The Gateway to Algorithmic and Automated Trading

QuantCore: Little fund in Big China

Published in Automated Trader Magazine Issue 36 Spring 2015

Martin Lockstrom is a Swedish national operating a quant fund out of Shanghai - a choice he made with his family after falling in love with the country from his days as an academic in the region. Automated Trader talks to him about the opportunities and technical realities of starting a little fund in big China.

Martin Lockstrom, Chairman and CIO, QuantCore Capital Management

Automated Trader: Choosing mainland China wasn't a strategic business decision for you exactly, but the region is certainly interesting to investors. What brought you there?

Martin Lockstrom: I don't have a traditional financial career background. I finished my PhD in 2006 in Germany in the operations research area, heavily influenced by mathematics and physics. I got an offer to continue research at a local university, then afterwards I spent another three years at a business school called China Europe International Business School, which is one of the top business schools in Asia.

In 2012, with a second daughter on the way, my wife and I thought it would be nicer to be closer to home in Sweden. We headed back to Germany where I lived before coming to China and stayed there for about two years, but after realising it was easier to move to China than from China, we moved back in September 2013. Some people say we are crazy, but I usually reply that China is like French cheese: in the beginning you think it stinks and is disgusting, but after acquiring a taste for it, you realise it's a delicacy and you just want to have more.

Our fund often gets this question, is there a strategic investment reason for being in China? Let me put it this way, most companies are well-established but in my case it is the other way around. We are in the very early stage of companies existing here.

A lot of change is taking place. It is a big market, big growth, a lot of capital seeking investment opportunities. So all of these are nice side effects of the fact that I am pretty much already here. It is very interesting to tap into the local market here.

AT: What are the major challenges of operating out of Shanghai?

ML: Our servers are not in China. One thing to mention when you are talking about China, you cannot avoid talking about IP (intellectual property) protection.

Everyone wants to get rich yesterday and employer loyalty is not really the best in the world. There is a high attrition rate, in some professions up to 30% per year.

IP is pretty much the only asset we have. If we can't keep our IP proprietary, what is the edge?

In our case, we have chosen to keep our core IP outside of China. This also has an impact on the kinds of decisions you make in terms of hardware and software localisation. If you keep it outside, then you have a problem with higher latency - instead of 10 milliseconds latency maybe you have 30 milliseconds latency from Hong Kong, and higher than 300 millisecons from Europe or the US.

Other things are, you have to be more wary, and that means less outsourcing, better data security, better encryption, more vetting of your recruits. It adds complexity across the board. Especially in our area, technology and IP is where we get our edge. That is one of the key challenges here.

Our biggest challenge is simply the fact that we are small and not a well-known brand. When we tell about our performance many people think we are too good to be true, but we are happy to share our transaction statements and show line-by-line how we accomplish this. People should be skeptical for good reasons, but usually what we see is our clients are investing smaller amounts to test the waters, and then when the see returns, they want to invest more.

AT: Where do you fit within the larger quant community in China?

ML: We don't operate any domestic fund for the moment, however we plan to set something up later on this year. We have an algorithm for CIS300 index futures that we are testing right now doing mock trading, and we plan to set up our own structure in the shape of an investment management firm later on this year.

We work with asset managers and futures brokers to manage capital here in China, but that will be more a local-for-local product because we discovered that even though our returns for our flagship fund Quantum Leap II have been very good, there is still a preference for local products in the market. This strategy is based on the same principles as the other strategies that we already have.

Source: QuantCore Capital Management

We aim at serving as an investment manager for sub-funds of existing funds which are offered in the market here. We don't plan on setting up our own full blown asset management operations here at this stage. We are doing it slowly but steadily. There is a lot of capital here, both in the retail and institutional investment community.

When it comes to investing, the Chinese have basically two options historically - real estate market or go to Macau for gambling. But this is starting to change. Property markets are coming down a bit so people are getting more wary about that sector. People are becoming more interested in alternative investment products, especially quantitative hedge funds and. In general, the appetite for risk is much higher here than elsewhere, based on my own experience.

