This options trader and philosopher bets against the crowd. He explains why traders continue to underestimate the role of randomness in the markets and how betting on the possibility of rare events provides an edge.
Nassim Nicholas Taleb has traded options for more than 20 years, either for major investment banks or on his own as a fund manager and pit trader — but he bristles at being labeled a trader.
Taleb describes himself as a philosopher who argues that traders underestimate the role of randomness, and fund managers’ successful track records might be more easily explained by luck than skill or experience.
Unlike many options traders who collect premium from selling out-of-the-money (OTM) options, Taleb focuses on buying far OTM options that have a low probability of making money but post extraordinary returns if they do.
Instead of studying how the market did move, Taleb analyzes all the scenarios that might have occurred as well. Wild market spikes and crashes are admittedly rare events, but their reward is so large it offsets their infrequency.
“The likelihood is completely irrelevant compared to the payoff,” Taleb says.
Taleb first realized that long OTM options can produce unusual returns after he began trading currency options for Banque Indosuez in late 1984. When leaders of the G-5 countries (U.S., Japan, West Germany, France, Britain) agreed to devalue the U.S. dollar by signing the Plaza Accord on Sept. 22, 1985, he held long OTM options that surged in value, which caught his bosses off-guard.
“They said it’s impossible for such a small position to generate so much profit,” Taleb says. “So I learned two things — the market could deliver these crazy events, and people didn’t understand these payoffs.”
Taleb has spent the past two decades studying the characteristics of options, but it was his windfall following the 1987 stock market crash that put him on the map. His long position in frontmonth eurodollar options gained 67,000 percent as the market gapped up 370 basis points on Oct. 20, the day after the crash.
Despite this success, Taleb had a love-hate relationship with the markets, so he decided to become a scholar and research the science of uncertainty. In addition to holding top trading positions at Credit Suisse First Boston, Union Bank of Switzerland, CBIC-Wood Gundy, Bankers Trust, and BNPParibas, Taleb earned an MBA from Wharton and a Ph.D. from the University of Paris.
He also wrote two books: Dynamic Hedging (John Wiley & Sons, 1997) and Fooled by Randomness (Textere, 2001). Taleb is perhaps best known for Fooled by Randomness, which explains why we’re genetically disposed to misjudging probability. The book, now in its third edition, explains the relationship between luck and the markets with simple, entertaining examples and describes why market moves are too unpredictable to analyze with standard statistical techniques.
After finishing his Ph.D. in 1998, Taleb founded Empirica, a private investment fund, which he actively managed until 2002. He won’t discuss its performance record, although he continues to trade options and earned a seven-figure income from trading last year, according to 2004 tax returns disclosed to Business Week last fall.
Currently, Taleb is a professor in the sciences of uncertainty at the University of Massachusetts at Amherst and also teaches at NYU and the University of Paris. He spoke with us in early December about his market experiences, the nature of options, the flaws in using the bell curve to price options, and why our brains have trouble judging probability.
AT: You mentioned in Fooled by Randomness that the 1987 stock market crash really made you as a trader. What happened?
NNT: I was long out-of-the-money OTM options in about anything that traded. People were laughing at me.
AT: On both sides of the market? Calls and puts?
NNT: Yes, particularly in financials and currencies — eurodollars, the deutschemark, and the Japanese yen. My largest position was in the yen because its volatility was tremendously low. I also had a collection of [other long OTM currency options], like Swiss franc/Australian dollar and Swiss/Kiwi (New Zealand dollar).
In October 1987, I went to a symposium in Philadelphia. The market had effectively been dead, particularly the financials and currencies. I was on stage with five other traders, and they all said this is the death of volatility. Their idea was central banks now run the world. The banks are getting sophisticated and can force stability just like they can control inflation. There’s no reason to buy an option because the world is moving into a far better regime of managed movement. So anybody buying an option was an idiot.
These guys depressed me. I thought my life was over, because we’re not going to have volatility — or at least no large deviation. Of course, the stock market crashed a few days later and the rest is history. The first thing I noticed [as the market tanked] was stocks didn’t move the most.
AT: What did?
NNT: The eurodollar front-month option. Someone was squeezed and forced to liquidate their position, and it opened up 370 basis points the day after the crash (Oct. 20 — see Figure 1). I had to check the screen to see if it was right.
Click here to view Figure 1: Eurodollars and the 1987 Market Crash
I also had so many out-of-the-money options in the deutschemark I didn’t know what to do with them. The position was so large, it took me a week to go delta neutral. That was when investment banks did not compensate based on income. But I didn’t really care about the money — it was just the feeling that I was right.
Here’s one thing you can learn from this: If you owned an option that was 20 standard deviations out of the money — and I had plenty of those — how many cumulative months of time decay could you sustain if it moved into the money?
AT: I don’t know. A few dozen?
