With Damian Mitchell, Managing Partner of FX trading boutique – Dsquare

First Published: e-Forex Magazine 30 / Tradertalk / January, 2008

With Damian Mitchell, Managing Partner of FX trading boutique – Dsquare

Damian, what prompted you to set up the specialised FX trading boutique Dsquare and can you tell us a little about your trading style and approach?

Firstly and foremost trading is a passion for me. I love the thrill of beating the markets and each trading day provides a different opportunity. Fortunately I have worked with some of the world’s best FX traders in my career. Having traded a few different trading styles in my career I realized when the market opened up to the buy side electronically that this was my time to turn my hundreds of different ideas into a piece of software. I have always loved the thrill of finding an edge in the market and I wanted to capitalize on that edge.

Secondly, I have always been fascinated with systems and started developing horse racing systems from the age of 12!

Finally, I always wanted my own firm where I could have the flexibility to create and implement strategies swiftly with the team of my choice.

Dsquare trades high frequency and in high volume. This volume gives us a lot of valuable data. We study the market for consistent patterns and use them to create new algorithms which we trade electronically via API’s. We trade thousands of times a day and are always looking to hunt value. Our knowledge of the microstructure of the market, and this includes our technological expertise, gives us our edge. We have many strategies and multiple micro strategies.

Has Dsquare always operated at the high frequency end of the spectrum since inception? If so, why was this? (E.g. more inefficiencies, easier diversification, smoother equity curve). If not, what was the prior trading style?

We have been high frequency since inception but were really quite surprised by how high frequency we actually were! Fortunately we were early adopters and this gave us a head start over much of the competition. Our success has been aided by the huge increase in FX business and intra day volatility over the last 18 months. Our initial strategies were very high frequency but as the market has become more efficient these strategies have become less fruitful and we have moved onto many other strategies. The fact that we have many strategies means we continue to trade large volumes (Dsquare has traded more than a trillion dollars worth of volume in 2007) but rather than split between a few strategies there are now well over 100 micro strategies. Whilst Dsquare trades in a high frequency manner many of our algorithms may only trade a few times a day.

Does Dsquare look to trade strategies that the competition doesn’t and do you feel that the ready availability of technology and data to all participants is reducing the shelf life of trading opportunities in your space? (I.e. inefficiencies that four years ago lasted six months now last one month.)

When we wrote the business plan we decided that we must be unique and able to beat any new algorithm that we create. We will not compete with any strategy that we know anyone else is trading. If someone else is trading in this way we know it is only a matter of time before it becomes obsolete. Our efforts are centralised on new strategies where we know we are alone. We strive to have more knowledge about the microstructure than our competition and create models where we have the edge. Technology and data are commodities in which everyone should have the same opportunities. We like to be quick but we don’t pursue speed as a competitive advantage as everyone will have the same capabilities soon. Knowledge of the microstructure is our strength and this knowledge enables us to build strategies which we hope are one step ahead of the competition. The ready availability of technology and data reduces the shelf life of trading opportunities in which the players all have the same strategy. It is so important that our strategies are different.

Dsquare’s Web site describes it as a highly specialised FX trading boutique. Are there further high frequency opportunities outside FX, such as synthetic inter market arbitrage, that may be of interest to you?

Dsquare is committed to FX and we still have a lot to achieve. This includes trading a wider variety of currencies, trading in different time zones, improving our algorithms, extending our time frames, trading on more marketplaces and implementing new strategies/micro strategies. We have also been approached by other algorithmic traders who want to use our trading and technological expertise to access the market. Whilst we love FX, we remain opportunistic and open to the idea of trading other markets which can be accessed electronically. The team spends a good amount of time studying other possibilities. The global growth in marketplaces continues to expand at great speed and if we can access these markets easily and they match our liquidity and volatility criteria then we will pursue them. In a nutshell, we may evolve from being a highly specialised FX trading boutique to a highly specialised trading boutique!

In your strategies are you more commonly a maker or taker of prices?

Overall we are more commonly a maker but this is totally dependent on the model. Market making also gives us some extra advantages. It often gives us a strong feel for where a market is trading rather than where it appears to be trading. If we support a market and we trade in a large size or there are a few quick trades in the same direction then we are well informed about where the market could be trading in the next few minutes. We also want to feel that we are making a contribution to the smooth running of the market and to act in its best spirits as this will benefit us and all the players in the long term. If we were purely taking prices then I also feel we would be jeopardising our reputation which is one of our assets.

Tell us about how you test your strategies?

We test the technology rigorously on our internal simulator but I am a firm believer of testing the profitability of the strategies in the real market. Live market conditions are the only true way of telling whether an algorithm works. We start trading in small size in normal trading conditions. After a few trades we analyse and check it is behaving efficiently. We will then increase the size in the same conditions. When we are fully comfortable with the outcome we will stress the strategy in various market conditions. Of course we monitor closely the returns of our strategies on a daily, weekly and monthly basis and have indicators that warn us of underperformance.

