubt that this event was as unpredictable as Taleb claims. Given some large percentage of the world that was unexplored (and known to be unexplored), and that previous exploration and communication with distant lands had introduced a wide variety of new animal types and variations on old animal types, I rather doubt that the idea that a new variation of swan that was of a different color could be entirely inconceivable.

It seems a bit like saying that the election of Barack Obama is inconceivable, because we've never had a black (or at least partially black) president before.

He belittles a fictitious individual for responding to this experimenter's question (paraphrased), "I have a fair coin (50/50), and flip it 99 times, all heads. What are the odds of tails on the next flip?" The foolish respondent answers 50/50, instead of a very low number (because the odds of a fair coin coming up heads 99 times in a row are far lower than the odds that the experimenter is lying, mistaken, etc.) I agree with the logic, and frankly think he is setting up a straw man (I doubt very many would give the foolish man's response.)

But then Taleb commits roughly the same mistake himself, taking the limited conditions of his examples a bit too strongly to describe a [weak] proof of his arguments. He tries to illustrate survival bias thusly: Conventional wisdom posits that Russian mobsters are particularly tough, because they have survived the Gulag, which toughened them. But Taleb says the logic is reversed - only the tough survived the Gulag. Even if the Gulag weakened each individual who went through it, the average person emerging from the Gulag is tougher than the population as a whole, because the weak ones who went into the Gulag did not survive.

I find his argument weak - it is possible that there is some aspect of a survivor bias at play with the Gulag, but that does not invalidate the conventional wisdom that being in the Gulag, surrounded by violence and cruelty, toughens one and makes one more suited for the kind of behavior that mobsters exhibit.

In essence, Taleb is quick to belittle conventional wisdom by suggesting alternate theories, but the alternate theories do not disprove conventional wisdom - in all likelihood they supplement it (i.e. Gulag survivors are tough because the weak ones die AND because being in the Gulag hardened them). If he was stating this (conventional models are too simple and other factors are at play), he'd be ok. But he is suggesting that the main premise of the conventional model is flat out wrong, not merely incomplete.

Finally, Taleb name drops heavily, apparently a Who's Who of top philosophers, French intellectuals and the like. (I say 'apparently' because my knowledge in this arena is limited.) Sometimes this is interesting, but on the whole, I find it mostly distracting.

The best part of the book was the only 'hard science' part of the book - a brief review of experiments by psychologists and others of how people make incorrect forecasts and the like, backed by reasonably hard data. Unfortunately, this section is rather short.

The book did make me think about some issues that I do not usually consider - the limits of knowledge and so forth. But overall, I was less impressed than others were by the book as a whole. It felt like a ~300 page rant on a topic better confined to a 20 page magazine article.
An interesting review of Swan Psteinx. Thanks.

I agree that Swan is 90% attitude BS, but sometimes attitude can be interesting in its own right. I did note Taleb’s statement that the major anomalies in the market, like the 1987 27% loss, absolutely can’t be made to fit inside any normally distributed stats. (I remember when taking “practical” engineering statistical analysis courses we were told to throw out lab measurements outside the 1-in-20 range as likely due to anomalous error.) Anyway if Taleb is right about that, and I assume he is, then, all our fidgeting over standard deviation and portfolio mix returns might be of little significance if the market just turns upside down periodically; wouldn’t you think? I don’t quite know what to make of it myself so I hope the authors can explain it all to me. That is probably foolishness on my part, to think that I can plunk down $30 or so and get all the answers. :cry:

Anyway, I have laid the Swan aside to take up Mandabrot’s “The (Mis) Behavior of Markets” and I am in the early stages of it now. Hope springs eternal....

Chas
Chas | | The course of true love never did run smooth. Shakespeare
psteinx
Posts: 4696
Joined: Tue Mar 13, 2007 2:24 pm

Post by psteinx »

The fact that events like the October 19, 1987 market drop are extraordinarily unlikely under a normal curve (and therefore, stock market returns are almost certainly not 'normal') is well known, and has been for a long time (even before that date, I'd venture, though I was too young to be paying much attention).

That doesn't render SD or other conventional risk measures moot, it just means that they are incomplete and we should regard them with caution.

An asset class with an SD of 10 is still riskier than one with an SD of 5. But the incidence of very bad returns is likely higher than the normal curve would suggest (i.e. fat tails).

