Fooled by Randomness

[RussianMike]

Just finished reading the book last week, will start re-reading again. I think this is a great book. To me the book emphasizes that one must think and this is something that is not practiced enough, for sure.


<iframe src="http://rcm.amazon.com/e/cm?t=quantfinaneng-20&o=1&p=8&l=as1&asins=0812975219&fc1=000000&IS2=1&lt1=_top&lc1=0000FF&bc1=FFFFFF&bg1=F7F7F7&f=ifr" style="width:120px;height:240px;" scrolling="no" marginwidth="0" marginheight="0" frameborder="0"></iframe>

 

[davethemovie]

Fooled by Pompousness

Fooled by Pompousness: Why I don't like Taleb's book

The book is an indictment, followed by an acquittal, followed by the author committing the very same crime.

Specifically, it begins with the indictment that traders / money managers ignore outliers and foolishly generalize from past experience.

Then we come to the acquittal, where concepts such as Hume's riddle of induction* are used to suggest that it's part of the human condition to try to generalize from the past and fail.

Lastly, the author boasts of sitting back and profiting by betting on the outliers which others have been too myopic to account for. In doing this, he commits the same crime of generalizing from the past.

Betting on outliers is not a dichotomous issue; in practice it's a matter of degree.
Examples of questions that demonstrate this are:

How far out of the money do you want to straddle the market?
How far would the market move when the next 'outlier' occurs?
What is the cost if the next 'outlier' doesn't occur within a certain amount of time?
How much should be bet on volatility?

To answer these questions requires data and decision making that borders on total subjectivity. There is nothing in Taleb's book which demonstrates that he, in practice, transcends the limitations of those he criticizes.

The above is just an artful way to tour the book and critique it; but if this is not persuasive or just not relevant to the quality of the book, then here is the real problem, stated plainly:

Taleb does not justify his contempt for the common man. In fact, at the end of the book, you are told, and I mean YOU, the reader, that you are probably part of the problem.

If I want to experience the limitations of my human epistemic condition, I'll get a book on optical illusions; it won't insult me and also note the irony that I have purchased a book that is insulting me. As far as the revelation that outliers are a methodological problem... that's not much of a revelation.

Fortunately for us, Joe MBA and Joe MFE (and Joanne too), there are more popular books such as Surowiecki's "The Wisdom of Crowds" which popularizes research demonstrating that a cooperative group outperforms its most intelligent member under the condition that all members are independently informed. In contradiction to the world view Taleb presents, regular people make up industry by way of effort and cooperation. There's plenty of room in the back office for abrasive geniuses.

DaveTheMovie

*The riddle of induction asks 'what is the justification for generalizing from past experience?'. The riddle becomes challenging the moment you note that you can't use past experience as a justification because that would be circular. A philosopher named Nelson Goodman argues that if we assume that some previous events are legitimately generalized from and others are not; then the "riddle of induction" can be reduced to "the new riddle of induction" where we ask 'which events are valid to generalize from and which are not?'.

 

[RussianMike]

I dont know, if they guy currently makes $ and if he will not blow up it sounds good to me. I am not claiming I understand the book completely, the bottom line that I get from books like these is that u must think, the book is interesting.

 

[DominiConnor]

NNT is like that in like that in real life. He's not a consensus man, he thinks the quality of truth is basically a function of how hard you've tried to kick holes in it.
When arguing with him, the third bottle of wine brought the insight that this was a like one of those exercise you do in martial arts where everyone else in your group runs at you in turn for you to sharpen up your skills, and pinpoint defects in your stance and technique.

You are supposed to be annoyed by his books, if you're not, you don't understand them.

Occasionally we get him to torture entry level candidates we think show exceptional promise, most amusingly the next victim is a straight engineer who although getting through the screen has exactly 17 days reading knowledge of finance. So when I told him that that NNT was having coffee with him on moday, he showed no fear, or even recogniton.
We ought to film this one...

 

[RussianMike]

My partial understanding was that we do not read meaningful clues from history, this causes us to trade incorrectly. I believe you cannot predict the future from history unless you can state with some degree of certainty that one event causes another (I'm not knowledgeable enough in probability but I believe causality is tough to prove). For bonds we know if yields go up prices will go down and vice-versa. It is a function of the formula we use to value the cashflows of bonds. The question is are we using the right parameters to value the cashflows and who is the judge of this correctedness.



