The Gateway to Algorithmic and Automated Trading

Enter the dragon. How bleak trading in China heralds a bigger focus on algos

Published in Automated Trader Magazine Issue 27 Q4 2012

It's been a grim year for human traders in China, as volumes and prices on Chinese exchanges have fallen to multi-year lows. Yet the challenging environment is also driving firms to look for efficiencies - and that spells an increased focus on algorithmic trading. The use of algos may be starting from a very low base, and the market may have its share of regulatory constraints. But there are plenty of signs that automation will only increase as traders grow ever more sophisticated. Mary-Ellen Barker reports.

From Shanghai, Hongjie Qu, head of China electronic trading for UBS, has been watching the mainland China market hitting new lows almost every day. Still, it isn't all bad news.

"It could be positive in a way," Qu said. "When the market really turns down, institutions pay bigger attention to trading costs, impact, and the quality of execution. So from that point of view it's good for brokers like us who can offer them better solutions."

Hong Kong, the world's second largest exchange operator by market value, said in August that average daily turnover had fallen 23 percent from a year ago.

On the mainland the numbers were not as dire, but were still weak by any standard. On the Shanghai Stock Exchange, number six in the world rankings for domestic market capitalisation, domestic market cap fell 14 percent in the first half of 2012, and by the end of August the daily trading value of $5.6 billion was the lowest since December 2008. The value of the Shenzhen Stock Exchange, number 10 in the domestic market cap rankings, fell 10 percent in the first half.

Banks that staffed up to capture the benefits from strong growth in Asian developing economies, driven particularly by China, are cutting jobs on trading desks. Deutsche Bank was the latest, shedding 10 percent of its Asian equities sales and trading staff in early September. The Financial Times forecasts more cuts ahead for other banks.

Hua Zhang

Hua Zhang

"If the future continues down this path, with lower liquidity, lower volumes and less revenues and profits, then algorithms will become more important for people trying to automate workflow, so they can do more with less," said Greg Lee, who is based in Hong Kong for Deutsche Bank and is head of the bank's Autobahn Equity electronic trading service for the Asian region. "And if markets pick up, people are going to look to algorithms to gain an advantage, to be more efficient."

Based on current levels, there is plenty of room for algorithmic trading to grow even while turnover shrinks - especially in mainland China. Market participants and experts say algorithmic trading there is gathering steam, but is still at a very early stage of development.

Hua Zhang of financial services consulting firm Celent, who is based in Beijing, estimates that algo trading now accounts for less than 1 percent of securities trading in China. He forecasts growth to about 2.5 percent in 2013. Algo-based trading is now concentrated in futures, where it accounts for 10 percent or more of volumes.

Qu agrees that algorithmic trading in China is increasing. "In the past two to three years algorithmic trading has come from mystery to a real and known practical trading tool in China," he said. He estimates that the dozen or so large mutual funds operating in China now execute around 30 percent of their volumes using algorithmic trading, accounting for about $0.4 billion a day of transactions in A shares - the Chinese share class quoted in renminbi and open only to Chinese investors, or foreign investors licensed under the QFII (qualified foreign institutional investor) scheme. B shares are quoted in foreign currencies and are open to foreign investors.

He expects institutional trading to reach 30 percent of volumes over the next two or three years, and algorithmic trading to account for more than a third of their A share flows. If overall daily volumes return to a more 'normal' level of $35 billion from the current $25 billion levels, daily algo trading volumes should hit $3.5 billion a day by then.

Lack of liquidity has hampered the growth of algorithmic trading in available fixed income instruments, which include bonds, repos and convertibles, he said.

But where they are used, styles of algorithms in operation are increasingly sophisticated. "UBS rolled out China's first algorithmic trading services in late 2010, so our mutual fund clients have now been using algorithms for about two years," he said. At the beginning, they would use mostly basic algorithms such as TWAP and VWAP; now, requirements are for much more sophisticated and customised varieties.

Hongjie Qu

Hongjie Qu

The T+1 blues

Growth is constrained by a lack of diversity in financial instruments, along with tight regulation, according to Zhang of Celent.

The Chinese stock market is heavily dominated by retail investors, who account for about 70 to 80 percent of transaction volumes. UBS's Qu said Chinese mutual funds account for about 20 percent. Only around 1 to 1.5 percent is from foreign investors under the QFII scheme.

There is little high frequency trading, market watchers say. In the cash equities market, the biggest obstacle is the 'T+1' rule, which forbids sale and purchase of a stock on the same day.

"Clearly, when you think of high frequency traders and what they do in and out of various positions, that rule is not going to help you," dryly noted Dr Richard Bentley, vice president of banking and capital markets at Progress Software. The same restriction does not exist for futures, which are also exempt from the stamp duty of 10bps which applies to stock sales.

But some in the market are preparing for the arrival of HFT. China's largest investment bank, CITIC Securities, has bought Progress's trading platform. "One of the reasons we won that business was because they were specifically looking for a platform which was suitable for the demands of HFT in the future," Bentley said. "A lot of the bigger firms in China take a pretty long view around this because of the pace things happen."

Tanuja Randery, the chief executive of MarketPrizm, which provides low latency market data and order entry infrastructure, said she has noticed "stirrings of interest" from her client base for trading in mainland China, but that most were content to watch from the sidelines to see how things develop in terms of foreign access.

Randery noted that exchanges on the mainland have not got the equipment in place to allow HFT.

"They do not offer high-speed engines, and the reason they do not is because of this retail market. They're concerned that if there are any glitches, that will cause real issues, real repercussions."

