Today, about 6% of assets under management in China are invested using algorithmic trading strategies. > Song
All you have to do is register, get information on a certain strategy on your phone, buy the one you like and start trading. > Li
Regulators in some countries are concerned that automated trading could disrupt the market. > Liu
What people should know is that in order to hit that right timing once, they will fail many times. > Yong
Quantitative trading, which is gaining traction in several Asian countries, is set to make inroads into the local market in the near future
Malaysian retail investors will soon be able to use algorithmic trading strategies to boost their investment returns with the introduction of platforms that use a combination of mathematical tools and technology to seize opportunities in the market faster than human traders can.
With the popularity of financial technology (fintech) growing fast, some individuals who used to work in banks have branched out to start their own algorithmic trading business. Two of them are Song Nian Hui and Li Guang Zhen, both from Smart Trade Inc, a fintech start-up based in Japan that aims to offer retail investors algorithmic trading strategies. Both were colleagues at a Japanese bank in the US before they co-founded Smart Trade. Back then, Song was involved in creating quantitative trading tools while Lee was a derivatives trader.
“I used to develop algorithms for professional traders to help them trade and manage risk. Many people outside the industry don’t know who we are and what we do. So I thought, ‘why don’t we go out and raise funds, build our own tools and provide them to the general public?’” says Song.
Li had the same thought. He says both of them attended a fintech programme in Japan and later, decided to set up their own start-up to provide algorithmic trading strategies to retail investors. “Algorithm use was limited to financial institutions then and the public either had no access to it or doesn’t know what it is. So I thought it was an opportunity for us to start it off.”
Meanwhile, Chris Liu, the founder and CEO of Tixguru, also decided to follow the fintech trend by launching a start-up. Like Song and Li, his aim is to develop more tools and algorithms to help him trade on the market and at the same time, extend the products and services to financial institutions and retail investors to generate revenue.
“The fintech trend gave me the idea of starting our own company. I know many traders in several countries and we are working together to develop deep learning algorithms (a type of algorithm strategy). Each of us is in charge of investing in different markets on behalf of our clients,” Liu says.
Song says in recent years, algorithmic trading has been gaining traction in several Asian countries, particularly China, Li’s and her homeland.
“Today, about 6% of assets under management in China are invested using algorithmic trading strategies. These are the people who do not have time to monitor their trades every day. Things are picking up very fast,” says Song.
Liu says this trend is growing in Asia. This is due to the busy lifestyle of city people and the lacklustre performance of mutual funds and unit trust funds in recent years, he adds.
“These people are looking for higher returns but do not have the time to monitor their trading daily. So, this is where algorithmic trading comes in. We can make recommendations based on our algorithm and their trades will be executed when the algorithm sends them a ‘ping’,” says Liu.
While algorithmic trading is not new in developed countries, especially the US, it is still difficult for retail investors in emerging markets to access the technology.
This is because many local exchanges do not allow or encourage the practice. Also, it requires an initial investment — the purchase of a computer and software. Investors also need to acquire the necessary knowledge to formulate an algorithmic trading strategy and test it using historical data.
More importantly, hedge funds and financial institutions are able to invest heavily in trading machinery and hire software engineers to formulate sophisticated investment strategies. Retail investors, on the other hand, lack of resources to do the same and will lose out to large firms in terms of trading speed and strategies.
Examples of algorithmic trading strategies
There are various forms of algorithmic trading strategies. A simple example is the five-day and 10-day moving average, says Song.
“A typical [algorithmic] trading strategy is the use of the double moving average. For example, we smooth out a five-day and 10-day moving average line. When the five-day moving average moves up from the bottom and crosses the 10-day average, it is a strong ‘buy’ call and vice versa. This is a very simple example that requires only 10 lines of coding,” she says.
“The trade order will be created and executed automatically by the computer.”
Quants can also derive different strategies by combining different technical indicators. Besides the moving average, the other indicators include stochastics, moving average convergence divergence and Bollinger Bands.
Apart from the back-tested algorithmic strategy, there are also other strategies that can carry out arbitraging and pair trading on different exchanges.
Investors can also use deep learning algorithmic trading strategies to trade stocks. In this case, quants install stock and market historical data and various algorithmic trading strategies in a self-learning computer. The computer will then be able to forecast movements of stocks and markets and provide recommendations.
“Based on the forecast, we will make a recommendation to brokerage firms [we partner with] on what stocks to buy in the next one week or so. The brokerage firms can also provide this service to their clients, depending on the rules and regulations of the respective countries they are in,” says Liu.
