On May 6, 2010, $1trn was wiped from the US stock market in a matter of seconds. Traders and other observers of the market around the world were aghast and baffled by this 600 point dip, yet almost as quickly as they fell, prices recovered. It happened so quickly, many missed the dramatic fall and recovery all together, and the sudden plunge and rise would become known as the ‘Flash Crash’.
What had happened left many, financial authorities included, somewhat perplexed. While the Flash Crash was, as the name suggests, over in a flash, the surprise volatility unsettled analysts. This time around long-term investors, to whom many entrust the future of their pensions and savings, had not lost out as prices quickly recovered, though the incident remained a cause for concern.
In a 2011 speech the Bank of England’s Chief Economist Spencer Dale said that this brief crash had “taught us something important, if uncomfortable, about our state of knowledge of modern financial markets”, he declared. “Not just that it was imperfect”, the economist continued, “but that these imperfections may magnify, sending systemic shockwaves… Flash Crashes, like car crashes, may be more severe the greater the velocity”.
While the Flash Crash itself was just a freak blip, the implications were troubling
While the Flash Crash itself was just a freak blip, the implications were troubling. Relatively harmless though it was, future such incidents could prove more dramatic. Financial regulators and authorities began investigating and five months after the incident, the Securities and Exchange Commission published a report indentifying the culprit. It was all a big accident, apparently. A Kansas City mutual fund had mistakenly placed a large sell order of stock market future contracts on a Chicago exchange.
Uncovering the truth
Fast-forward five years and it appears the real force behind the Flash Crash may have been discovered. This time the alleged culprit was an individual accused of wilfully manipulating the stock market through fraudulent means. In April 2015 the US Department of Justice and Commodity Futures Trading Commission accused London-based high-frequency trader Navinder Singh Sarao of using algorithms to place spoof bids and sales, for which he was arrested.
From his suburban house on the southwestern edges of London, Sarao was accused of helping to trigger the Flash Crash. The 36-year-old is alleged to have made £27m ($42.38m) through fraudulent means on the Chicago Mercantile Exchange over the past five years. Despite being accused of contributing to the infamous Flash Crash and profiting handsomely, the casual observer would hardly have known. Living in a modest house with his parents and driving a Vauxhall Corsa, it was also reported that Sarao’s clothes were usually tracksuits from discount sports outlet Sports Direct. The unassuming stock-market whizz was allegedly placing spoof orders and manipulating the market to make his unspent fortune. By placing an order bid for a large amount of shares, the shares in question take on the appearance of experiencing strong demand pressure, which then raises the share price and attracts buyers. The spoof order will then be cancelled, however the demand created allows the spoofer to sell at a higher price.
Abiding by the same basic principle, the accused is said to have placed large sell orders just above the price of the lowest prevailing prices, with the intention of giving others in the market the impression of large selling pressure, and, in doing so, depressing prices. Once the value was down – as a result of Sarao’s own doing – he would purchase stock at the deflated price. His initial orders would then be cancelled, allowing prices to rise again as the illusion of selling pressure disappeared. Once prices had bounced back he would sell the stock he had bought at the lower price he manipulated, thereby turning a profit.
Sarao’s alleged strategy requires speed, and lots of it. The use of automated algorithms, otherwise known as high frequency trading (HFT), gave him the speed required, to react to each stage of his process at exactly the right time. HFT is the “use of complex algorithms which analyse financial markets and synthesise data faster than other traders”, according to the Securities Arbitration Clinic.
Stock markets are often imagined either as dominated by fast talking traders on top floor city desks in expensive suits, or perhaps alpha male-types dressed in colourful jackets shouting from the trading pit. These images are mistaken. HFT now accounts for, according to consultancy Tabb Group, “as much as 73 percent of US daily equity volume, up from 30 percent in 2005”.
Every second counts
In his book Flash Boys: Cracking the Money Code, Michael Lewis recounts the story of Brad Katsuyama who worked at the Royal Bank of Canada, buying and selling large quantities of stock for clients. In the mid 2000s Katsuyama noticed something strange in the stock markets he worked in. When he would attempt to place a sell or bid order on shares merely by pressing enter on his keyboard, the price would suddenly change – higher if he was buying, lower if he was selling. The baffled banker thought perhaps it was an IT problem and requested his company’s tech team look into the matter. They were just as stumped.
The entire process made no sense to Katsuyama at first. He assembled a team of financial technology experts to look into it. Eventually they worked it out. The computer centres of different exchanges were located in different locations, with different lengths of internet cable between themselves and brokerage houses. The time for an order to go from one exchange to another depended on the length and path of these cables.
When banks such as RBC place large orders of stock, they purchased different amounts from different exchanges. These orders, although all placed at the same time, were received by the different exchanges at slightly different times due to how long it took the order to travel through the internet cables burrowed underground.
High frequency traders using algorithms could exploit this delay of a few seconds. When Katsuyama was placing orders for a large amount of shares from one company at one exchange, the algorithms were able to predict that large orders of the same company’s shares were going to be ordered at another exchange, delayed by a few seconds. This slight delay – miniscule to humans – was open to exploitation by computers. High frequency trading firms would then be able to place orders at these other exchanges, raising the price he would have to pay to complete his order. Katsuyama was face to face with the new, fast paced, computer-driven world of HFT.
The use of computers on the stock exchange has been around since the 1970s, with early initiatives allowing for orders to be sent to the correct trading posts. In the 1980s ‘program trading’ was introduced, in which orders are entered into the market to be executed automatically when certain price points are reached. The use of this was cited as a cause of the 1987 stock market crash.
