LONDON Whatever the precise trigger, the sudden plunge in Brent prices on Monday is a timely reminder that liquidity is discontinuous, even in a market as deep and heavily traded as crude oil.
Markets can switch from quiet calm to a raging storm with frightening speed for any reason, or no rational reason at all.
Chart 1 shows the sudden drop in oil prices and upsurge in trading just before 1800 GMT.
The most actively traded November ICE Brent crude contract had been under light pressure throughout the day, with prices easing lower on turnover of around 500-1,000 contracts per minute.
The gentle selloff appears to have accelerated after 1700 GMT. By 1751 GMT, prices had already fallen by $1.81 per barrel to $115.20, from a peak earlier in the day of $117.01.
But then the market plunged downwards without any apparent support. Prices dropped 57 cents in just 60 seconds at 1752 GMT, 26 cents more at 1753 GMT, and a massive $1.87 at 1754 GMT. Turnover soared from 152 contracts a minute at 1751 GMT, to 2348 at 1752 GMT, 3090 at 1753 GMT, and an extraordinary 10,246 at 1755 GMT.
Each contract is for 1,000 barrels. So in 60 seconds at 1755 GMT, the market traded futures contracts equivalent to 10.25 million barrels of crude, more than the entire daily output of Saudi Arabia, and over half of all the refined products consumed in the United States every day.
Whether the abrupt selloff was unusual depends on the time horizon used for comparison.
In the otherwise calm conditions which had prevailed during the previous few trading sessions, the selloff was extremely unusual. The 1.64 percent price drop at 1755 GMT was equivalent to more than 24 standard deviations, based on recent low levels of volatility, which ranks it in the highest category of exceptional events (Chart 2).
The last person to complain of a 25-sigma move was Goldman Sachs Chief Financial Officer David Viniar following the market volatility in August 2007. "We were seeing things that were 25-standard deviation moves, several days in a row," Viniar said.
By the end of the day, however, prices had reversed some earlier losses. November Brent futures closed down just 2.46 percent. For the day as a whole, the price move was only a little over 1 standard deviation, which was unexceptional (Chart 3).
The point is that the level of volatility is itself highly variable. In fact volatility is more variable than underlying prices: the volatility of volatility is more volatile than plain volatility itself.
The oil market, like any other asset market, experiences abrupt regime shifts from calm trading to wild trading conditions and back again, as the legendary French mathematician Benoit Mandelbrot explained ("The (mis)behavior of markets", 2004).
Just before 1800 GMT on Monday, the market experienced one of those sudden regime shifts, switching from dead calm to exceptional levels of volatility within less than five minutes.
The initial disruption was followed by a series of aftershocks - with other unusually large price moves, both gains and falls - over the next three and a half hours, before calm trading conditions returned by around 2130 GMT.
HUNTING FOR A REASON
Monday's oil price plunge and partial recovery is reminiscent of previous flash crashes in the oil market (May 2011) and equities (May 2010) when prices dropped abruptly without any apparent fundamental cause.
Commentators have already suggested various possible triggers for yesterday's move: an input error by a trader ("fat finger"); liquidation by a hedge fund; falling prices triggering stop-loss orders placed at $115; heavy turnover by computer-driven trading program; or some complex interaction among all these factors.
But as with previous flash crashes, it will take time to trace the chain of events back to the original cause, assuming it is ever known. In previous selloffs, the ultimate cause has rarely been established, unless regulators initiate a lengthy and detailed enquiry, which seems unlikely at the moment.
According to behavioral scientists, humans, including traders and analysts, have an over-riding need to make sense of the world by creating narratives of cause and effect. In a narrative approach, a big event (such as an abrupt drop in oil prices) must have a big cause (such as a big hedge fund selling or a fat finger data entry error).
In practice, however, many major developments are the result of an unexpected combination of small factors, none particularly significant in themselves. It is just as likely that several unconnected factors (light hedge fund selling, a concentration of stop-loss orders and destabilizing feedback among computer-driven trading programs) came together to produce an unusually large and fast price move, defying the desire for a neat explanation.
(Editing by Alison Birrane)