Do you rotate your micro strategies or are they typically all used? If you rotate them, is the process for deciding whether to use/rest them qualitative, quantitative or both?

We rotate the strategies depending on a number of factors including liquidity, volatility and opportunity. Only on two occasions have we used all the strategies together. It is rare that all the conditions exist for this to happen. Typically we are running between 40 and 70 micro models at any one time. Generally these are evenly split throughout the currency pairs that we trade but it is possible that we will focus on one currency if it is extremely busy. For example, after the release of certain UK figures we tend to run more strategies in GBP. Also, certain marketplaces have capacity issues which make it tough sometimes to support all the currencies we like at any one time. We use a qualitative and quantitative methodology for deciding which strategies to run but it is difficult to go into detail without giving out some secrets!

There was some bad press in August regarding model traders coping with the extreme volatility….how did you manage and what did you do differently?

It seems that the model traders that were caught out were trading equities and were running similar strategies to one another. Whilst FX also traded in volatile conditions I believe the conditions were far worse in other markets. Nevertheless at Dsquare we traded a lot more cautiously as these FX conditions were new to our software. Whilst I have traded through similar episodes in the last 20 years we remained cautious and applied our ‘IF IN DOUBT: OUT!’ mantra when considering which strategies to use. We had a strong month but asides from the reward we gained a huge amount of information about the robustness of our models and how they performed in these new conditions. Whilst the conditions were rare we are expecting more of the same in 08.

Do you feel we may be entering an era of ‘disposable models’ or do you find that most of your existing strategies remain viable if tweaked appropriately?

Some of the earliest models adopted by the banks 3 to 4 years ago are now non existent. We always want to be one step ahead and part of the process involved in creating a new strategy is to beat it. We are constantly looking at other strategies and for improvements. We put a shelf life on each model and if it continues to perform beyond that time period it’s a bonus. I am a strong believer that one strategy’s loss is another strategy’s gain. We even have proof of that: we had a low frequency strategy that used to run between 10 and 30 times a day. It seemed to be squeezed out but this gave us further information and enabled us to produce another algorithm that is now even more profitable than the original.

Does Dsquare have a strong preference for building its own tools? If so, in what areas does this apply? (E.g. database management, quantitative/statistical modelling, order routing.)

We like to build our own tools unless it makes business sense to purchase off the shelf. The trading software is a robust and easy to use framework in which we implement the new strategies and improve the existing ones. We re-use and share as many technical and functional blocks as possible across the different models in order to reduce our time to market. This way any improvement will benefit all strategies whilst the algorithm building process becomes more efficient. We also have a large number of indicators and statistics that are automatically used by the models to improve and increase their success ratio. Overall we like the tools to be simple to use and robust. 

You constantly supervise your strategies. Is this more a matter of risk management or refining strategies on the fly to enhance already positive performance?

Both. We like to keep an eye on our strategies as closely as we can. Whilst we are aware that our performance has been robust I have learnt over the years that anything can happen at any time and to counteract that we must be in a position to react quickly in an unlikely market event. We don’t want to be in a position where we have to rewrite some code in an extraordinary event where we need to react quickly. As you correctly say we sometimes refine the strategies on the fly. This is generally early in a model’s lifetime as the more we refine it the more data we have about the strategy and this data can then be written into the software so that the algorithms can react themselves to certain patterns. Risk management and security are very important at Dsquare. We have many security features written into the software which warn us of any potential risks.

With Damian Mitchell, Managing Partner of FX trading boutique – Dsquare

Do all the currencies trade in a similar manner?

All of the currencies have commonalities but they all trade differently because of the individuals trading them. I firmly believe that each currency has its own personality and behavioural patterns. Whilst the currencies appear to react in a similar style to the outside world I actually think from a micro perspective they trade very differently. Typically in extreme conditions they become even more individual. I love to study and read books/papers on trading and sports psychology and I use some of the principles of sports psychology in the strategies. There are some close correlations between trading professionals and sports professionals and they react in similar ways to certain events. This is especially the case under extreme conditions.

Is there any particular advance that has taken place since Dsquare started that has particularly benefited your trading?

I would say the new entrants to the space have added a lot more liquidity and this has definitely benefited us. The different interests of the entrants makes it such an interesting market to participate in and provides extra opportunity. We certainly welcome all new players!  Also, the prime banks have certainly improved their technology solutions in the last two years. These improvements now save us a lot of time especially at the end of the day when we are looking to reconcile. Time is most important to us. It gives us the freedom to create even more strategies.