NNT: I quizzed traders, and they were telling me two or three years. But it was 67,000 months of time decay. You get paid 67,000 times your bet.
AT: Is this what you meant in Fooled by Randomness when you discussed the importance of asymmetrical bets — that to measure an outcome you need to consider both probability and the size of the payoff?
NNT: Exactly. It means you don’t need the event to happen often for you to be compensated. And you don’t need to be right on the event, because you can bleed for 67,000 months and still be ahead. After the crash, I told management at Credit Suisse First Boston, “Let me bleed 67,000 months before you question my strategy.”
AT: How did they respond?
NNT: They couldn’t quite understand. When I quizzed them, they said “Well, I’m sure you got 40 to 50 times your payoff.” I then realized I no longer had anything to prove. All you need is a 20-sigma (standard deviation) event.
But if you have a 24-sigma event on an option that’s 24 standard deviations out of the money, your payoff is 750,000 times your bet, which is what happened in eurodollars. It is totally irrelevant whether these events happen every 20 or 50 years. Secondly, the further out of the money an option is, the more complicated it is for the human mind to calculate its properties.
AT: So could you conclude that extremely out-of-the-money options are undervalued?
NNT: No. I don’t know whether those options are over or undervalued. But look what can happen if you’re wrong. A single large deviation can make you right, but 750,000 months of no deviation can’t prove you wrong.
After the 1987 crash, I decided that people didn’t quite understand option payoffs. At that point, I decided to leave options and become a scholar. But I kept a foot in options out of addiction. I specialized in exotic options because they also have complicated payoffs.
AT: What do you mean by exotic options?
NNT: Binary options — options on more than one instrument and either-or scenarios, where you can get either coconut oil or a Treasury bond. I also took a sabbatical as a pit trader at the Chicago Mercantile Exchange [beginning in 1992].
AT: What was that experience like?
NNT: I suddenly discovered a whole sector of self-employed traders. It was nice not trading for a large bank where people tell you what positions you should have. These traders were scalpers, making income around the clock — $500 an hour just like lawyers. They were their own bosses and didn’t really care about what others thought.
AT: That appealed to you?
NNT: I liked it. Before I became a pit trader, I was managing director at UBS and had to wear a tie and go to meetings. In the pit, you’re making moderate income, but you’re in control of your life.
AT: If you enjoyed trading in the pits, why did you stop?
NNT: I wasn’t made for the pits. And I specialized in exotic options, not executing trades. Also, I was in the pit in 1992-93, when the market had zero volatility, which is bad if you’re buying options. It was like watching grass grow.
AT: Are the options you buy so far out of the money that you have to use the over-the-counter market or can you buy them on exchanges?
NNT: You can buy them on the exchange. But you may have to wait years for them to pay off while using other strategies. You have to be extremely patient or not care at all about performance. After the 1987 crash, I had the luxury of not caring. If one event can pay 2,000 years of time decay, you can really afford to wait a few years [for another one].
AT: But meanwhile you could sell near-the-money options?
NNT: Yes, but it’s a lot of work. Now I have an economic interest in other traders (through the Empirica fund) that sell nearthe- money options.
AT: Why is it so much work?
NNT: If you want to place butterflies in 500 securities, it’s a lot of work because you have to adjust them dynamically. But if you want to buy strangles in 500 different securities, it’s a nobrainer. (A butterfly options strategy sells options with strike prices near the current price and buys options further away from the money to protect them. A long strangle buys options both above and below the market in hopes that the market will exceed those strike-price thresholds by expiration. See “Long butterflies”)
If you care about performance, you should short at-themoney options, which expire and have very unstable deltas. Sometimes they bite you at expiration, so you have to monitor them. The amount of labor involved in strategies that have both long and short options is astronomically higher than just buying options.
AT: So the long out-of-the-money options aren’t as risky because their gamma — delta’s rate of change — isn’t that high?
NNT: Exactly. When they pay off, [the reward] is huge. But when you’re selling options, you need a lot of traders. For example, two or three traders can trade long out-of-the-money options on 500 instruments, but when you’re using long and short strategies, two or three traders can only monitor 50 or 60 positions. We invest in traders who sell at-the-money options, and we concentrate on just buying the “wings” (the out-of-themoney puts and calls of the butterfly position).
AT: Are your other traders placing butterfly positions or simply selling ATM options?
NNT: I call it a mixed strategy. Some of the traders sell at-themoney options and buy the wings (creating a complete butterfly position), and some just sell these options, while we buy the wings for them. But you need a lot more that just a butterfly position. I buy butterflies and also buy a lot more wings.
AT: What’s the benefit to this approach?
NNT: A butterfly position allows you to wait a lot longer for the wings to become profitable. In other words, a strategy that involves a butterfly allows you to be far more aggressive [when buying out-of-the-money options].
When you short near-the-money options, they bring in a lot of cash, so you can afford to spend more on out-of-the-money options. You can do a lot better as a spread trader.
