Advantage: Algos

Algorithmic trading techniques have gained strong traction in Asia in the past couple of years. How have these strategies performed during the current bout of market volatility, and what's the future outlook? By Pamela Tang

Equity markets became volatile in Asia at the turn of the year, which has prompted many investors to park their money in cash. But a growing number of market participants believe the turbulence requires the deployment of aggressive trading techniques, as long as there is sufficient liquidity to provide for exit strategies.

Whatever the market conditions, however, proponents of algorithmic trading - trades that use a computer algorithm to determine the timing and price of large orders - say its performance in the first few months of 2008 has proven strong.

"Algorithmic trading has done fine in the past few months," says Kevin Chapman, head of trading at Nicholas-Applegate Capital Management (NACM), an investment management firm based in San Diego. "Algos are better in difficult markets, because I would rather have them systematically executing my strategy than have a broker trying to take a view and inputting my price. He (the broker) may think that might improve my prices, but often it doesn't work as well."

NACM, which had used algo trades for about 10% of its Asian trades, has seen its use increase by 20-30% in the last few months.

Chapman says another advantage of trading strategy is that it allows for greater confidentiality. "I would rather not have us showing our hand in difficult markets," he says. "If I give a broker a big order, and it's showing on their system, others might start making calls on it."

David Sellen, director for e-commerce at Barclays Capital in Singapore, appears to agree. "On the whole, we've seen algos perform surprisingly well," he says. "We're not seeing people running back to broker intervention and manual execution in these environments. There's a level of confidence building in the way these algos work."

Sellen points to Barclays Capital's automated real-time execution system (Barx), which offers enhanced volume-weighted average price (Vwap) strategies with directional indicators. He says the service has outperformed in the past three months and prompted more demand for algorithmic execution.

Algo application

But it is not always so simple to make trading decisions. "If you don't know what you want out of the trade, then the algo doesn't do it for you," says Gavin Williamson, head of connectivity for Asia-Pacific at HSBC in Hong Kong.

Barclays Capital's Sellen adds: "There are some strategies out there that are dependent on presumptions that don't necessarily apply in a volatile market. Do you want to risk trying to tease out hidden liquidity when the markets are extremely volatile? The focus is on moving your position and proving you did that efficiently. The risk of failing to complete your order is much higher than any potential gain." Barclays Capital's clients prefer the dealer to keep trades very simple in the current market conditions, he says. "That's producing real results," adds Sellen.

Bill Capuzzi, president of G-Trade, the global electronic trading unit of BNY ConvergEx Group in New York, seems to have observed the opposite. In the past six to nine months, he says, demand for algo trades beyond those offering Vwap, time-weighted average price (Twap) and implementation shortfall have increased from his clients, which are mostly hedge funds. "Our clients have been asking questions about differentiating algos," says Capuzzi. "The more sophisticated hedge funds are looking for competitive advantage and for ways to mine liquidity within these markets and are expanding beyond the three basic benchmark type of algos."

While there has been escalating interest in algo trading in Asia, adoption rates are still considerably lower than in Europe and the US, say market observers. They believe algo trading accounts for 3-5% of volume in Asia, while direct market access (DMA) - which allows clients to send orders electronically to a market without manual intervention when the client is not a member of the market - is responsible for 15-20%. Several parties believe algo trading will reach 10-15% of market share within the next few years, with the most advanced electronic trading markets being Japan and Australia, followed by Hong Kong and Singapore.

Monthly volume at the Australian Securities Exchange (ASX), for example, has grown from around two million trades to as high as eight million from mid-2005 to March 2008, while average trade size fell from about A$35,000 a trade to below A$20,000 per trade over the same period. The rising trade count and the falling trade size suggest more algorithmic trading is occurring in the ASX market, says David Stocken, manager for trading and derivatives at the exchange in Sydney.

In the past year, trading volumes on the MSCI Singapore Index - the domestic equities contract on the Singapore Exchange (SGX) - have grown by 81% and transactions in the cash equities market have doubled in terms of average daily value. "Some of this can be attributed to traders arbitraging the domestic derivatives contract and components in the underlying cash market," says Rama Pillai, head of intermediaries and channels at SGX. "Based on increasing enquiries from customers, we see that this is just the tip of the iceberg in terms of the full extent of algo trading that could come to our marketplace."

