High-frequency-trading – why is it the subject of increased global regulatory attention?
By Ingrid Goodspeed
Governor of the Institute of Financial Markets
Computers are incredibly fast, accurate, and stupid: humans are incredibly slow, inaccurate and brilliant; together they are powerful beyond imagination.
n 1 August 2012 an error(1) in its trading software caused losses of USD440million (ZAR3.6billion) for Knight Capital. Knight Capital is a global financial services firm that engages in market making and trading in global equities, fixed income, foreign exchange, options and futures. Knight Capital’s electronic market making unit offers designated market maker services at the New York Stock Exchange (NYSE) and executes about 10% of US share trade volume, taking advantage of regulations that spread trading across many on- and off-exchange venues including electronic communication networks(2) and dark pools(3) . Within two days of the event, the company’s share price had fallen by 75%. Had the firm not received a rescue (rather than takeover) package from Blackstone, Getco, Jefferies and others, bankruptcy was virtually inevitable.
During the first 45 minutes of trading on 1 August 2012, Knight Capital’s algorithmic or algo trading system (see Box 1: Algorithmic trading) malfunctioned and executed erroneous orders in 148 shares listed on the NYSE. The scale of orders pushed prices sharply higher or lower. Since May 2010 circuit breakers are in place to limit wild swings in the price of shares beyond 10% during a five minute period. However the circuit breakers did not kick in as they only start operating fifteen minutes into the trading day. The NYSE reviewed trades in all 148 shares but broke (or cancelled) trades in only six shares, which left Knight Capital facing substantial losses having amassed a USD7billion position it had to off-load at a loss . Of course breaking trades underlies the problem of using and being dependent on high-speed trading technology as once a trade is executed it triggers a chain of positions that become costly to break.
Box 1: Algorithmic trading
Automated (or algorithmic or programme) trading is trading using algorithms (i.e. sets of rules or instructions) at some stage in the trading process. Computers and advanced mathematical models are used to make decisions about the timing, price and quantity of an order. Automated trading ranges from simple algorithmic execution to complex algorithmic trade decision-making.
Algorithmic execution involves the use of an electronic trading programme to execute a trade after the decision to trade has been made by a human trader. For example automated trading may comprise a basic algorithm to feed portions of an order into the market at pre-set intervals to minimise market impact cost. Alternatively the programme may use smart-order routing, which helps traders seek out where prices are best across a range of competing exchanges, platforms and dark pools, and routes their orders accordingly.
Algorithmic trade decision-making involves the use of algorithms to initiate trades based on key input parameters such as order-book imbalance, momentum, correlations within or across markets, mean reversion and response to economic data or news headlines. Algorithms may also be designed to predict the presence and actions of other algorithms, thereby attempting to stay one step ahead of them. Once the algorithm has made the decision to trade, it also executes the trade. Thus trades can be made without human intervention using information received electronically.
High-frequency trading (HFT) can be thought of as a subset of algorithmic trade decision-making. High-frequency traders attempt to generate mainly arbitrage profits by doing a large number of small-size, small-profit trades with short holding periods of frequently less than one second.
The Knight Capital episode was a stark reminder of the flash crash in May 2010 that temporarily erased USD862billion in US equity value and highlighted that stock markets are vulnerable to glitches in high-speed computer trading systems. On 6 May 2010 the prices of almost 8 000 US-based individual shares and exchange traded funds experienced extremely rapid price declines and reversals within a short period of time, most falling from 5% to 15% before recovering most if not all of their losses. However over 20 000 trades across more than 300 securities were executed at prices more than 60% away from their values just moments before. After the market closed, the stock exchanges and Financial Industry Regulatory Authority(4)(FINRA) met and jointly agreed to break such trades in terms of their ‘clearly erroneous’ trade rules. Importantly no market-wide circuit breakers were triggered on 6 May.
The report to the Joint Commodity Futures Trading Commission (CFTC) and Securities and Exchange Commission (SEC) Advisory Committee on Emerging Regulatory Issues regarding the market events of 6 May 2010 found that the key lessons of the flash crash were that (i) under stressed market conditions the automated execution of a large sell order can generate significant price changes. In addition the interface between automated execution programs and algorithmic trading strategies can quickly erode liquidity and result in disorderly markets and that high trading volume is not a reliable indicator of market liquidity, (ii) many market participants have their own versions of a trading pause(5) based on different market-signal combinations. The withdrawal of a single market participant may not significantly impact the entire market. However the withdrawal of many market participants at the same time, may result in a liquidity crisis, which in turn may cause the price-discovery process to breakdown, and (iii) market participants’ uncertainty about when trades will be broken can impact their trading strategies and willingness to provide liquidity.
To address these issues: (i) the CFTC, SEC and exchanges recalibrated market-wide circuit breakers to pause trading across the U.S. markets in a security for five minutes if that security has had a 10% price change over the preceding five minutes and (ii) the SEC, exchanges and FINRA clarified the process for breaking erroneous trades using more objective standards (see Box 2) and (iii) the SEC approved a rule in July 2012 requiring exchanges and FINRA to establish a consolidated audit trail (CAT) that will enable the reconstruction of a market crisis and allow surveillance across 13 exchanges, 10 options markets and 200 broker-dealers that execute share trades off-exchange i.e., in alternative trading venues such as dark pools. Implementation is not soon, but rather in about two years’ time.
Box 2: Standards for breaking erroneous trades
Where circuit breakers are not applicable, trades will be broken as follows:
- For shares priced USD25 or less, trades will be broken if the trades are at least 10% away from the circuit breaker trigger price.
- For shares priced more than USD25 to USD50, trades will be broken if they are 5% away from the circuit breaker trigger price.