In other parts of the world, investors expect three to five years of solid track record. In China, six months and you can start attracting capital.

AT: Who are some of the market players?

You have some of the bigger foreign players here, like Och-Ziff, Winton, BlackRock, Goldman Sachs. And then you have the local asset managers and brokers, like Galaxy Securities, Shenying and Wanguo Securities, Haitong Securites etc. These are the bigger firms. Most of the big banks offer some kind of wealth management products geared towards domestic clients.

There is a huge number of small individual traders and prop trading firms. I have seen numbers like 25,000 registered financial services firms, whether they are wealth managers, or prop traders, or individual traders, or whatever. But like many other industries in China, it is very fragmented.

By fragmented, I mean you have a very large number of very small market players.

AT: Do you see it continuing to develop that way?

ML: The success of China is three-fold; modernisation and liberalisation of the market is one thing. But China GDP growth has been about access to technology and also upgrading people so that they are able to draw on that technology.

These are the levers behind China's success. The financial industry is no different. It is a mission for them to become the most successful and prominent financial firms in the world.

So, on the one hand, you have a huge number of retail investors, who are like the 70-year old grandmother going to the securities firm to invest pension savings in the stock exchange. Then you have the more big established players.

In terms of talent you have a lot of Chinese overseas graduates returning, bringing the knowledge that they have acquired from top business schools, engineering schools, with the ambition of joining one of the bigger established firms or setting up their own shop.

Even though the hedge fund industry is in its infancy here, in China it is very rapid development taking place but like every other industry the fragmentation will persist for the foreseeable future.

AT: What are some of the opportunities developing from the point of view of algo trading?

ML: Our criteria is liquidity, and usually liquidity in the index futures. So in China, the CSI 300, which is the combination of the 300 largest stocks on the Shanghai and Shenzhen stock exchanges. It is the main index, and shows that Shenzhen and Shanghai are equally important, although different in nature

The Shanghai-Hong Kong Stock Connect doesn't have a very big impact on us, because it is not liquid stocks. According to the latest information, you currently cannot short sell through it although this will be possible later on this year, however we are primarily trading futures hence the Stock Connect has relatively little relevance for us.

There are four key futures exchanges - the Shanghai Futures Exchange, China Financial Futures Exchange, Dalian Futures Exchange, and Zhengzhou Futures Exchange, each of which focuses on different underlying commodities. The latter two most foreigners have never even heard of.

AT: China is just opening up commodity futures trading to foreign firms, crude oil in Shanghai for example, how do you see that developing?

ML: You have the base metals, precious metals, lean hogs, agricultural futures, things like rebar and plywood, not just metals but refined products that go into construction, which can be standardised in one way or the other.

Futures trading has been introduced, options still have not launched, this has been postponed repeatedly. The government regulatory bodies are still in discussion on what the legislation should look like. Maybe later on this year we will see futures options being launched in China.

AT: Tell me about your fund QuantCore, you've only been running about a year?

ML: We went live in April 2012 trading our personal assets. We opened up for external investors in the beginning of 2013. Basically we started out as a private project and eventually it went so well that I decided to take it to a more commercial level.

We are still very small, we have now been going from embryo stage to the stage where we learn how to walk. We are about $4 million in AUM. However, if you look at the growth rate in terms of number of clients, we currently have about 45 clients, most of them are retail investors, but with an increasing number of institutional investors like family offices, fund of hedge funds, etc.

We are still a fairly small team because all our trading is fully automated. We work pretty much project based, so develop new algorithms, testing new algorithms. We have three staff in Europe, US and Asia pretty much monitoring the systems, making sure the operations run smoothly. Then it is me, my business partner and admin local in China. We have grown over the year in terms of marketing and PR, we are currently working with roughly a dozen or so agents. At the beginning of last year we were three people, and now we are roughly 15 in number.