You can make some money in options, but the larger the deviation, the less we understand. Secondly, it doesn’t mean all out-of-the-money options are priced wrong.
AT: So there’s no way to value extremely out-of-the-money options?
NNT: Right. They’re so far out there, you don’t have enough months in history to figure out [the correct price]. There are techniques by Professor Benoit Mandelbrot that seem to be convincing, though.
When you bet against the tails (i.e., selling out-of-the-money options and hoping large deviations won’t occur), it takes a long time to be proven wrong. But when you’re wrong, you lose everything. In our society, if something doesn’t happen for awhile, we forget about it. On the other hand, there’s what I call the “suckers’ volatility” — whenever people have a plausible reason to buy volatility, don’t get close to it.
AT: Because it’s too expensive already?
NNT: Let me give you an example. If you went to JFK airport and offered terrorist insurance to passengers, you can get them to pay up to X dollars. But if you try to sell them general insurance, which of course includes terrorism, how much would they pay for it? They would pay less.
NNT: Because there are some properties of the human brain such that the vividness of the terrorist possibility may cause you to overpay. You won’t pay on valuation or probability, but how explainable it is to your brain. You’re likely to pay more for an option that delivers a given payoff.
We’re not programmed to deal with variables that can take very large deviations. We tend to not pay at all for things when we don’t have reason to pay for them, but overpay when we see a reason.
AT: I inferred from your book that using historical data may be a good idea to generate trade ideas, but it isn’t a good way to measure risk. Is any historical data worth testing?
NNT: It is worth analyzing. You cannot ignore past data, but you have to be careful about what you’re interpreting from it. You can’t [use it to determine the probability of rare events]. Large deviations can’t easily be seen from past data.
The bell-shape distribution is a fraud. For the bell curve, much of the deviations are delivered by regular volatility. But in the markets, much of the deviations are delivered by the tails (extremely large, infrequent moves). People use the bell curve because it simplifies things and gives the illusion of understanding what’s going on.
If you use past volatility to predict future volatility, it would hardly predict anything. The bell curve does not apply to something that has fat tails. Now, Mandelbrot’s theory of Power Laws distribution is used everywhere except in finance — in science, in sociology, and on the Internet. (For more details about the bell curve and Taleb’s argument, see “Taleb’s critique of the bell curve”).
AT: What is the Power Laws distribution?
NNT: It’s a different statistical paradigm in which the word volatility is not used. Volatility is a concept that is ingrained in the Gaussian bell shape distribution. In real markets, it doesn’t mean anything, because they can have very large deviations. The essence of a power laws distribution is that the ratio of deviations stays close to a constant. For instance, the number of people with wealth higher than $1 billion in relation to those with $2 billion or more is the same as the number of those with wealth higher than $200 million in relation to those with a fortune of $100 million. The inequality is the same regardless of wealth. The same applies to market movement.
If you shoot to minimize daily variations, you may even increase your risk. Instead, you want to minimize the number of losses over 10 percent. If, for example, you’re flying in a plane, you’re more concerned with the risk of a crash than how bumpy it may be.
There are two types of volatility — tail and regular. More of the volatility in the world is explained by a few deviations. Just like a small number of movies explain the bulk of movie sales, a small number of days in the market explain the bulk of its volatility. In the real world, a small number of days represent a huge percentage of returns in the S&P.
AT: In Fooled by Randomness, you distinguish between randomness in the physical world and randomness in the markets. How is this related?
NNT: Take height for example. You have zero probability of seeing someone in Chicago who is 10 feet tall. A one-million foot tall person isn’t possible, either. But a currency or stock can take any value. For example, in Germany in the 20s, the deutschemark went from four to four trillion. There’s absolutely no friction — nothing can stop it from taking a crazy value.
AT: Have there been other times besides the 1987 stock crash when you received a large payoff from this approach?
NNT: Yes, but all you need is one event to disapprove the Gaussian bell curve. You weaken your point by looking for more events. I just had to bet these rare events were possible. You don’t have to back test if someone tells you, “If you open this door, it’s possible to have this event.” The other guy (i.e., an options seller) is betting it’s impossible to have this event.
AT: Did your success in the stock market crash and your experience buying out-of-the-money options help keep your emotions in check afterward as you were buying out-of-the money options and consistently losing those small premiums?
NNT: No. I’m very emotional, which is why I don’t like trading. That’s why I wanted to retire after the crash of 1987. I like options intellectually, but I don’t concentrate as well when I have a position on.
AT: Do you just have to deal with the emotional side of trading, or is there a way to minimize it?
NNT: I have to deal with it. My strength is that I’m emotional, but that’s also my problem. I’ve always wanted to be out of the markets, but I have this love for options.
What bothers me is not the trading but what comes with it — the emotions and dealing with investors. Trading is like chocolate — a little bit is good for you, but a lot [can hurt you]. It should be done in small doses.