- For shares priced more than USD50, the trades will be broken if they are 3% away from the circuit breaker trigger price
Where circuit breakers are not applicable, trades will be broken as follows:
- For events involving between five and 20 stocks, trades will be broken that are at least 10% away from the reference price, typically the last sale before pricing was disrupted.
- For events involving more than 20 stocks, trades will be broken that are at least 30% away from the reference price.
In May 2012 the CFTC Technology Advisory Committee’s Sub-committee on Automated and High-frequency Trading, released a draft definition of HFT as follows: HFT is a form of automated trading that employs:(i) algorithms for decision making, order initiation, generation, routing, or execution, for each individual transaction without human direction; (ii) low-latency technology that is designed to minimize response times, including proximity and co-location services; (iii) high speed connections to markets for order entry; and (iv) high message rates (orders, quotes or cancellations). In addition the subcommittee is working on (i) categorising, for example by trading strategies and trading methods, the broad composite of HFT for focussed oversight and enforcement purposes, (ii) the supervision and oversight of HFT and appropriate HFT controls and (iii) how HFT activity impacts market quality metrics such as market depth and liquidity. Furthermore the SEC held a round-table discussion on 2 October 2012 to look at ways of protecting highly automated markets from algorithms that run out of control. The round table discussed preventing errors through system design, deployment and operation as well as responding to errors and malfunction and managing crises in real time.
Other international regulators have had algorithms and high-frequency trading on their radar screens for a number of years. In Europe, Australia and Hong Kong, regulators are investigating, setting out guidelines or introducing controls on automated trading.
In February 2012 the European Securities and Markets Authority (ESMA) issued ‘Guidelines on systems and controls in an automated trading environment for trading platforms, investment firms and competent authorities’, HFT being one form of automated trading. The guidelines were effective across the EU from 1 May 2012. In respect of trading platforms the guidelines require firms to (i) make use of clearly defined development and testing methodologies prior to using an electronic trading system, (ii) develop testing standards to ensure that systems being used to access a trading platform will not pose a threat to the fair and orderly trading on that platform, (iii) be able to limit access, cancel or change transactions and prevent the excessive flooding of the order book at any one moment in time, (iv) have limits to prevent capacity limits being breached, (v) have effective arrangements to prevent ping orders(6),quote stuffing(7), momentum ignition(8) and layering and spoofing(9).
On 13 August 2012 the Australian Securities and Investments Commission (ASIC) unveiled draft market integrity rules and guidance on automated trading, which include new market integrity rules requiring direct control over filters and automated controls to suspend orders and/or systems; rules that revise the process for endorsing systems and reviewing changes at least yearly; and regulatory guidance on automated trading such as the testing of systems, filters and controls and stress testing of order flows. The consultation paper says that ‘an aberrant algorithm generates not only costs that are borne by the firms using the algorithms, but also negative impacts for all market participants by impairing the fairness and orderliness of the market’
In July 2012, Hong Kong’s Securities and Futures Commission issued a consultation paper on the regulation of electronic trading. The paper noted that in line with other major international markets, market participants in Hong Kong are using automated trading algorithms for the implementation of complex trading strategies. The regulation requires intermediaries to ensure that pre-trade risk management controls such as trading and credit limits as well as post-trade monitoring are in place to mitigate the risk of HFT software malfunctions.
In July 2012 the JSE celebrated the launch of its new trading platform and the move of the platform from London to Johannesburg. The JSE expects these developments to have ‘a significant impact on trading volumes by attracting a greater share of high frequency trading when collocation and the appropriate billing models are in place’. The Financial Services Board is aware of the risks and benefits of HFT. No doubt regulatory guidelines will soon be put in place in South Africa to mitigate the potential risk of HFT causing adverse effects on the JSE and its participants.
In the words of SEC Chairperson Schapiro on the Knight Capital trading issue: ‘reliance on computers is a fact of life not only in markets everywhere but in virtually every facet of business’. There is no doubt that HFT is here to stay. However regulators will need to continue to review automated
trading mishaps and quickly implement additional risk management measures, particularly as a similar event to that experienced by Knight Capital could threaten domestic and global financial stability if it took out a systemically important institution.
(1) also known as a ‘bug’ or an ‘undocumented feature’
(2) Electronic Communication Network (ECN) is an order-driven system where the buy and sell orders of investors meet directly in an order book, either in a call auction or in continuous trading. Institutional investors, broker-dealers, and market-makers that are subscribers to an ECN, place their trades directly with the ECN. Individual investors must have an account with a broker-dealer subscriber to place trades on an ECN.
(3) Dark pools are orders and quotes in listed shares that cannot be seen by other market participants until the trade is complete.
(4) FINRA is the largest regulator for all securities firms doing business in the United States. It oversees nearly 4 380 brokerage firms, 163 150 branch offices and 633 000 registered securities representatives. Its chief role is to protect investors by maintaining the fairness of the U.S. capital markets. www.finra.org
(5) A pause is a halt in trading and is considered a useful way of giving time for (i) market participants to review their strategies, (ii) trading algorithms to reset their parameters, and (iii) an orderly market to be re-established.
(6) Small orders designed to detect hidden orders on dark trading venues
(7) Entering large numbers of orders and / or cancellations to disguise trading strategies
(8) Entering orders or a series of orders to start or intensify a trend
(9) Submitting multiple orders on one side of the order book to remove a trade at a certain value on the other side of an order book
IOSCO. July 2011. Regulatory Issues Raised by the Impact of Technological Changes on Market Integrity and Efficiency. www.iosco.org
U.S. Commodity Futures Trading Commission and the U.S. Securities and Exchange Commission. September 2010. Findings regarding the market events of May 6, 2010. www.cftc.gov