As Dr William Blewitt of Newcastle University says: “Algorithmic trading itself has been a part of equity trading for decades, with broadening use seen in the 1990s as electronic communication networks became more widespread.” The real boost came with a report in 2001 in which researchers from OBM published a report detailing their laboratory experiments using trading algorithms, with the findings showing the advantage of computers over humans. Throughout the 2000s banks and investment firms increasingly adopted HFT software as part of their trading strategies, which Katsyuama, working for RBC, encountered. As Bloomberg notes, “2007, traditional trading firms were rushing to automate. That year, Citigroup bought ATD for $680m”.
The growth of HFT can be, Blewitt tells World Finance, attributed to a number of factors. One of the most important was the decimalisation of US stock prices [in the 1990s], which allowed stocks to be quoted to the cent, as opposed to a fraction of the dollar”, he says. “Another was the ever-improving infrastructure for high-speed communications, which trading firms had been investing in since the 1990s. By 2009, HFT accounted for over 60 percent of equity turnover by volume in the US.”
The need for speed
The positives and negatives of this rise in HFT have been subject to much debate. Many maintain that HFT has made markets less predictable and more volatile. As Blewitt notes, one of the downsides is “that prices fluctuate significantly over very small periods of time – given the volume of overall trading made up of HFT activities; this can lead to overall market volatility”.
There are also potential benefits. Blewitt maintains that HFT has had some positive impact on stock markets, arguing “faster and more accurate updates of stock prices lead to narrower spreads and more competitive bid-ask prices”. The common line of defence of HFT generally is centred on its increased efficiency. Peter Kovac, himself a high frequency trader and author of Flash Boys: Not So Fast, argues, HFT has “dramatically reduced the cost of trading over the past decade, by five times or more. TD Armeritrade, the largest online retail broker, estimates that in the last 10 years transaction costs have declined 80 percent for retail investors… In short, every-one – retail investors, mutual funds, pension funds, whoever – has benefited significantly.” The increased speed and inefficiency is said to have made “markets substantially cheaper to invest in, reducing costs and adding a little bit to everyone’s investment returns”.
The algorithms used by HFT firms are also closely guarded secrets; in recent years there have been a number of people arrested for theft after taking the specific codes for certain bank HFT strategies. The secret nature of the logic behind certain algorithms makes the market harder to understand and predict.
This does raise one clear problem with HFT: it is used to identify a multitude of small gaps within the market to eke out a profit. As Jerry Alder at Wired notes, high frequency traders “are continuously testing prices, looking for patterns and trends or the chance to buy something in one place for $1 and sell it somewhere else for $1.01, or $1.001”. Where fundamental buyers may bid or offer shares based on their own learned opinion or evaluation of a company, HFT reacts blindly to signals.
Whereas the demand for a share may be determined by a multitude of investors having hope in the prospects of a certain company, HFT will merely respond to pre-programmed price patterns. The decisions of fundamental buyers or the cumulative demand of many smaller investors helps to determine the value – whether that is subjective or objective – while the actions of HFT do nothing of the sort, instead reacting to fluctuations as they have been programmed to.
These blind, reflexive responses of HFT are what allowed Sarao to allegedly take advantage of the stock market. While Sarao may have relied on algorithms to execute his trading strategy – thus consigning him to the category of a high frequency trader according to many – the only people who would have been damaged by his alleged market manipulation would have been other high frequency traders whose algorithms would have been tricked into transactions by Sarao’s.
Ranjiv Sethi, a professor of economics at Columbia University, argues that “the strategies that Sarao was trying to trigger were high-frequency trading programs that combine passive market making with aggressive order anticipation based on privileged access and rapid responses to incoming market data”. More advanced algorithms would have “detected Sarao’s spoofing and may even have tried to profit from it, but less nimble ones would have fallen prey”. Meanwhile, fundamental buyers and sellers would have been unaffected by Sarao’s alleged high speed chicanery, as their investment decisions would have been “based on an analysis of information about the companies of which the index is composed”. Professor Sethi continues, “Such investors would not generally be sensitive to the kind of order book details that Sarao was trying to manipulate.”
Algorithms are here to stay in financial markets. Now that Pandora’s box has been opened, attempts at expunging automated trading would be impossible. The question to be addressed is how exactly these new technologies are used. According to Blewitt, the activities that Sarao is accused of are an abuse of HFT. Known as ‘hype and dump’, such trading strategies – made clear by the arrest of Sarao – are illegal. Blewitt tells World Finance that HFT itself is not to blame: “HFT simply exists as an application of evolving technologies to the stock market infrastructure. As such, it is the market itself which has any exploitable vulnerabilities, and certain abuses of HFT can enable the morally bankrupt to leverage some of them.”
How HFT is used, or abused, is decided by institutional actors. The next big tax for financial authorities will be trying to find the most appropriate way to regulate HFT, in which the benefits are kept and its worst excesses curbed. Financial authorities in the US, evidenced by their pursuit of Sarao, are treating HFT as a bigger threat to the stock market. However, Sarao is, in many ways, being made a scapegoat. He was apparently able to make a tidy profit from the use of algorithms, yet only because of the prevalence of algorithms in the first place. America’s various financial authorities, rather than pursuing an obscure trader in the suburbs of London, should be creating and architecture of regulation that allows, as much as possible, the benefits of HFT to be realised while curbing its worst excesses.