Instead I would prefer to look at it from a knowledgeable series of what-if's. For example when last week a plane flew into a building in NYC bonds started going down, so did stocks(didnt look at anything else at that point, so I dont know what happened to gold). Then around 3:10 when it was reasonable to assume that it wasnt terrorism bonds started going back up. It was just interesting to watch, but it would have been more interesting and profitable to make $ on it. I realize this is not what Nassim Nicholas Taleb does because I take it he bets on extreme happpenings before they happen, not while they are happening. What if another plane/helicopter flew into a building. What would happen to stocks/bonds? They would probably go down. Same as this time? It depends on how quickly we see that it is terrorism or not as that seems to be the main concern. I prbbably sound like I'm rambling on and on about things that interest me but this is my respite after taking the programming tests.

 

[pauljack]

Fooled by randomness.

I read this book and liked it immensely. If you read "Trend Following" by Michael Covel, "Moneyball" by Michael Lewis
and the sections in "Market Wixards that deal with Trend Followers, Ed Seykota, Richard Dennis, and particularly Larry
Hite. , You will understand the basic Philosophy of What is being said.
It seems to me Probability is as much a philosophy as a Mathematical tool. Things happen you just cannot predict.
You have to respond other than predict, and try to understand the probability of your response being successful, given the
parameters of the situation. (the concept of a zero sum game).
Jonathan Hoenig summed this up perfectly a couple of weeks ago on "Cashing In" (a Business show on Saturday Morning on
Fox). All the panelists were debating and arguing about the direction of the economy to make their stock predictions,
and he simply said "I'm not an economist, I'm a money manager", and that was that.

PaulJack

 

[joe_bradley]

I also liked the book, although let me point out that this is a "mass market" text, and this give nothing but the broadest overview of Mr. Taleb's trading strategy. I don't know if it was necessarily supposed to impart particular financial knowledge. Rather, it's just a very long essay series on the ways people tend to misjudge probabilities, risk and reward. The discussions he gives of the financial world mostly seem like anecdotes he gives to prove a point. As such, it is half-math, half-psychology. It's actually fairly "light" reading, and very enjoyable at that. So I really quite recommend it.

I think he may rub some people in the finance world the wrong way by claiming that most successful traders are lucky rather than good and claiming that people who follow day-to-day movement of stock indices are wasting their time. But as I come from an engineering background rather than a finance background, that doesn't offend me at all ;-)

As far as some of the criticism that his investment strategy commits the same crime as what he criticizes others for, that would seem at the surface to be true, and this is a point worth investigating. As he describes it, his strategy seems like a simple arbitrage play - make investments that pay off big in the case of unexpected events - because people undervalue the probability of these events. And one would think that over time, like any other arbitrage strategy, it would start to lose usefulness. That is, if it works for him, then other people will recognize that, the market will correct the prices, and there will no longer be an arbitrage opportunity. However, as I understand it, based on the other points he makes in the book, he's betting this won't happen. Why? Because any trader following this strategy will compile a steady stream of losses, with an occasional big payout. Because of the way humans experience risk and reward, actually carrying out such a trading strategy is psychologically very difficult. (And maybe even more difficult to justify to your backers.)

Whether this is true or not I don't know. Without evidence of how his fund is doing, I suspect your opinion on this has more to do with one's beliefs in efficient markets and the like than anything else.

Joe

 

[Bastian Gross]

Why Value at Risk?

I agree with DaveTheMovie on Talebs (and Mandelbrots) arrogance of "know-it-all (but-give-no-solution".

But one point also I agree Taleb: To hold that Value at Risk is charlatanism and a dangerously misleading tool in risk-measurement.
In 1994 VaR was a evolutionary idea, but now in the 21th Century at awareness of coherent risk measurement VaR is unsophisticated. Because VaR does not give any information about the severity of loss by which it is exceeded.

To further reading:

1. Coherent Measures of Risk by Arztner, Delbean, et al. (paper in 1998)
2. Value at Risk: The New Benchmark for Managing Financial Risk by Jorion

That's getting me asking:
Why is not VaR-measurement replaced by a coherent measurement (like CVaR (conditional Value at risk)/Expected shortfall)?
This could save al lot of money!
To rate risks (or chances), wichever are far-reaching and unreckonable, I've in mind to arrange (at a price) some "Think-Tanks".