While technology has been a key driver for algo trading, dealers say some first-movers in bringing the technology to Asia have experienced growing pains. Subsequently, as the technology became commercialised, algos were deployed more easily and cheaply, as mistakes were ironed out.

NACM's Chapman says technology has caused trading speeds to increase considerably. This is particularly significant now, since phoning a broker could mean a missed opportunity, as others take advantage of that window. "When you've had those experiences, you'd rather have a smart algorithm that is programmed to stay on the bid side, but knows to take it should a share show up at a certain amount on the offer," he says.

Dealers say algos need to be tailored for the market to work well, given the peculiarities of the different markets. Australia, for example, has a staggered open, so algos need to be programmed to participate in that, as well as take into account the lunch break and second sessions.

The closing auctions that occur in certain markets in Asia also present challenges, because each market differs in the way it performs. Hence traders using algos need to make sure they can opt in or out and are getting the liquidity they are looking for.

However, one of the most significant challenges to electronic trading in Asia is the wide bid/offer spreads at the exchanges. In May 2007, Citi research found that there was a 13 basis point spread on trades on the ASX 200, 15bp on Hong Kong's Hang Seng Index, 20bp on South Korea's Kospi and 57bp on the MSCI Singapore Index. These figures are higher than those in Europe and the US.

"We are affected by wide spreads just like everyone else," says NACM's Chapman. "There is just no reason why the difference between bid and ask is 20 to 50bp or more. It used to be the case that spreads were wide because brokers would get compensated on that, but that ended with decimalisation. The brokers' fees are fixed based on the value traded, so the spreads should be smaller."

HSBC's Williamson believes some traders do not understand the market structure, which is why they are stumped by wide bid/offer spreads. "A lot of brokers just brought their algos out from Europe, with no concept of the market structure in Asia, including bid/offer spreads," he says. "Some algos were designed by people with a lack of understanding of the market structure in Asia. A number of these guys got burned because of that."

Robert Zhang, Asia head of equity core sales and trading application development at Morgan Stanley in Hong Kong, says the wider spreads underscore how electronic trading still has some way to go. "Most exchanges here have wider spreads and more stocks with less liquidity, which means investors are still comfortable placing their orders through human traders to capture the alpha with the right price," he says.

Some are also concerned about latency in Asia, which lags that of Europe and the US. The wide geographical spread in Asia - which creates trading bottlenecks - contributes to that, say observers, as do varying technologies and regulations at the different exchanges.

Latency concerns

However, dealers say the latency issue is most significant for certain market participants, such as statistical arbitrage traders. And Zhang feels latency is less of an issue than liquidity. "Everyone knows that latency is an issue in Japan, but they still invest there, because the spreads are driven narrower and liquidity is there," he says. "The main factor is cost of trade, not latency."

NACM's Chapman, who has traders working Asian hours, also believes latency is not a critical issue. "Our traders try to keep the latency as low as possible. They are monitoring the orders constantly, so they'll trade where they feel it will benefit us the most and reduce costs and latency," he says. "I don't see (latency) as a difficult problem, since it has improved greatly in the past few years due to technological advances with order management systems."

Market observers say exchanges are aware of the issues and have already made moves towards becoming friendly to electronic trading. The top exchanges have made inroads in terms of achieving narrower bid spreads, while upgrading technology for lower latency. Some markets, such as Malaysia and Thailand, have also opened up to DMA. Market observers say exchanges have realised they need to adapt to survive in the face of increasing competition from alternative trading venues (see box, below).

But algo trading and DMA still face considerable regulatory obstacles in countries such as India from authorities that want control over the market and are wary of foreign investments. Foreign brokers wanting to trade algorithmically or via DMA in India must send their orders to local brokers offering the relevant transaction, which adds latency to the process.

Some regulators are also wary of statistical arbitrage funds, due to concerns that they might have too much impact on markets. But Chapman holds the opposite view. "Regulators have to realise that statistical arbitrage is not bad; that even though prices flicker a lot, you get a lot more liquidity, as the short-duration alpha players come in," he says. "They narrow the spreads, create more liquidity and prices move faster, and that reduces cost for everyone."

Market participants are confident that regulators are open to change. They cite positive examples of regulations in Australia, which were relaxed to allow more seamless electronic trading. "There are definitely challenges, but they are being overcome, as more and more volume is being traded electronically in Asia every day," says Capuzzi. "Over time, it'll become easier and easier for the world to trade in Asia."

