3. Four real disasters and what they had in common

Derivatives 6: When it goes wrong

The two previous modules have described a mechanism. This one checks that the mechanism works, and it does so with four cases separated by twenty-three years, starring very different people in markets that have nothing to do with one another. What is interesting is not the morbid appeal of the figures, but that the four stories, told in the abstract, are the same story.

1995: Barings and a trader in Singapore

Barings was the oldest merchant bank in the United Kingdom, founded in the eighteenth century. At its Singapore subsidiary, a trader named Nick Leeson built up positions on the Japanese Nikkei index that combined two things: long futures, betting the index would rise, and options sold on both sides, which took in premium as long as the market stayed still. In other words, he was betting on direction and, at the same time, on nothing happening.

On January 17, 1995 an earthquake devastated the city of Kobe. The Nikkei fell sharply and volatility spiked: both engines of the position failed at once. Leeson responded by increasing size to win it back, hiding the hole in an error account that internal supervision did not review. Losses reached the region of 800 million pounds, far above the bank's capital. Barings was sold to the ING group for a symbolic one pound. It is worth underlining the detail that makes this case uncomfortable: there was no exotic trade involved. It was index futures and sold options, the two most elementary instruments in the course, in a size the institution could not stand behind.

1998: Long-Term Capital Management

LTCM was founded in 1994 with the most highly qualified staff a fund had ever assembled: top-tier bond traders and two future Nobel laureates in economics among the partners. Its business consisted of spotting small price differences between very similar assets and waiting for them to close. Since each difference was tiny, enormous leverage was needed for the trade to be worth doing, and the fund's balance sheet came to be tens of times its capital.

In 1998 Russia defaulted on its debt and the market stopped behaving normally. The positions, which looked independent, turned out to be the same bet repeated many times over: that spreads narrow and that there will always be someone on the other side. Capital went from some 4.7 billion dollars to around 400 million in a matter of months, and the Federal Reserve Bank of New York had to coordinate a contribution of several billion from a group of large banks to avoid a disorderly liquidation. Roger Lowenstein reconstructed the episode in detail in a book whose title sums it up: when genius failed.

February 2018: the inverse volatility products

For years there were listed products that replicated a short position in VIX futures. As long as volatility stayed low, they rose almost without interruption and attracted many retail investors who in many cases did not know exactly what they were buying: it was not an equity fund, it was a leveraged bet that the calm would continue.

On February 5, 2018 the VIX recorded its largest single-day percentage rise, more than doubling. The internal mechanism of these products forced them to buy volatility futures to rebalance, precisely as the price was rocketing, feeding their own collapse. The main one lost around 90% of its value between the close and the reopening, and the issuer exercised its right to wind it up early. No recovery was possible: the product ceased to exist.

November 2018: a natural gas options manager

A US firm specializing in selling options for retail clients held, among other positions, naked calls on natural gas. In November 2018 the price of gas rose with unusual violence, of the order of 20% in a single session, in a market already strained by inventories and cold weather.

The outcome is module 2 applied point by point: margin shot up, the margin calls arrived, the positions were closed at market in the middle of the stampede and the losses exceeded the balances in the accounts. Many clients not only lost everything they had put in, they ended up owing money to the clearing firm. The manager told his clients about the disaster by video, apologizing.

What they had in common

Four different markets, twenty-three years apart, four different profiles, and the same skeleton:

• Leverage. In every case the size of the position was a multiple of the capital behind it. Without that, a bad month is a bad month; with it, it is the end.
• A short volatility position, explicit or disguised. Selling options, replicating an inverse VIX or betting that spreads will narrow are variants of the same thing: being paid for nothing happening.
• No way of holding on until they were proved right. None of them was thrown out by a final verdict from the market, but by the margin mechanism, which acts first.
• Models calibrated on data that did not contain the scenario that arrived. All four had risk measures and none of them included an earthquake, a sovereign default, a doubling of the VIX in one day or a 20% jump in gas.

A fifth feature is worth adding, this one for its absence: none of the four protagonists was ignorant. Amaranth, which in 2006 lost billions on natural gas positions, was not either. The lesson is not that they were clumsy, but that technical competence does not offset an asymmetric payoff structure combined with leverage. It is a problem of the architecture of the position, not of the trader's intelligence.