Moreover, just as that one day in 1987 was far outside of the experience of the US equity market, it is entirely possible that we could at some time have a one year stretch, or a five year stretch, far outside of our past experiences.
User avatar
Topic Author
Chas
Posts: 846
Joined: Sat Mar 24, 2007 8:41 pm
Location: America

Market risk...

Post by Chas »

psteinx wrote: An asset class with an SD of 10 is still riskier than one with an SD of 5. But the incidence of very bad returns is likely higher than the normal curve would suggest (i.e. fat tails).
Well, let me try to put that into Taleb’s perspective as I understand it. How would the 5% SD fund have performed compared to the 10% SD fund on Oct. 19, 1987? Aren’t all turkeys just as dead come Thanksgiving?

BTW, I think Taleb said the 1987 drop was 1 in a trillion+ or so odds if the market were normally distributed. That is indeed a very, very, FAT tail! The problem is that this drop is factored into the SDs we use just as though it was only 1000 to 1 or so; right?

Chas
Chas | | The course of true love never did run smooth. Shakespeare
User avatar
AzRunner
Posts: 999
Joined: Mon Feb 19, 2007 6:18 pm
Location: Phoenix

Post by AzRunner »

Chas
Anyway if Taleb is right about that, and I assume he is, then, all our fidgeting over standard deviation and portfolio mix returns might be of little significance if the market just turns upside down periodically; wouldn’t you think?
That's basically my thinking and I am an engineer by training. Personal finance is not the same as engineering. These numbers and past market performance are interesting but the market looks forward not backwards. I think you need to take a lot of what you see and read with a grain of salt.

That's why a widely diversified portfolio works for the vast majority. You can basically say, I don't know what's going to do well going forward, so I'm going to own a wide portfolio of stocks and bonds.

Norm
User avatar
Topic Author
Chas
Posts: 846
Joined: Sat Mar 24, 2007 8:41 pm
Location: America

Market fluctuations

Post by Chas »

AzRunner wrote:[That's basically my thinking and I am an engineer by training. Personal finance is not the same as engineering. These numbers and past market performance are interesting but the market looks forward not backwards. I think you need to take a lot of what you see and read with a grain of salt.

That's why a widely diversified portfolio works for the vast majority. You can basically say, I don't know what's going to do well going forward, so I'm going to own a wide portfolio of stocks and bonds.

Norm
Norm,

No argument there, but I have just thought of something that makes me wonder about Taleb’s Swan book and my own arguments here. Taleb said in Swan that the 27% market drop on Oct. 27, 1987 was wildly out of line with a “normal” curve distribution. However, most of us are used to looking at a SD that is based on annual market returns, where daily returns are much more variable. So, is the 27% drop on a day, or on a week, really an enormous anomaly?
Chas | | The course of true love never did run smooth. Shakespeare
User avatar
stratton
Posts: 11083
Joined: Sun Mar 04, 2007 5:05 pm
Location: Puget Sound

Post by stratton »

No argument there, but I have just thought of something that makes me wonder about Taleb’s Swan book and my own arguments here. Taleb said in Swan that the 27% market drop on Oct. 27, 1987 was wildly out of line with a “normal” curve distribution. However, most of us are used to looking at a SD that is based on annual market returns, where daily returns are much more variable. So, is the 27% drop on a day, or on a week, really an enormous anomaly?
Very extreme short term volatility is much more prevalent than you think. In a paper posted here or on the M* forum they looked at short term volatility within a single trading day. Every few minutes and even seconds if they could get the data. Surprisingly they found volatility as high as 29 standard deviations within market data when measured every one or two seconds.

Paul
User avatar
LH
Posts: 5490
Joined: Wed Mar 14, 2007 2:54 am

Post by LH »

I suggest anyone who is interested in what Black Swan had to say, consider reading mandlebrots (mis)behavoir of markets.

Its a nice, interesting, book about fractals and probability, written in an entertaining, basically math free way. It is a large part of the underpining of what black swan talks about.

Taleb does seem to want to have it both ways a lot with his arguments in black swan. A lot of are mixed with belittling ad hominem type attitude/attacks, with a nice dollap of name droping..... Anytime that sort of stuff is bandied about one has to wonder what the strength of the underlying arguments are. It usually hides something that if just plainly set out, is either obvious/trivial, or intellectually weak in some fashion. It certainly makes his arguments more obtuse.

I also wondered, as an aside, in a half serious sort of way, if a potential black swan for his way of trading, a period of prolonged "normality", a bell curve with no significant fat tails over 10-20 years? Lack of "randomness" would be pretty random and unexpected for him perhaps? I wonder what the probability of that is, compared to the probability of his nice large fat tails?