Also, there is none of the market fragmentation in mainland China that has helped drive some forms of algorithmic trading elsewhere in the world. There are no dark pools, and short-selling and margin trading are tightly restricted.

Tanuja Randery

Tanuja Randery

Chinese markets are gradually opening up, but the pace of change is slow and careful. "The government and the regulator do not want to see stock volatility," said Zhang.

Expectations about the pace of change have faded a little in the face of market performance, according to Bentley: "Up until a year or so ago, with volumes increasing and a huge focus on the Chinese market, people were a bit more bullish about some of these regulations changing in the near term, in the next two to three years," he said. "The slump in volumes has wound expectations back."

"Anyone who gets impatient looking at the Chinese regulator really should just give up and move on," said a senior broker at a major international bank who did not want to be identified. He is not champing at the bit. "To their credit, they've been very prescriptive in how they've done things, and very intentional in the changes they've made to open up the market and try to protect their retail investors, which is what they're all about."

China loosened requirements for QFII scheme investors in June, though hedge fund managers are still excluded. The system sets quotas for foreign investment, and the overall size of the quota was raised to $80 billion from $30 billion in April.

Trading has to be done via local brokers. That is also one factor that has slowed the pace of adoption of algorithmic trading, said Rob Laible, head of Asia ex-Japan cash sales trading at Nomura, because it makes trading electronically into China more complex.

"Most Western broker dealers that do have all the algos, and do have [access via] QFII don't have an onshore licence, so they would need to partner with a local broker to provide access. As soon as you start doing that you definitely start layering on more costs and less reliability, because you're not controlling the infrastructure from start to finish."

Qu said most larger institutional clients in China operate 'point-to-point' - they trade everything via Direct Market Access, using their own trading teams and their own trading platforms.

In other markets, competition between brokers for better execution has helped spur development of algorithmic trading tools and technology. "In China, institutions like mutual funds and big insurance companies are operating in a more isolated environment, so there's no contribution from brokers in terms of trading technology and execution services," he said.

Another factor cited by both Laible and Hua Zhang is the small size of the quantitative fund management industry in China, which did not get underway until 2010 when index futures and restricted forms of margin trading and short selling were introduced.
The growth of the sector is constrained by the T+1 rules, the shortage of hedging instruments, and relatively high costs of stock borrowing. High-touch cash sales trading desks can handle most of the volumes from Chinese fund managers, Laible said.

Rob Laible

Rob Laible

One counterweight pressure, though, is the reliance Chinese mutual funds put on relative performance rankings against each other, rather than benchmarking against an index. If funds can cut trading costs using algorithms to climb the rankings tables, algorithmic trading becomes much more attractive. "So in this kind of situation, where the market isn't good and they have bigger difficulties raising money, they care more about technology," Qu said.

In August, Chinese regulators said they will extend their pilot securities lending and margin trading programme, begun in 2010 to allow brokers and investors to use their own stocks as collateral. The new rules allow them to borrow from banks, funds and insurers, and lend money and stocks to clients for margin trading. Those changes are expected to help boost liquidity and increase the sophistication of the Chinese market.

"Definitely the regulator is very cautious about the stability of the stock market in China, especially in terms of adopting advanced technologies - for example HFT," said Qu. "But I don't think they're slow in opening up to new products or technologies; actually, they're quite fast in my view, as evidenced by the recent rollout of index-futures, margin-trading, cross-border ETFs, et cetera." He, too, says mainland China regulators are carefully walking the line between opening up markets and protecting retail investors.

Automating arbitrage

In mainland China's commodities markets, the adoption of algorithmic trading is also gathering pace. Mainland China commodities investors are developing automated trading strategies to arbitrage local commodities markets against London, Chicago and other offshore commodities exchanges, according to Jeremy Goldwyn, head of Asia business development at futures broker Sucden. "It's not a pure arbitrage as we see in some commodities. It's much more a way of reducing risk, a spread trade."

Jeremy Goldwyn

Jeremy Goldwyn

Goldwyn, who is based in Hong Kong, said in the last 18 months he has seen several clients build software to help human traders handle that arbitrage trade. "The next stage on from that is obviously more of an algorithmic system that takes that sort of information and trades automatically, and we're beginning to see those this year," he said.

It is only a handful of clients now, but the trend is accelerating. "It may not be high frequency trading as we see it in other markets, and it varies a lot from client to client, but it is growing very rapidly."

Chinese mainland clients use Sucden for the offshore leg of trades. The push towards algorithmic trading started in metals markets, with initial interest in Shanghai copper against LME.

But investors have shown a lot of interest in agricultural commodities this year, along with rubber, precious metals and base metals. "It's out of the embryonic phase," Goldwyn said. "Now it's in the toddler phase, but on an accelerated learning programme."

Most market-watchers predict steady rather than explosive growth in algorithmic trading in mainland China markets, as institutional investors battle to balance the drive to cut execution costs against the valuable soft-dollar services, such as technology and research, provided by the sell-side.

It is a struggle that may be more challenging in Asia than elsewhere. Consulting firm Greenwich Associates notes that Asian institutions have a relatively heavy dependence on research provided by the sell-side. It says a typical institution in Asia uses 67 percent of the commissions it pays to brokers on trades of Asian equities for research and advisory services, above the 55 percent in the European market and 59 percent in the U.S.

But whether slow or explosive, the trend is clearly to further growth. Deutsche Bank's Lee sums up: "There isn't a downside from the electronic point of view - it's not a new fad, it's simply how the market has evolved, and it's core to every part of trading now."