Arnold Kling  

More on Taleb

PRINT
Jolie on Pregnancy: Do You Bel... Public Opinion About Fetal Tes...

From the London Times


the good investment strategy is to put 90% of your money in the safest possible government securities and the remaining 10% in a large number of high-risk ventures. This insulates you from bad black swans and exposes you to the possibility of good ones. Your smallest investment could go “convex” – explode – and make you rich. High-tech companies are the best. The downside risk is low if you get in at the start and the upside very high. Banks are the worst – all the risk is downside. Don’t be tempted to play the stock market – “If people knew the risks they’d never invest.”

My opinion would be that most of the people who go with Taleb's approach will have a negative return on their investments. Typically, you lose everything on your 10 % flyers, and you make less than 10 percent return on your safe investments. Ergo, you lose. Of course, a few people win. Their flyers turn out to include Google before it went public, or somesuch.

I just finished The Black Swan. Toward the end, Taleb goes on a long rant, in which "Gaussian" becomes an epithet, which he flings at finance Nobel laureates Markowitz, Sharpe, Scholes, and Merton.

I want to ask Taleb a simple question about the non-normality of stock price changes. Why?

Either the "news" that affects stock prices is not normally distributed, or prices react irrationally to news, or both. Taleb might want to argue that "news" is not normally distributed. But what is the mechanism for this? I understand the mechanisms that make book sales or academic reputation or wealth have highly non-normal distributions. Basically, a small advantage can cumulate over time. But "news," in the financial market sense of the term, does not cumulate over time. If you know something is going to cumulate over time, then it is not news!

Moreover, the classic example of October, 1987, does not appear to reflect news. It seems, to me anyway, to be more of a case of market prices going off independently of news.

The recent run-up in oil prices represents a similar puzzle. I think that it's difficult to tell a story for the rise in crude prices for the last six months that is based on the rational digestion of news. Either six months ago folks were overly optimistic about long-term supply and demand conditions or now they are overly pessimistic about those conditions. I don't think that what changed in the last six months was the news about supply and demand.

If the Black Swans come from the skewed distribution of returns, then investing in some highly uncertain projects makes sense (until enough people start to do it that the expected returns get driven way down). However, if the Black Swans come from bubbles developing and popping, then there is no great opportunity in trying to bet on Black Swans. So I think the question of why stock price changes are non-normal is a key question.

Overall, the book was a disappointment. As I feared, I agreed with much of it, and from what I agreed with I learned relatively little.


Comments and Sharing





TRACKBACKS (3 to date)
TrackBack URL: http://econlog.econlib.org/mt/mt-tb.cgi/850
The author at Jim's Blog in a related article titled Rational Oil Prices writes:
    It seems that Arnold Kling has been in Afghan cave during the past six months. He does not seem to think that there has been any bad news on oil supply over the last six months. ... [Tracked on June 12, 2008 7:57 PM]
COMMENTS (11 to date)
Dog of Justice writes:

Normal distributions arise from the sum of many relatively small, independent random variables. In practice, most of the time the stock market can be approximated in this manner, but every once in a while something happens that causes independence to spectacularly break down, such as the Russian default that brought down LTCM. As Taleb points out, the magnitude of such events is better described by a power law than a Gaussian.

He is too nihilistic, though. As long as your downside risk is bounded during a calamity (and for better or worse, mostly worse, I don't see hedge fund employees begging on the street; there are too many mechanisms that allow them to push all the worst downside risk on taxpayers*), there is positive utility in using tuned Gaussian-based models.

*: Consider a simple game where you flip a fair coin twenty times; if you flip heads at least once, you win $1, while if you don't, you lose $2000000. It's easy to work out that the expected value of playing this game is negative. However, if others will partially bail you out when you lose, suddenly it may be worth playing. I fear that variants of this game have been played too often lately.

D. haeck writes:

I think the most interesting literature on this topic is to be found in the field of econophysics. It explains the power law in the distribution of returns with interacting traders. There are striking similiarities in the data with interacting particles models out of physics. Criticall mass, phase transistion etc.

eccdogg writes:

The thing is almost no one assumes a normal distribution when pricing options. So Taleb is railing against a belief that no one believes in practice.

That is why you observer a volatility (smile,frown, smirk) in different markets. Out of the money options particularly way out of the money options trade at a much higher implied volatility than at the money options.

Traders have a saying for this "Sell a teenie loose your weenie" traders will never sell an out of the money option cheap.

I have read both Fooled by Randomness and Black Swan and have been mildly disapointed by both.

I am currently reading Taleb's first book "Dynamic Hedging" which is muchmore technical, but far better that the first two.

