The gradual adoption of electronic trading by exchanges in Malaysia, Indonesia, Thailand and the Philippines presents specific challenges for brokers in the construction of algorithmic trading platforms suited to the four ASEAN markets.
Characterised by wide bid-offer spreads and large numbers of illiquid stocks, these markets require algorithms that are designed to cope with these specific traits and hence will behave differently from those in more developed markets like Singapore and Hong Kong.
“The regular algorithms do work but there are always tweaks for the different market dynamics. The algorithm needs to understand the trading speed of the market in terms of your time to fill; it needs to be much more patient in these markets,” says Greg Lee, head of Autobahn equity, Asia at Deutsche Bank.
Deutsche Bank in 2009 rolled out its Stealth algorithm globally. Designed to ”silently' execute difficult orders in over 30 markets around the world, including Asia, the liquidity seeking strategy trades when it's optimal to do so.
“Algorithms like VWAP use historical volume curves to determine how to allocate today's trading. We can make them somewhat dynamic and look at today's volume but the problem comes when a stock trades 10 million dollars one day and 100,000 dollars the next day. These sorts of fluctuations in volumes and how a stock trades are the things that make those algorithms sometimes unsuited to those stocks. So you need to look at the more practical algorithms that look at spread and liquidity,” Lee adds.
On the Malaysian, Indonesian and Thai exchanges, algorithmic trading volumes as a percentage of overall trading activity have risen steadily from a low base and are expected to begin soon in the Philippines. The Stock Exchange of Thailand is in the process of selecting a shortlist of vendors to supply a new trading engine that is expected boost capacity of the exchange by 100 times when implementation is completed in two years. Bursa Malaysia has also seen a steady increase in the adoption of algorithmic trading following the migration of its derivatives products onto CME Group's Globex electronic trading platform in September 2010. In Indonesia, UBS become the latest broker to launch algorithmic trading in March, while buy-side block crossing network Liquidnet added Malaysian and Indonesian stocks to its platform in November 2010 and January 2011 respectively, bringing to eight its number of markets covered in the region. Liquidnet's average execution size in Malaysian stocks was US$1 million in Q1 2011, while Indonesia recorded an average execution size of US$1.1 million since launch on 25 January 2011.
Keith Ducker, Tora's chief investment officer, adds that although the four ASEAN markets have some common characteristics, they must be viewed independently from a development perspective. “These markets share similar challenges in that they have a strong retail bias and volatility. They don't have a typical distribution of volume across the trading day.
Stocks routinely trade on spikes of volumes and that's hard to model. It's a chicken-and-egg scenario. You need VWAP adoption to create that normal distribution but you're not going to have it until it is an accepted product.
To develop a volume-based algorithm, you're going to have to expect a lot of volatility within the results of that algorithm. There's nothing that says you can't do it, but your expectations have to be different from that of a more mature market like Tokyo,” he noted.
For Glenn Lesko, CEO, Instinet Asia, although the use of algorithms in ASEAN will always differ from larger Asian markets, electronic trading is already effective in an increasing range of circumstances. “It's not going to go up to 100% algorithmic trading in these smaller markets. But people can trade the top 10-20 names and selectively trade other names with algorithms, and we are already doing that for clients. The only strategies that we can't implement are those that interact with other off-exchange sources of liquidity,” he said.