Some hedge funds have been applying deep learning algorithm strategies to trade different asset classes and on different markets to obtain higher returns. One example is Two Sigma, a hedge fund company that adopts a scientific method to investment management.
Two Sigma, with US$45 billion worth of assets under its management, acquires data from more than 10,000 public and private data sources and deploys 95,000 central processing units with 1,695 terabytes of memory, according to its website. It also proudly shows that 72% of its employees are from non-financial backgrounds and two-thirds of its employees work in its research and development department.
In January, the company made it to the 20 best-performing hedge fund managers of all time index, according to The Financial Times. The index, compiled by LCH Investments, measures the total amount (in US dollars) made for investors by a particular hedge fund since its inception.
A Financial Times article last August said the underwhelming performance by hedge funds is among the factors that have contributed to the growing popularity of algorithm trading platforms. Some industry observers believe that “this new wave of amateur algorithm traders can be harnessed using crowdsourcing techniques to disrupt one of Wall Street’s elite professions — the hedge fund manager”.
“If our model is successful, there will be no need for hedge funds anymore,” says Martin Froehler, an Austrian mathematician who created Quantiacs, one of several online platforms for DIY algorithm traders.
“A smart guy with a laptop will be able to start his own hedge fund. It will be very challenging to the big incumbents. A very simple idea can prove very powerful.”
Gaining access by year end
Smart Trade and Tixguru are looking to offer their products and services to Malaysian investors by the end of the year.
Smart Trade’s Li says the company is set to launch its mobile application in October. The app will allow retail investors to purchase different algorithmic trading strategies and execute them via brokerage firms.
“All you have to do is register, get information on a certain strategy on your phone and buy the one you like and start trading,” he says.
While the details of the app are not out yet, Li says it will only allow its users to trade equities in China and Japan as the company has established partnerships with some brokerage firms in both countries. The Chinese and Japanese partners will execute the trades of the app users automatically.
The average purchase price of the strategies is US$30. This could go up to US$200, depending on how sophisticated the strategy is.
“It does not necessarily mean that the more expensive the strategy, the higher the return. It would only mean that it has a better risk-adjusted return,” Li says, adding that the app user will also have to pay an additional fee of about US$18 per month for the more sophisticated trading strategies.
“If the creator of a trading strategy sets his price high and does not provide sufficient details, we will not allow him to sell it via our app.”
Smart Trade has an online platform that allows quant traders to create and use their own algorithmic trading strategies, which they can also sell to other traders and investors. As at July, there were more than 100 strategies on the platform.
“The quants generously share their trading strategies online for two reasons. The first is that they want to show their talent and build a reputation. The other is because some of these strategies require a very large capital to execute. Thus, they show it in the hope of pooling investors’ funds in order to trade,” he says.
Tixguru is partnering one of the largest local banks to offer the bank’s clients algorithmic trading services.
Liu says the company will provide the bank with stock recommendations by using its deep learning algorithms. The bank will, in turn, offer them to its clients.
He says the firm will not push the product directly to investors because that will require them to obtain a fund management licence and, on top of that, take all the necessary measures to adhere to the country’s trading rules and regulations.
Unlike Smart Trade, Liu says, local investors who use Tixguru’s algorithm trading strategies will have to execute trades manually through brokerage firms after they are prompted by the ‘buy’ or ‘sell’ signal. This is because there are separate rules for automated trading. Besides that, the brokerage firm will have to support such a function.
“Regulators in some countries are concerned that automated trading could disrupt the market. For instance, the computer could create a lot of buy and sell orders on the market in a short period and cancel them subsequently. This could create a false sentiment on the market and, therefore, cause disruption,” Liu says.
He notes that brokerage firms have to take into account the possibility of investors getting upset with them when they lose money through automated trades. “Thus, it really depends on whether the brokerage firms want to do it. They have to know who the investors are and if they know the risks they are taking. The KYC (know-your-customer) procedure is very important and their clients have to sign certain documents. These things have to be done carefully.”
Risks and returns
The risks and returns of algorithmic trading can be high, depending on the return target.
For example, Liu says one of Tixguru’s algorithmic trading strategies has been able to generate 3% return per month and is set to bring in 36% by the end of the year.
“You may ask, ‘Is this possible?’ Yes, it is, but it is also relative to the risk you take. The higher the risk, the higher the return. For instance, if you want to increase your return from 36% to over 100%, you can do it by using the same algorithmic trading strategy and leverage it by three times. But the risk is huge,” he says.
Yong Cheng Chook, founder and director of Straits Index Sdn Bhd and a professional trader, cautions investors to know what they are getting into before going for algorithmic trading.