Alex J. writes:

Investor psychology can build up over time, via herding, conformity, jealousy and short-sightedness. During a run up, people hear of the gains being made and join the crowd. During a crash, everyone runs for the exits at the same time. That is, you head for the exit, and I see you leaving and follow you etc.

Another way of thinking about it is the long-term or collectively irrational digestion of news about other investors' behavior. The other investors might themselves be behaving irrationally.

Patrick Molloy writes:

Taleb’s investment strategy appears to be very similar to that of Boston University’s Zvi Bodie, author of Worry-Free Investing. Here’s an excerpt from a recent interview:

What investment strategy do you advise?
I never recommend the stock market. I say, load up on Series I bonds in taxable accounts (because taxes are deferred until cashed in) and make sure you have plenty of TIPS in your retirement account (to steer clear of paying taxes on accrued interest). Then no matter how wild the markets get, you don't have to worry.

But with this strategy you give up the chance of making money in stocks.
If there's a fat tail event on the upside, buy an out-of-the-money call option on a market index. [A call is a bet that the market will rise. Calls are "out of the money" when the strike price is greater than the market price of the underlying security.] It has a low probability of paying off, but if it does, you'll make a killing. I implement this in my own retirement portfolio. I have about 95% in TIPS and 5% in out-of-the-money call options that go out as long as three years on the market.

http://www.businessweek.com/magazine/content/07_37/b4049090.htm

eccdogg writes:

I don't think the Taleb/Bodie strategy is a bad one.

I just don't think it makes above normal risk adjusted returns.

The strategy is tailored for someone with a very specific utility function. A person who is very afraid of losses but still wants the oportunity to make money if the everyone is making lots of it. To accoplish this goal they are willing to lose compared to the S&P 500 in periods where the market slowly and steadily out performs treasuries. In that case they probably lose money on thier out of the money calls and the market outperforms thier T-bill position.

In other words they are long volatilty and if the market is not more volatile than the implied volatility on their calls they are net losers.

fundamentalist writes:

Does Taleb not know about inflation? If you invest 90% in government securities, inflation will kill you and I doubt that the average success from the other 10% will sufficient to overcome that loss.

Brad Hutchings writes:

The striking thing to me about most of the criticism Taleb gets is that it's all focussed on his trading advice. He wrote one book about trading and two that move much more generally philosophical in direction. In The Black Swan, he brings in some tales of the market as an anchor to his story, but they are not all of it.

In the London Times story, it's pretty clear that he's on a path of being the next Tom Peters, but with a more overtly philosophical approach and message. I find it a little weird that saving $10 on a pricing error or missing a flight is a "Black Swan" in his story for the masses, but I'd guess he's joking and the author doesn't convey the humor.

The open question on Taleb is whether he has a novel wide contribution to make to how people think about their lives and careers or whether he's repackaged "Who Move My Cheese" for the dumbed down masses. His trading strategies seem, at best, a side note.

8 writes:

People experienced in the market are not suprised by crashes, mathematicians are. Stock price movement is often "normal" to market participants, but not to mathematicians.

eccdogg writes:

"The striking thing to me about most of the criticism Taleb gets is that it's all focussed on his trading advice. He wrote one book about trading and two that move much more generally philosophical in direction. In The Black Swan, he brings in some tales of the market as an anchor to his story, but they are not all of it."


In his "trading" book he does not advocate the 90% in bonds and ten percent in options on flier
strategy. At least I have not seen it through the first couple 100 pages. The book is for practitiones mananging a portfolio of vanilla and expotic options and it is very good.

He advocates the 90% safe 10% fliers in his latter two books and argues that you should employ a long volatility strategy (you win when crazy stuff good or bad happens) in markets and in life. The question is when is insurance against crazy stuff overprice and how do you know when it is?

Sometimes Taleb seems to be arguing that crazy stuff is always over estimated or overpriced, but other times he aregues that people are irrationally afraid of some risk. How do we know which is which? Especially in light of the fact that Taleb tells us that we can't trust past history to be a good guide for the future.

I appreciate Taleb's skepticism but I don't think he is skeptical enough of his ove ideas.

David Shor writes:

Basically, it has to do with the central limit theorem. A bunch of random variables summed up will approximate a normal distribution(They usually don't need to be Independent or uncorrelated, unlike what the first poster said).

But in order for the central limit theorem to hold, you need the distributions to have a finite variance. Otherwise, convergence will go to a more general levy distribution.

Teleb was not the person to invent this, Mandelbrot showed a while ago that futures market prices were not normal.

The problem, is that levy distributions are extremely difficult to estimate, and shift over time, making them pretty much useless to finance. Normal approximations work well enough.

Comments for this entry have been closed
Return to top