Historically, fat finger trades — where a trader presses the wrong key or adds a zero too many on an electronic trading system — were considered exclusive to equities markets. Not anymore. They are now stretching their piggy little digits across into other asset classes.
Most recently, on June 8, an alleged fat finger wiped eight per cent off of natural gas prices on NYMEX in an after-hours trade from Asia. The natural gas market recovered almost immediately, but not before some savvy traders saw what had happened and jumped in to buy and profit from the mistake, according to Reuters. The same thing happened during the May 6, 2010, equities flash crash, although it was the smarter machines with clever algorithms that did most of the jumping in, rather than human beings.
Fat fingers, abusive or manipulative trading and mini flash crashes, are increasing as high frequency trading penetrates alternative asset classes, such as energy and commodities, at a rapid pace. The Reuters article said that about a third of energy futures volume is now done by computers, and HFT accounts for half of that. Issues with high-speed algorithmic trading are growing concurrently.
Consider that on May 5 the crude oil market saw its second largest ever daily drop when sell-stops were triggered again and again by trading algorithms. This time there was no fat finger, and no Greek-tragedy style news such as that of May 6th, 2010. The $13 drop in the price of Brent was almost unprecedented, yet it made few headlines. Oil prices dropping are like equities prices rising – good news for most people.
But good news or not, the fall was exacerbated by several machines which had similar sell-stops programmed in. If a machine hits a sell-stop and the market plummets, it will likely encounter more sell-stops and push the market further down. This could cost investors a serious amount of money. If algorithms are not programmed to sniff out when they should stop selling, perhaps even to jump back in to buy, then money will be left on the table.
On May 5, human traders with charts could have predicted the fall, and some did — just as they did during the later natural gas drop. There is nothing illegal about taking advantage of a sell-off, whether it is event-led or whether it is human or machine-created. There are concerns over whether human beings could trigger the sell-off intentionally, however. That is the domain of the already-overburdened regulators. But the velocity of the drop and the ungainly actions of the trading algorithms are my concern here.
As HFT-style algorithms migrate to asset classes with higher volatility, and oil is certainly a prime candidate, the margin for error increases. Fat fingers, market abuse or manipulative trading will trigger increasingly quick market moves. Many will happen before a human being can say “what the…?” and capitalise on them.
As we saw on the day of the flash crash, many algorithms sensed there was something wrong and pulled out of the market — which decimated liquidity. While some of their owners were scratching their heads in puzzlement, some of the more clever algorithms recognised a buying opportunity. A report by four economists, “The Flash Crash: The Impact of High Frequency Trading on an Electronic Market,” said that the buying activity of so-called “opportunistic” traders on May 6, 2010, could have translated into “substantial profits.”
In markets that are relatively new to HFT, such as energy and commodities, issues with mini flash crashes, fat fingers and market abuse are just beginning. High frequency trading is still fraught with obstacles and issues, some of which cannot be regulated away. But out of some of these threats can arise opportunities for machines and algorithms that are flexible enough to face down and profit from the adversity.