We are operating as a registered investment advisor, so we are managing segregated accounts, and if we extrapolate our AUM growth, by the end of this year we intend to have $15 to $20 million in AUM. And for having catered to mainly retail investors, we are now focusing mainly on high net worth individuals, fund of hedge funds, family offices, and so forth.

AT: What do returns look like?

ML: Performance since inception dating back to April 2012, we have returned 334% gross so far. For the full year 2014 gross return was 119%. So annualised return ranges between 80% and 90%. We have a Sharpe ratio of approximately 2.0 for our flagship fund, Quantum Leap II.

AT: How do you expect that to change as you grow in size and AUM?

ML: We expect the performance to remain very stable, similar to what we have seen in the past, because all our trading systems are defined equally well, regardless as to whether markets go up or down as we utilise short selling selectively.

Usually what we see is, when volatility is very low, as measured by the VIX index, there tends to be fewer trading opportunities because there is less momentum in the markets during those periods. Our performance simply stays flat, because our algorithms sense that there is not enough momentum in the market and so it chooses to stay on the sidelines. In our forthcoming system, we have enhanced it so that it can trade profitably during market regimes with low volatility as well.

If we are seeing the sharp drop of our algorithms, they short positions so we can generate returns even in periods of decline. We usually say that we look forward to the next stock market crash and that is part of our investment philosophy - there's no greater selling point than helping your investors sleep well at night and worry less about the future. We want to benefit from decline rather than becoming a victim of it, in line with the concept of "antifragility", if you've read Nassim Taleb's latest book with the same name.

AT: What kind of technology underpins your trading systems?

ML: When we talk about algorithmic trading, the discussion often leads to high frequency trading, which is very technology intense, capital intense, you need to have the latest, greatest hardware, and fastest programming language, colocation, all this sort of things.

We are still a very small company, with limited resources, so we are not even trying to compete with the big HFT guys. The way we see it, you can either trade faster or smarter. We have opted for the latter, as we believe that HFT will become increasingly regulated in the future, but also as it tends to become a sort of rat race where it's only a matter of time before someone gets ahead of you. By utilising highly proprietary systems, chances are lower that someone would come up with something exactly identical by chance. In fact, by minimising the amount of inspiration we get from others, we minimise this risk.

This also means that means we don't need colocation, FPGAs or any of these fancy concepts. We do develop software in C++, but that is where we have our expertise not necessarily because we have to be faster.

When we talk about latency, of course we can't have latency or lag of five minutes, but if it takes 10 milliseconds or 20 milliseconds for us to submit an order that has very little impact as of today.

We use a combination of Matlab and C++ for software development. We don't use retail hardware of course, we use real servers. We have redundant power supply, redundancy in the switches, which is hosted by our IT service provider in order to ensure that we have stable and reliable operations. But in terms of speed, there is nothing extraordinary about our hardware.

AT: What can you tell me about your trading strategy?

ML: Our flagship fund is only trading S&P E-minis. We do fully automated trading, there is no human decision-making involved in any trading buy or sell decision. Even the optimisation is automated. We rarely if ever do any changes to any parameters in the software. The software reconfigures itself. But there is no manual redirection in what we do. Our software only has an on switch, there is no off switch. We can interrupt our software and execute orders manually of course, but it is really designed to run perpetually in any kind of market condition for an unlimited period of time.

Quantum Leap II, if I had to categorise it, is predominantly a trend following, and more proper term would be order anticipating, because it also takes contrarian positions.

If we have followed a trend for a certain period of time, and we see signs of weakness it might take a contrarian position. If it is long, it might take a short position.

I would say it is medium frequency trading. Typically we do maybe one to three relative trades per day, usually not more.

Furthermore, we are aiming at substantial returns, so we don't hedge, but instead rely on our predictive capabilities. And this is also the reason why we have the magnitude of returns that we have. But we cannot tolerate the overnight risk so we close out before every trading session.

We like futures, and particularly index futures, because they usually have very high trading volumes. Liquidity is very good. We like simplicity because our trading is automated, so we have to ensure that our algorithm is stable and reliable - the fewer parts, the lower the complexity of the product, the lower the risk of something going wrong, pretty much like in the aerospace industry.