Black swan, and fooled by randomness, his other book, are both worth reading imo.
User avatar
LH
Posts: 5490
Joined: Wed Mar 14, 2007 2:54 am

Post by LH »

Surprisingly they found volatility as high as 29 standard deviations within market data when measured every one or two seconds.

What kind of data is that? Whats the scale and such? Is that like the dow going from
13,590.31
13,590.32
13,590.31
13,690.31
13,590.31
or something akin to that? The variance is miniscule and one value has a huge percentage change relative to the variance even though its a small percentage change relative to the dow number itself?

I guess when looked at from that perspective, when one has basically "quantum jumps" or large discontinuity in stock prices (huge instantaneous changes), the deviations can be huge on a small time scale, yet if look at on a different scale, the deviation would not be so big?

A 29 sigma jump is pretty big, do they happen often? If you look at it by hourly data, and still include the jump, is it still a 29 sigma jump?
User avatar
RiskAverse
Posts: 205
Joined: Wed Jun 20, 2007 1:41 pm

Post by RiskAverse »

LH wrote: Taleb does seem to want to have it both ways a lot with his arguments in black swan. A lot of are mixed with belittling ad hominem type attitude/attacks, with a nice dollap of name droping..... Anytime that sort of stuff is bandied about one has to wonder what the strength of the underlying arguments are. It usually hides something that if just plainly set out, is either obvious/trivial, or intellectually weak in some fashion. It certainly makes his arguments more obtuse.

I also wondered, as an aside, in a half serious sort of way, if a potential black swan for his way of trading, a period of prolonged "normality", a bell curve with no significant fat tails over 10-20 years? Lack of "randomness" would be pretty random and unexpected for him perhaps? I wonder what the probability of that is, compared to the probability of his nice large fat tails?

Black swan, and fooled by randomness, his other book, are both worth reading imo.
I thought Fooled by Randomness was the stronger book. Taleb's attitude/cockiness/arrogance got a bit out of hand in the second book. But alot of hedgefund people are like that, especially if they think they are beyond psychology as a result of knowing about behavioral finance.

He should have hired Malcom Gladwell to edit "Black Swan,", since he wrote a good article about NNT for the New Yorker "Blowing Up"
User avatar
Topic Author
Chas
Posts: 846
Joined: Sat Mar 24, 2007 8:41 pm
Location: America

Black Swans

Post by Chas »

stratton wrote: Very extreme short term volatility is much more prevalent than you think. In a paper posted here or on the M* forum they looked at short term volatility within a single trading day. Every few minutes and even seconds if they could get the data. Surprisingly they found volatility as high as 29 standard deviations within market data when measured every one or two seconds.

Paul
Paul,
I don't know that second-to-second volatility matters much to fund investors like us. What we notice, unless we plug our ears and wear blinders, is day-to-day change. I know, I know, buy and hold is the name of the game, but if only for academic and emotional reasons, curiosity exists about aspects of the market. Which brings me back again to wondering exactly how well, or badly, do daily variations fit into the “normal” bell curve? The answer to that might refute Taleb’s basis for the Swan book, though I seriously doubt that such an error could be made in a book such as his.

Chas
Chas | | The course of true love never did run smooth. Shakespeare
User avatar
LH
Posts: 5490
Joined: Wed Mar 14, 2007 2:54 am

Post by LH »

Chas,

Daily stock prices graphed out do not follow a bell curve distribution well at all. The point he makes, as well as mandlebrot, is that there are multiple market pricing events that by a bell curve would basically hardly ever happen, but in reality happen relatively routinely.

Image

There is the fat tail of the DJIA daily returns versus a normal curve from a google image search.

LH
Warner
Posts: 184
Joined: Thu Apr 19, 2007 2:35 pm

Post by Warner »

Taleb is this week's guest on BBC's "The Interview". It is about 25 minutes long. BBC streams the show on their web site. Unfortunately, I can't post the link:

In order to try to prevent spammers, we do not allow our members to post links or images until they have posted at least 5 legitimate posts and have been with us for more than 7 days. We appreciate your understanding in this matter in order to help us eliminate spam from this forum. If you have somehow gotten this message even though you meet both of the criteria, please email the Site Admin at bhadmin (at) diehards (period) org

Thanks!
User avatar
gatorking
Posts: 1267
Joined: Tue Feb 20, 2007 6:15 pm
Location: MA

Here's the link

Post by gatorking »

Post Reply