Ben Read, Electronic Sales, Merrill Lynch, examines the ongoing trends and potential pitfalls in electronic trading in Australia.
To the year ended July 31, turnover on the ASX had fallen 46% over the previous year on year, with eight down months out of the 12. (Please see diagram1 on next page). August provided some relief, with turnover up 23%, but globally equity volumes continue to decline through the seasonally quiet summer months; driven by a continued rotation out of equities amidst global economic uncertainty – the European Sovereign Debt Crisis, the US election and “Fiscal Cliff”, and Chinese growth moderation at the same time as a once-in-a-decade leadership change, all weigh heavily.
Beyond the macroeconomic (cyclical) challenges over the past 18 months, there has been a significant structural shift in the Australian market. Most significantly, in executing orders on behalf of investors, participants now have two public, or lit, venues and exchange-run/broker dark pools to consider. Best execution obligations set out in the corresponding ASIC Market Integrity Rules have introduced a new set of market integrity rules, which bear resemblance to the European Union’s Markets in Financial Instruments Directive (MiFID). Each require every market participant to now have a policy in place which sets out the framework in which it will meet its best execution obligations to its clients. The similarity to MiFID is its principles-based nature of this best execution requirement, which allows each broker to decide how it determines the best outcome for clients.
This is driving electronic trading/execution to become a more significant part of how Australian markets trade. Globally, we have witnessed this over the past decade and recent data estimates that 75% of US and 55% of European volumes are now being executed electronically. In Australia, we estimate this figure to be closer to 35%; up significantly over the past 2-3 years, but still very low by global standards, which, to us, implies that there is still substantial growth ahead. ASIC has been following this shift towards electronic, or direct, execution and in its recent consultation paper (CP184) provided guidance and rules on automated trading. One point was to ensure that all participants maintain robust filters and controls across all platforms that access the market, with a key requirement to have a kill switch in place that can shut down parts of the execution infrastructure when behaviour outside accepted patterns is detected.
A kill switch can take two forms: it can either be software coded, or hardware-based. The former is a functionality that exists across the execution platform. It allows the trading desk, in combination with oversight functions, to quickly shut down either flow from a particular client, or a particular algorithm, that they believe is not behaving correctly. A hardware-based solution is more complex and involves the physical disconnection of parts of a market participant’s order routing platform from the exchange infrastructure. This is typically accomplished by closing or blocking network switches that exist between the two.
Also vitally important is the real-time monitoring that alerts trading desks to any form of execution that is occurring outside acceptable parameters. Most market participants with established electronic trading platforms have pre-trade filters based on a percentage of AD V, value of single orders, total notional and net delta. Taking this further is the implementation of systems that can alert on abnormal patterns of messages, trades and realised profit and loss. Significant deviations from historical patterns can significantly speed up decision making in whether to involve a kill switch.
After the recent events in the US involving Knight Securities, SEC Chairman, Mary Schapiro commented that “when broker-dealers with access to markets use computers to trade, trade fast, or trade frequently, they must check those systems to ensure they are operating properly.1 ”
In Australia, the investment community is supportive of such initiatives to ensure that overall market integrity and confidence is maintained and that investors and participants are protected from similar events occurring. In the immediate aftermath of Knight Capital, sell-side lines of business right across the industry conducted thorough reviews of their internal risk frameworks and there was plenty of dialogue with clients on what their trading limits were with different trading desks they faced. The majority of this work started before ASIC formally published CP184.
Algos: Definitions Matter
We think it is also important to distinguish between the different types of “algorithmic trading” given the air-time this contentious issue has received over the last 12 months. Benchmark execution algorithms used by sell-side trading desks and accessed directly by the buy-side are for the most part relatively safe. They execute a specified number of shares in a single or basket of stocks, governed by a framework of parameters that controls how they post and take liquidity from lit and dark venues. Typical algorithms of this type are Implementation Shortfall, VWAP, Percentage of Volume and Market on Close. They are some of the tools used by many participants to help achieve the best execution outcomes for clients.
The algorithms or automated strategies run by proprietary trading organisations and high frequency trading (HFT) firms have more discretion in what they buy or sell based on the market conditions they encounter. They can generate orders without the direct oversight of a portfolio manager or any human intervention at all. For example, a market making strategy can trade across a whole index of stocks, generating orders to both buy and sell the same stock concurrently and auto executing futures contracts to manage risk.
The volume of orders and turnover generated by a high frequency trading strategy is likely to be significantly higher than a regular execution algorithm. However, the right risk and control framework at the exchange and participant levels, can offer significant protection to the Australian market from the systemic risk that may be posed by the combination of algorithms operating on a daily basis.
Finally, a lot has been spoken about HFT algorithms that are deliberately setup to manipulate or game other investors. We do not believe that the majority of HFT strategies behave in this way; they operate legitimate business models, which are fully compliant with ASIC market integrity rules.
Dark pools are currently a focus and there has been much discussion on how much of market turnover is getting transacted within them. Our research shows that the percentage of total turnover that has been transacted in dark has remained relatively stable at 11-15% in 2012. For the purpose of our research, we have designated volume executed in Centrepoint, priority crossed or National Best Bid and Offer (NBBO ) trades as dark. We think the right metric for the Australian market is the amount of volume that’s getting executed within ASX’s Centrepoint, and the crossing mechanisms that report to market. Block trades negotiated in the upstairs market have been classified as offmarket volume for many years and should continue to be so.
The key is finding the right balance that protects price discovery and liquidity in the lit market, but still affords investors the ability to keep their footprint minimal. The majority of investors are focused on minimising their overall market impact. Hence, it is important that the market structure moves towards an equilibrium with an appropriate balance between the protection of price discovery and liquidity in the lit market on one hand, and the ability of investors to minimise their footprint via the dark pools on the other. A quality independent research on the links between the quality (price discovery and liquidity) of the lit markets and the growth in dark pool activity would be a welcomed addition to the current debates.
If ASIC were to impose minimum sizes for dark pool crossings, as the ASX is campaigning for, then we believe something is needed that is a more flexible than an arbitrary dollar value. As turnover increases or decreases and the price of individual stocks moves then a single value can go from being appropriate to inappropriate very quickly. One suggested solution is a value that is a function of the stock’s liquidity. For example, five times the average trade size based on the previous 20 trading days.
Global regulators are certainly aware that that every round of wholesale regulatory change adds a significant cost burden to market stakeholders both on the buy- and sellsides. The majority of resources are going into technology, compliance and legal functions. What the industry globally is still trying to evaluate is whether the more stringent regulatory requirements and all of the investment made has yet led to both the buy- and sell-side being able to offer a better product or service to their clients.
As a final point, the majority of Australian participants are global firms where Australia is just one of many territories they operate within. If the cost of running an equities business within Australia becomes prohibitively high then there is a danger that, in an environment of reduced activity, focus and resources are pulled back in favour of other regional markets. This in turn could reduce the breadth of services that are available to our domestic managers.