We prefer to have very streamlined, very simple algorithms, so typically we only trade one instrument per algorithm. We started out with the S&P E-mini futures simply because liquidity is good, volume is good, and you also get natural diversification by only trading this instrument - the whole spectrum of S&P 500 stocks.

AT: If your software is reconfiguring itself, is that because you use machine learning?

ML: We have tried using artificial intelligence and machine learning, but results in our case were quite varied. We saw that it works really well, but for a limited period of time. In the long term, what you see in financial time series, the data are so non-linear, so high degree of randomness, such a high degree of efficiency, that we were only able to produce good results for a limited period of time, then it started to deteriorate very quickly.

Our algorithm is based on mainly price information and the derivatives thereof, like volatility. We filter price information and have a couple of other indicators as well. We only use standard exchange data feeds, we don't apply big data or pull a lot of unstructured data from the internet. It is all highly standardised data feeds that we are filtering in order to extract the information out of the noise, giving us predictive power.

AT: Where do you get your data feeds from?

ML: We use several brokers, but for Quantum Leap II, we mainly use Interactive Brokers.

We filter the datafeed information to reduce the level of noise and subsequently generate our own internal indicators. At the beginning of each trading session, we are gauging the direction of the market. If according to our indicator, we predict an uptrend, we take a long position. If we detect a downtrend, we take a short position. If there is a lack of trend, we simply stay on the sidelines. It sounds very easy in theory, but the algorithms have grown quite complex over time in order to cope with all nuances of the market.

Depending on how the trend evolves, we continue riding the trend or we might take a contrarian position. And then we have daily cumulative stop losses as well. What our algorithm does at the end of each trading session, I mentioned we close out our positions, the system also tries to predict based on its historical performance whether or not tomorrow will be a good day to trade.

That helps us to avoid a lot of the drawdowns that you see among CTAs and other quantitative hedge funds. We have built in an analytical layer which helps us to predict on a day to day basis whether the market conditions are favourable enough to be in the market. If not, then we stay on the sidelines.

AT: What are your future plans? Any new markets or asset classes?

ML: Our next step is Quantum Leap III. We have developed algorithms for the Nikkei 225 and EuroStoxx 50 indexes. We are only trading 6.5 hours per day in the US. By trading Japanese and European markets, we will be able to utilise time more effectively and trade virtually around the clock. We trade three markets in sequence on a daily basis, and by doing that our aim is to at least double our current returns while at least keeping risk-adjusted returns constant.

AT: Do you think you can maintain that as the fund grows?

ML: If you introduce a lot of complexity as you grow bigger, you run the risk of slippage. But by diversifying into other instruments, you can easily expand capacity. And by going into different markets using the same capital, instead of trading one market, you trade three markets with the same capital on a daily basis, this kind of ramp up has the potential to substantially increase returns without taking on excessive risk.

If we can do 100% in one market, why not three hundred percent in three markets? Returns add up linearly, whereas the risk is regressive. We aim at growing returns faster than we grow the risk.

Of course as our AUM grows we will have to pay attention to the complexity of simply getting bigger. Our algorithm is already developed to handle bigger AUM and to consider the existing liquidity in the market, so that we reduce the risk of becoming a victim of lack of liquidity and slippage.

300% is our vision. We may or may not achieve that, but realistically speaking definitely about 200%.

AT: Are you considering any of the other BRICS for trading?

ML: I don't have a lot of experience in the other BRICS countries, but what I can say is that when we were considering which markets we should be active in, we considered the Nifty Index in India. The reason why we opted out was simply because of the time zone overlaps. We want to have as little overlap over regular trading hours as possible due to the setup of our trading

But probably the most mature market of the BRICS in addition to China is India. They have an English-speaking work force, a very strong tradition in mathematics and also a lot of talent there. They have a strong track record in IT, which makes them very strong in the quant hedge fund area.