According to him, real algorithmic trading should include automated trading and the use of the back-tested strategy.
“Based on my experience, it is true that simple algorithmic trading such as one that uses two exponential moving averages to trigger ‘buy’ and ‘sell’ signals can bring you huge returns once you hit the right timing. But what people should know is that in order to hit that right timing once, they will fail many times,” he says.
“For instance, let’s say you will fail 30 times before hitting that one right timing ... you will definitely lose money every time you fail. Ask yourself, ‘Are you able to fail 30 times and lose a sum of money for that one right timing?’ This is a simple explanation of the risk you are expecting to take.”
However, Yong says, this is based on his experience in Malaysia, where most of the traders and brokerage firms only use back-testing of their algorithm. He adds that in other countries where the traders are more sophisticated, they also conduct forward-testing of their algorithmic trading strategy.
“It essentially means that traders will try to randomise all factors to simulate ‘real market’ situations to test their strategies. It is more accurate than conducting only back-testing.”
Meanwhile, there are “trading gurus” who claim that they have made a lot of money by using their own algorithmic trading strategies and are opening classes to teach others their methods for a fee.
Yong’s advice is to remain cautious.
“Yes, it is possible that a few of them got huge returns using their strategies. In my opinion, it is merely luck that they managed to hit the right timing. But not everyone is that lucky,” he says.
Still a nascent industry
Yong Cheng Chook, founder and director of proprietary trading firm Straits Index Sdn Bhd, says algorithmic trading has been slow in picking up in Malaysia, compared with countries like China and Singapore.
Only a few investors apply it and it is mostly on the futures market where there is more liquidity.
At the same time, their equipment and methods are very basic, involving only back-testing of certain strategies that use various indicators. This is unlike sophisticated traders in other parts of the world who are able to conduct forward-testing or deep learning.
“Also, most of the local traders perform their trades manually instead of automatically [where the computer creates and submits the trades].
In comparison, Singapore is moving much faster as the Singapore Exchange had taken early steps to innovate its markets eight years ago. “Their traders are much more sophisticated than ours in this area,” says Yong.
Things are moving even faster in China. “It is typical of them. They copy and learn things very fast. And they implement them very quickly.”
Yong says there are reasons for the slow development of algorithm trading in Malaysia. He attributes it to the 1997 Asian financial crisis where then prime minister Tun Dr Mahathir Mohamad blamed George Soros for short selling the ringgit.
“Besides Soros, the short-selling mechanism, together with the futures market [which allows people to short sell], also carries the stigma of causing the meltdown of the country’s economy. Since then, local financial institutions and brokerage firms have been reluctant to introduce newer and more advanced trading methods.”
Algorithmic trading has been progressing slowly even though Bursa Malaysia has introduced the Direct Market Access, which allows traders to do algorithmic trading (including automated trading) on the local market.
Another reason is that most local brokerage firms do not take the initiative to push algorithmic trading. Also, there are not enough traders developing and using it.
“There are only very few local firms that have a platform to support algorithmic trading. Thus, even if traders have developed their own strategies, they are unable to apply them on the real market. Our firm is one of them,” says Yong.
“For the very few firms that provide such platforms, their clients are mostly from overseas. It is a chicken-and-egg situation.”
For now, he says, only a few fund houses, mainly big international players, have traded a bit of Malaysian equities through algorithmic trading (equipped with automated trading). This is because there is not enough liquidity here.
“There are more people deploying it on the [palm oil] futures market as it is much bigger and more liquid,” Yong says, adding that the local market will not be affected at the moment, but it will be hard to tell in the future if the trend will take off locally.
Although Yong remains open-minded on the changes to come, his friends who are also traders have told him about the downside.
“During our chat, they said they hoped that the Singaporean market could be like Malaysia’s where algorithmic trading is only minimal. They said once it kicks off, it will develop into a never-ending race for equipment and technology. If you don’t have enough money to buy good equipment and develop good technology, you will be left behind,” he says.
Part-time trader Jason Ching says he would want the market to remain status quo as algorithmic trading will disrupt the market and open up opportunities for other types of algorithm trading like high-frequency trading.
“It will be humans versus robots, like what is written in Michael Lewis’ Flash Boys. In the current situation, we do research and pick undervalued stocks. But when machines come in, they may take over the role of human traders.”
However, Li Guang Zhen, co-founder and marketing manager of Smart Trade Inc, says fundamental analysis and investment will not be replaced. “Just look at Warren Buffett. Hedge funds and algorithm traders can’t beat him [on returns].”