The Gateway to Algorithmic and Automated Trading

A Global Round Trip

Published in Automated Trader Magazine Issue 06 July 2007

Exchanges around the world are investing in new technology and services to attract a greater share of algorithmic and automated trading volumes. Chris Hall takes a whirlwind tour of the exchanges that are pulling out the stops to bring much-needed sources of liquidity to local markets.


Overshadowed perhaps by the efforts of US and European exchanges to attract algorithmic and automated trading volumes, competition between less established markets is equally fierce. Exchanges across all continents are updating legislation, upscaling technology and upgrading bandwidth to attract algo and auto traders that are looking to apply their models to new markets.

"There's a huge amount of investment going on in IT infrastructure by exchanges to increase message processing capabilities, and thereby increase volumes," says Yousaf Hafeez, Director of Product Development, BT Global Financial Services. "But the investment in technology has to be an ongoing process; the more information you can provide to algo traders the higher their volume of trading." Hafeez, responsible for exchange partnerships at BT, suggests exchanges' efforts should be focused on three key areas: low latency data feeds that supply every price tick; colocation services to further reduce latency; and use of pricing models that reward higher order flow volumes. "If they can embrace these changes, exchanges in emerging markets will soon see rising volumes on their venues from algo traders," he says.

Grigory Gankin, MICEX:

"We can see the real increase of algorithmic and automated trading with our own eyes."

Coming in from the cold

Russia is one market experiencing volume growth - and significant IT investment. The battle for algo and auto trading volumes is just one of the several fronts on which its two exchanges compete. Established in 1992, MICEX (Moscow Interbank Currency Exchange) is an integrated exchange that trades, clears and settles equities, corporate and government bonds, foreign exchange, money markets, commodities and derivatives. The largest exchange in Russia, the CIS and Eastern Europe in 2006 (MICEX reported daily average transaction volumes of RUR211bn (USD7.8bn)), technology has been key to the exchange's growth. More than half of the exchange's securities transactions are carried out via Internet-enabled broker systems; foreign financial institutions can trade on MICEX's equity exchange via GL Trade's proprietary order routing network.

RTS (Russian Trading System) has traded equities, bonds, futures and options since 1995, and migrated its eight most liquid blue-chip stocks to an orderdriven trading platform in November 2005. Orders are submitted to its anonymous electronic order book either through RTS's Plaza trading system or via the exchange's API (minimum order value USD5,000). The exchange's futures and options platform (average daily open interest 1m+ contracts) is also accessible via direct market access and RTS plans to move the rest of equities and bonds to an order-driven platform in due course.

Both exchanges have already attracted order flow from algorithmic and automated traders and are expanding services and capacity to taken on greater volumes. Grigory Gankin, Head of Derivatives, MICEX, says research into client technology (conducted as part of the exchange's IT upgrade programme) shows algorithmic and automated trading to be a driving force behind rising volumes. "We know well the technologies used by clients to trade on MICEX and can see the real increase of algorithmic and automated trading with our own eyes," he says. Daily volumes on MICEX have doubled to around 600,000 orders and 300,000 trades in the past year, with both orders and trades expected to rise by a further third by October. In response, MICEX is expanding its processing capacity to handle the higher volumes. Recent performance tests claim processing speeds of up to 1,500 orders per second, enabling MICEX to accept up to 1,000,000 orders and match up to 800,000 trades per day from a maximum of 500,000 clients.

Based on HP UNIX servers, the MICEX trading platform's cluster architecture is designed to quickly extend capacity and performance as well as the number of client connections supported, says Gankin, who notes that larger clients are improving trading performance by using a local gateway installation rather than connecting to the gateway located at the MICEX Trading Center. "This means that trading session information is instantly delivered to the client allowing him to make a quicker decision and submit an appropriate order, algorithmically or otherwise," says Gankin. To further improve trading efficiency, data from MICEX's technology infrastructure upgrade programme are being shared with clients. "We issue recommendations regarding minimum hardware configuration, gateway requirements, connection channel bandwidth, etc. in order to trade reliably, in real time, with minimum latency," says Gankin.

Dmitry Shatsky, Russian Trading System:

"…participants must either adapt their systems to process such high volumes of data or apply other trading approaches, …"

In response to the rise in message traffic that is accompanying higher levels of automated and algorithmic trading via direct market access, RTS is upgrading its data services, for example by offering variable depths of order book data and by enabling use of 'composed messages' via the RTS platform's messaging system layer. "This means that a message router can accumulate small messages over a specific period (for instance, 100 msec), then compose them all in one big message, remove the control headers and deliver one compressed message," says RTS vicepresident Dmitry Shatsky. "We have now at least ten times reserve of carrying capacity, but our participants must either adapt their systems to process such high volumes of data or apply other trading approaches, for instance working with the two best quotes instead of full depth of market."

In anticipation of further pressures to process a higher volume of orders more quickly, RTS plans to launch colocation facilities to market makers and refine its margining arrangements. "All our orderdriven markets already operate using online orders and position margining. As the markets further develop, we will switch to post-trade position margining. This will improve the scalability of our matching system," says Shatsky. Currently, RTS's electronic trading platform can process between 260 to 1500 trades per second depending both on the software and hardware options chosen by clients and whether the market is quote-driven or order-driven.

Rama Pillai, SGX:

"… we've seen a 30 per cent increase in connections … for algorithmic and automated trading since closing the derivatives trading floors."

Follow the sun

Automated and, in particular, algorithmic trading is well-established in a number of Asian markets, as reported by AT in January 2007 (Slow Burn or Damp Squib?), and growing volumes are driving change and investment in the region's exchanges.