The ASX is cited by many for significantly narrowing spreads and reducing latency. There is a one-cent tick-size increment for all stocks over A$2. Latency has also dropped to 24 milliseconds after the ASX moved to its integrated trading system for trading of cash equity products. Stocken says the exchange has achieved this lower latency while its trade count has risen aggressively over the past two-and-a-half years.

SGX revised the minimum bid sizes for its securities market in December, with the aim of improving trading efficiency and market liquidity. As a result, minimum bid spreads are now smaller. "One of our brokers serving algorithmic trading clients has feedback that the Vwap performance has reduced significantly from -20bp to -4bp after the revised minimum bid spreads were introduced," says Pillai.

HKEx upgraded its trading system on January 28 and can now support up to five million trades a day. The system's average response time has dropped to less than 0.5 seconds from about two seconds before the upgrade - this facilitates various order types in algo trading.

Other Asian markets will no doubt be watching such moves closely, as algo trading gains momentum in the region.

Challenging exchanges

While exchanges are likely to be the main mechanism for price discovery in the next few years, they are seeing their traditional iron grip loosen in the face of developing competition.

Europe and the US have seen the proliferation of alternative trading venues, such as electronic communication networks (ECNs). Such price-formation venues operate on a model similar to an exchange, creating a central venue where buyers and sellers can meet in an open order book to trade with each other. They are attractive to traders because of their lower cost.

According to Hong Kong-based research firm and brokerage ITG, competition has triggered a drop in trading costs in the markets where it is most prevalent. Research shows that in the US over the past five years, where alternative trading venues have proliferated, average trading costs have declined from 135 basis points to 38bp.

Some of these alternative trading venues have been set up in Asia, but they don't appear to have taken off strongly. Gabe Butler, head of sales and trading at ITG, believes the long-term outlook for such venues is good. But short-term prospects are not so bright, he says, as broker desks in Asia still control much of the trading flow. There are also other challenges, in the form of multiple time zones, currencies, regulatory structures and stages of economic development.

But some very sophisticated investors are attracted to using such alternative venues. "If we have control of the order via an ECN, DMA (direct market access) or alternative trading system, we can pull the order out much faster than the broker, in seconds - and that makes a big difference," says Kevin Chapman, head of trading at Nicholas-Applegate Capital Management (NACM), an investment management firm based in San Diego. "We can act quickly if we see rumours coming out of the market. It encourages us to use these newer venues, and as more people do, they'll see the benefit of it and continue to do so." About 10-20% of NACM's Asia trade is executed on these platforms, and the company wants to increase this figure.

The other development is the rise of price-taking venues, such as crossing networks and dark pools of liquidity. Such venues let buyers and sellers come together anonymously to match up large blocks of buy and sell orders without 'leaking' information to the market and thus moving the price significantly. This benefits large institutions, such as Australia's superannuation funds, as they do not get penalised in terms of cost when trading in large sizes. It also helps them to avoid moving the market price in a big way.

Some see dark pools as an upcoming thing that have yet to really gain much foothold in Asia. Understanding of the ECN-type of environment will increase in Asia, which will pull some of the liquidity off the primary exchanges into the dark pools, says Bill Capuzzi, president of G-Trade, the global electronic trading unit of BNY ConvergEx Group in New York. G-Trade will trade on the dark pools when liquidity appears, but that 99% of the liquidity is still on the primary exchanges, he adds.

Gavin Williamson, head of connectivity for Asia-Pacific at HSBC in Hong Kong, is confident dark pools will take off. But, because liquidity is a big issue in Asia and initially liquidity will fragment, providers will need several success stories to change trading patterns. He sees commercial organisations exploiting these business opportunities by building smart order routing and access to these liquidity pools.

Morgan Stanley is already in the game. It set up a dark pool in Japan in May 2007 and is planning to go live in Hong Kong this year. "Our biggest challenge is to integrate all the flows that go through there," says Robert Zhang, Asia head of equity core sales and trading application development at Morgan Stanley in Hong Kong. "Right now, it's moving at a slow pace. It's not that people don't want (these options), but it takes time to integrate the systems to take advantage of the dark pools.

"From the day we went live, volumes have grown five times, and we can see we're really providing liquidity in there," adds Zhang.

NACM's Chapman backs up that view. "Now when we're trading in Hong Kong, we're trading wide spreads and would love to trade at tighter spreads," he says. "If a crossing network can allow that, we can immediately save up to 20-25bp."