The exponential growth in high frequency trading (HFT) over the last few years and some recent high profile HFT glitches on US markets have attracted the close attention of regulators around the world, including ASIC.
Automated HFT is big. Exactly how big is unclear, with estimates of ASX trading attributable to HFT ranging from 15% to 25%.1 On US markets the estimates are as high as 75%.2 Whatever the exact number, HFT is big and getting bigger.
Opinions on the merits of HFT are sharply divided. Proponents claim that HFT adds liquidity and reduces spreads. Critics, on the other hand, argue that the computers unfairly front run trades and could engage in market manipulation or insider trading.3 The sheer volume of HFT on various markets and its perceived threat to market integrity by some, has brought HFT into the focus of regulators around the world, including ASIC.
What is HFT?
HFT refers to trades implemented through computer algorithms that independently generate trading messages without human input. Algorithmic trading is generally based on pricing movements and market information within predetermined parameters, rules and conditions. It is often characterised by speed sensitivity where a very small fraction of a second can mean the difference between hitting or missing an opportunity. So much so that Australian HFT traders are allowed to physically locate servers adjacent to ASX servers to facilitate the fastest possible order entries. HFT traders generally hold positions for short periods of time, often with zero end-of-day holdings.
Polarised views on HFT
Proponents of HFT argue it facilitates the efficient implementation of legitimate transactions and trading strategies. It can improve liquidity and spreads and reduce volatility across the market. For exchange traded funds (ETFs) in particular, HFT by authorised participants and market makers improves price transparency and liquidity. HFT can also assist market participants to meet best execution obligations, particularly where trading can be implemented on multiple markets.
Some HFT strategies, sometimes referred to as “stealth” or “gaming” strategies, are more controversial. Algorithms can be programmed to read real-time market data and information to identify trading patterns including when a larger order has been placed. The program can then essentially front run the order to make a spread selling into it. Other strategies involve generating trading by others, for example by creating trends, and then attempting to profit from that trading. Investor concerns about the potential impact of these HFT strategies on market integrity is claimed to be one of the drivers leading an increasing trend for transactions to move from the lit pools of the ASX and Chi-X to private dark pools.
Some investors have called for greater regulation of HFT on the basis that it can result in market manipulation and insider trading at the expense of retail investors.4 While others see these claims as over-the-top scare-mongering, regulators around the world have taken notice and increased their focus on HFT generally and on controls around flawed trading programs in particular.5
The last ten years have seen exponential growth in the volume, scope, speed and complexity of HFT, particularly in the US. With the extreme speeds at which HFT trades are placed and executed, come increased consequences of an algorithm glitch. For example, the May 2010 US “flash crash” saw extreme price drops stemming from alleged faulty algorithmic trades. The market impact of the initial alleged faulty trades was compounded as other trading algorithms responded to those trades and rode the trends they established. This led to a market crash of almost 6% in a matter of minutes, most of which was recovered a little later.
The Nasdaq trading delay fiasco in this year’s Facebook IPO was another recent incident which some claim was the result of HFT. The volume of transaction orders was so high that it crashed Nasdaq’s trading systems.
More recently, Knight Capital, a leading US broker responsible for approximately 10% of all trading on US equity markets, teetered on the brink of collapse after a glitch in a new trading algorithm caused the firm to make faulty trades. In a matter of minutes, Knight Capital lost $440 million. One of the principal reasons for the extent of the losses was an inability to turn off the trading machines for a certain period even after the faults were identified.
Market regulators around the world are currently focussing on HFT. ASIC has released no fewer than four consultation papers in the last two years addressing HFT.6 In CP 184, ASIC proposes amendments to the ASX and Chi-X Market Integrity Rules in respect of automated order processing (AOP), through which HFT trading is implemented. Broadly, the amendments would require market participants to:
- have direct control over pre-trade filters (ASIC acknowledges that a market participant’s clients can have some discretion to change filter parameters within a defined range);
- undertake an annual review of their AOP systems and certify compliance with the Market Integrity Rules relating to AOP;
- have controls that enable immediate suspension, limitation, or prohibition of the conduct of all AOP for a client; and
- have controls that enable immediate suspension, limitation or prohibition on orders or series of orders that would interfere with the efficiency and integrity of the market (referred to as a “kill switch”).
In addition, ASIC’s draft regulatory guide sets out ASIC’s raised expectations for market participants to:
- put in place real time monitoring of all trading messages submitted to the market; and
- test AOP filters, filter parameters, controls and systems. In testing AOP systems, ASIC states that a market participant should use a non-production test environment to stress test algorithmic programs and order flow.7
ASIC acknowledges that certain HFT strategies could involve market misconduct and that these strategies are subject to increased ASIC scrutiny. However, HFT practices can be very difficult to monitor and regulate, and it is not always easy to distinguish legitimate HFT from more predatory HFT strategies. Indeed, to date there has been little international regulatory impetus to try to weed out undesirable HFT strategies. Stock exchange operators, dependant on HFT generated revenues, face commercial pressures to facilitate HFT traders’ requests. In some markets, such as the US, HFT now represents such a substantial proportion of transactions that a major overhaul may face insurmountable resistance.
Could a Knight Capital style glitch happen here?
Australian HFT programs are not immune to gremlins. The measures proposed by ASIC will go some way towards minimising a Knight Capital style meltdown in Australia. In particular, a kill-switch requirement might have limited the extent of the Knight Capital losses as it could have facilitated a speedier termination of faulty orders. In addition, if ASIC’s proposed guidance is followed and HFT traders properly test their AOP systems and filters as well as all trading algorithms, this might result in stronger protections against faulty HFT algorithms.
Ultimately however, the success of the AOP system depends on the rigor of each market participant’s tests and monitoring systems which necessarily involves an element of self-regulation. It is certainly possible to imagine scenarios where flawed algorithms could lead to defective messages and trading losses for an Australian HFT program.