David Sellen, Associate Director, Ecommerce Sales, Asia Pacific, Barclays Capital, sees a lot of growth, especially in the more liquid Japan, Australia, Hong Kong, Korea and Singapore markets both for equities and increasingly futures. But he cautions that any firms planning to launch into programme-based trading in Asia need to understand the many differences between the exchanges. "With exchangetraded products, traders understand the behaviour of a number of exchanges, but occasionally get caught out when they trade on an exchange they're new to. For example, the way the auction works and the volumes traded in Hong Kong is different from Japan and Korea. These features affect the efficiency of each algorithm, so we have spent a lot of time experimenting, measuring and tweaking our algorithms," says Sellen.

A further consideration is the different approaches of the exchanges to measures that protect the integrity and performance of their matching engines. "Exchanges can set limits on the number of orders per second or introduce order throttling if a member exceeds an orders-to-executions ratio," comments Sellen, who says such differences must be regarded as part of the landscape, "providing all members are treated the same and there is transparency around the rules".

SGX, Singapore's equities and derivatives exchange, estimates that 20 per cent of trading volume on its derivatives platform stems from algorithmic and/or automated trading models. On the one hand, the exchange believes it has benefited from a global growth in trading automation, quantitative trading and a growing appetite for Asian-based products. On the other, it has tried to keep closely in step with clients' trading preferences. "SGX now operates fully electronic derivatives and securities markets, and we've seen an increase of approximately 30 per cent in connections to the exchange by members for algorithmic and automated trading since closing the derivatives trading floors in September 2005," says Rama Pillai, Senior Vice President, Head of Intermediaries & Market Access, SGX.

With SGX currently experiencing an increase in the order-to-trade ratio of between 30 and 50 per cent, capacity is a key factor in the exchange's investment plan, although round trip times of 16 milliseconds reported by clients stand comparison with most exchanges. "We constantly monitor trade flow and increase our network capabilities as required. Our trading engine (designed by OMX) is a scalable model," says Pillai. "Our system capacity is multifold and expandable to accommodate increases in order flow; additional capacity can be deployed within a month."

Another example of growing algorithmic trading activity in Asia can be found in the derivative warrants market on HKEx, Hong Kong's equities and futures exchange. Market makers are required by the exchange to supply liquidity "by means of continuous quotes or quote request" throughout trading sessions. As such, market makers use algorithms to distribute continuous streams of quotes in response to requests from market participants.

Described in its 2007-9 strategic plan as "essentially an IT-based enterprise", HKEx lists development of a "more user-friendly market infrastructure in order to attract overseas securities firms" as a priority. AMS/3, the exchange's equities trading platform, can handle 1.5 million trades per day within its benchmark response time, i.e. orders executed within a mean of one second. AMS/3's Open Gateway device supports automated and algorithmic trading by brokers and also investors through 'broker-supplied systems', i.e. trading facilities developed by exchange participants to capture investors' electronic trading requests.

David Sellen, Barclays Capital:

"…traders occasionally get caught out when they trade on an exchange they're new to."

One of the ways in which SGX is attracting further algo and auto volumes is through direct market access. Indeed, it claims to be the only exchange that allows trades to be entered directly into its trading engine without first having to pass through the pre-execution checks. The exchange is looking to extend its DMA capabilities by supplying colocation solutions to members. According to Pillai, this is just one element of SGX's investment programme. "We are in the process of renewing our IT infrastructure and capabilities for trading and clearing to raise overall efficiency and enhance service quality," he says.

A fair wind for investment?

Barclays Capital's Sellen says enabling remote access to the exchanges by non-domestic market participants for algorithmic/automated trading purposes is taking off in Asia. "It's surprisingly common. We are working with a number of clients to offer remote access facilities and we're seeing an increase in flows to those exchanges with well defined policies in this area." In Asia and elsewhere, remote access often requires legislative as well as technological change, but this is seen in developing markets as a necessary step toward greater inward investment and financial sector growth. "A lot more exchanges around the world are enabling remote access because they know that hedge funds (and other high volume traders) cannot build trading operations in every country in the world," says Hafeez of BT Global Financial Services. "If an emerging market starts to allow remote trading, it encourages traders to start sending order flows direct to that venue, rather than through a broker."

Operational since 1998, MexDer, the Mexican Derivatives Exchange, has enabled non-domestic participants to access its trading applications and market data services via BT Radianz since November 2006. The service was launched following the passing of new legislation aimed at supporting liquidity on the exchange that permits foreign investors greater trading flexibility and allows MexDer to sell futures contracts to non-domestic trading firms. Brokers and traders globally can access MexDer market data and Mex-FIX, a proprietary API for order generation, via BT Radianz's proprietary network. In a press release announcing the deal, Jorge Alegria, chief executive officer, Mexican Derivatives Exchange, explicitly invited non-domestic players to "to diversify their listed derivatives product offering."

BT's Hafeez asserts that many Latin American financial markets are positioning themselves to command greater auto and algo trading volumes. "Two or three years ago Mexico would not have been seen as a large derivatives exchanges; last year it was in the top ten by contracts traded, largely due to algo traders moving into the market," says Hafeez. "Brazil's Bovespa equity index has risen from around 11,000 in 2002 to 39,000 in 2006 and a lot of this is to do with them making it easier to trade."

Indeed, ease of use should be part of any exchange's mantra, according to Hafeez. "Exchanges that want to attract more volumes need to make life as easy as possible for the algo trader. This means having systems they can easily integrate into - creating feed handlers can often be a very painful process - supporting protocols such as FIX, and offering data feeds that give every price tick," he says.