The Gateway to Algorithmic and Automated Trading

Quantitative Approach with a Human Touch

Published in Automated Trader Magazine Issue 04 January 2007

As electronic trading continues to gain ground, one of the casualties has been the reduction of personal contact. While things like data, quantitative calculations, speed to market, etc, have become essential parts of this new trading frontier, the demand for the exchange of ideas and better understanding of the market place has never been greater. What are the tools available? How do they work? How can you get the best out of them?

Brian Schwieger

Brian Schwieger, Head of EMEA Algorithmic Execution and Trade Analytics, Merrill Lynch

From EMS through to post trade analytics and Execution Consulting, users need an individually tailored solution that meets their needs. Understanding this has driven us to constantly evolve our algorithmic strategies, develop new trade analytic tools, expand into Execution Consulting and develop strategic partnerships with Portware and ITG Block Alert.

t's important to remember that tailored solutions include the way in which client information is conveyed and handled. Anonymity has always been a much discussed topic in this industry, but it is generating more interest as brokers increase crossing capabilities to maximize opportunities within their own "dark pools". Anonymity is an important part of any tailored package, and that relies on well defined and executed infrastructure security. This is something which we take very seriously in designing our platforms and handling electronic flow.

Approach to Algorithmic Trading

From the very beginning of the design process, our approach to Algorithmic Trading is to combine detailed quantitative research with human trading skills. We have experienced quantitative researchers throughout the world studying the microstructure of individual stocks. Their job is to find ways of incorporating new data and intelligence into the behaviour of the algorithms. At the same time, we have an extensive consultative process with our own traders, discussing trading tactics and strategies which are then incorporated back into the algorithms. A great example of this is our current work on enhancing the behaviour of strategies used for trading European mid and small cap stocks.

We're leveraging our acknowledged expertise and trading capabilities in this space by forming a development team that includes both traders and quantitative engineers. While tailoring algorithms based on our own trader input is an important part of our process, we believe that it is as equally important to listen to our client users, many of whom are now very experienced in electronic trading. Incorporating their comments, even developing customized algorithms for their specific needs, is a key element of our approach.

Process of evolution

The development of good algorithms is a process of evolution. The nature and structure of markets is constantly changing, and it's important for a provider of algorithms and electronic trading tools to understand this. Today's "next generation" is tomorrow's dinosaur. Early innovation is not enough providers need to offer products and services that evolve with the markets and clients' needs.

Evolution can be seen in practice through the behaviour of the Order Book Manager - the part of the algorithm that handles and trades slices on the order book. It's at this stage that the importance of data like trade frequency, average trade sizes, spreads, interval volatilities, queuing times and trend analysis can be seen. If the Order Book Manager is too aggressive, you will end up paying the spread unnecessarily. On the other hand, if behaviour is too passive, you risk the market moving away from you resulting in higher execution costs. The role of the Order Book Manager is often overlooked, but it clearly plays an important role in delivering the best possible execution price.

Another excellent example of evolution and the combination of quantitative research and human trading skill is in the performance of our OPL-X algorithmic strategy. It is an aggressive, statistically driven strategy that can be used as an alternative to a market order - in effect, "smart DMA". OPL-X removes liquidity from the market using both passive and aggressive slices simultaneously to quickly and efficiently execute the order. As such, it can take longer to execute than a straight market order, but its aim is to reduce short term impact and post-execution reversion. The success of OPL-X is demonstrated by our clients' acclaim and the strategy's performance. A recent study by our analysts found that over a three month period, OPL-X provided our users with an average improvement of over 150% versus a straight market order. That's not only a significant saving, but it underlines the trust that can be placed in an algorithm to execute intelligently while the user focuses on other issues.

Tailoring is key

Although technology and quantitative research are the cornerstones of our offering, we understand that it's the human touch that is crucial for our users. Tailoring the combination of quantitative research and human expertise is the key to providing a successful and profitable electronic trading experience for our clients. This is done by providing an end-toend service, where the sales team oversees the relationship with the users In terms of day to day interaction, the Execution Desk monitors client flow and offers on the fly advice for traders, acting as the human interface into the algorithmic process. Lastly, our Execution Consultants provide not just Transaction Cost Analysis, but a review of performance with the aim of helping clients tune their execution style to match their investment style. Tailoring the combination of quantitative research and human expertise is the key to providing a successful and profitable electronic trading experience
for our clients.

For further information please contact:
Brian Schwieger or Yvonne Hansmann
Direct Line + 44 207 996 5117