These engines house numerous investment teams, working collaboratively within the framework of Man Group, with a high degree of investment autonomy. Each team benefits from the strength and resources of the firm’s single operating platform, enabling their primary focus of seeking to generate alpha for clients. The teams invest across a diverse range of strategies and asset classes with highly specialised approaches, with long only and alternative strategies run on a discretionary and quantitative basis in single and multi-manager formats. Man Group’s investment teams are empowered and supported by our institutional infrastructure and technology, which aims to facilitate the most efficient exposure to markets and effective collaboration across the organisation.
We asked Damien Lempriere, Principal Trader FX at the firm, to tell us more about his job and use of FX execution algorithms.
Damien, please tell us a little about your career and how you became involved in the world of currency trading?
I had been introduced to the world of finance from an early age. I can remember seeing pictures of open outcry on trading floors around the world, and this had a huge influence on me. Shortly after leaving school, I started my first job as an FX Trader for an investment bank, surrounded by experienced traders on a busy trading floor. Their enthusiasm for their jobs sparked my own interest in currency trading – a career I’ve pursued since. I joined Man AHL almost twenty years ago before becoming responsible for FX Execution in 2008, and recently I took on a new role as Principal Trader FX at Man Group, focused on all aspects of FX Execution for the group.
What do your day to day responsibilities usually involve?
The role of the execution trader at Man Group has changed dramatically since I joined, from a time when the trading day would have been taken up by trading on phones with little or no electronic market access, to today where voice trades would count for only a few per day. Trading and managing execution risk is still a huge part of my day-to-day responsibility but now, with the electronic capabilities, you have time to think and consider improvements to the finer details of trading styles and execution.
I am also much more involved in the research area of our business and work with portfolio managers closely, whether this is on liquidity in the market or the best way of implementing a strategy.
I also find that keeping banks informed of hit rates and volumes via our different avenues of execution, and making sure they are aware of our current technology developments, are an important part of my working day.
The trading desk retains full ownership of trades and how they are being processed in the market place.
What do you particularly like about the currency markets and what are you looking to leverage from them?
Most FX currencies are very liquid instruments to trade and, although correlations between individual pairs can be higher than some other asset classes, the opportunity to be able to hold significant positions appeals to our portfolio managers.
We have numerous signals for each strategy that will calculate different criteria before making an investment decision. We are naturally looking to optimise the alpha that can be generated from the market – we believe this can be done by strategies that take views over short, medium and long term from a broad range of signals. Volatility and therefore opportunity in currency markets can come from numerous areas be it through central bank decisions, local and geopolitical situations or even herd mentality.
What instruments do you usually focus your FX trading and investment activities around?
As an example, Man AHL, Man Group’s diversified quantitative investment engine, has approximately 70 currency pairs set up to trade within the system from the most liquid EUR/USD to non-deliverable forwards and now going beyond into frontier markets. Man AHL actively trades over 50 currencies virtually during every minute of every trading day.
Due to the way certain currencies trade, some of Man AHL’s strategies lack streaming electronic prices. These can often take longer to price, trade and require voice trading relationships, particularly in comparison to our short-term trading models, which are fully electronic in execution.
We have invested heavily in technology in recent years, so that we can continue to be a market leader
How much intervention do you and your team make on orders as they are being processed and executed or is almost everything done entirely electronically?
The vast majority of orders are now executed electronically. The trading desk retains full ownership of trades and how they are being processed in the market place. The desk will intervene should execution risk parameters be breached in our fully automated trading venues, at which point the trader can assess liquidity or price movement before making a decision on execution. Fund releases and portfolio reallocation decisions will always have a trader present to advise stakeholders on the potential execution strategy to achieve best execution for our clients and investors.
Does the Trading Team have much direct engagement with clients and investors?
Yes, the trading team often engages with both clients and investors. Man Group has specialised traders in every asset class, so we believe we can answer any questions a client may have with regards to the micro structure of the markets we trade and provide insight about how orders are being processed in the marketplace in real-time.
We have a rigorous approach to when it comes to reviewing bank algorithms which I believe will have a huge part to play in the FX market over coming years
How do you go about turning an idea into an FX trading strategy? What does the process involve?
We have a structured process for implementing new signals and models into our systematic portfolios. This involves: a researcher coming up with an initial idea or concept, which is then assigned to independent reviewers of the project, and together they undertake an analysis and research with the in-sample dataset. Once the research is completed, a committee meets (including the CRO and CIO), and they will assess the out-of-sample analysis. If the model is approved, we will then utilise proprietary capital first to test the strategy fully and check the implementation. If all the tests have proved successful, we will then release the signal/model into client portfolio trading.
Why does designing and employing quantitative trading strategies necessitate state-of-the art technology?
The ability to process large datasets – be that huge amounts of tick data or non-price based information – and, more recently, our ability to adopt sophisticated algorithms with the strategies through machine learning, clustering analysis, and reacting quickly to events in the market, all require state of the art technology and equipment. We have invested heavily in technology in recent years, so that we can continue to be a market leader and create trading models that our clients will invest in.
Do you encourage competition between the various FX trading pathways you use and how do you measure the quality of execution to ensure you are always trading optimally?
Any trade that falls within a certain order size criteria is split between what we call a ‘Matched Band’. This ultimately means that our fully automated algorithm is benchmarked against the desk trader. We then utilise a machine learning algorithm which chooses the percentage split within this matched band, based on historic slippage performance, and is recalibrated weekly. Our benchmark is taken from arrival price in all cases.
We have a further execution analysis, which involves looking at a number of different trading criteria, be that a mixture of market impact, reject rates, skew for all our liquidity providers within our internal algorithm, and our human EMS.
Let’s talk about FX trading algorithms. When did the firm start deploying them and what factors may influence whether your team deploy AHL-developed algorithms or best-of-breed third-party solutions?
Any trade that is executed away from the human traders is traded by our in-house FX algorithm, which has been designed and tested over many years. Man AHL first started trading fully electronically in early 2005 through a very simple RFQ model. This has developed into a fully streaming pricing model across any value date in any size bands we require, where we constantly look at the heartbeat of a currency pair and trade when the algorithm believes it is the opportune time.
I think we are all waiting for that one algorithm that is a cut above the rest
Human traders have used third-party bank algorithms for a few years now and have been impressed with some of the execution experiences. We are in the midst of evaluating whether to integrate a more automated utilisation of the route to market and are embarking on this exciting project with our chosen liquidity providers.
We have a rigorous approach to when it comes to reviewing bank algorithms, which involves fully testing the performance of their algorithms, whilst seeking to remove any potential human bias. With the volumes we transact over time, we will build up a strong dataset that will help us identify which banks are the best-of-breed in the algorithm space.
How do you go about achieving Best execution for clients and what role do algorithms play in that?
Algorithms play a big part in the selection processes for all the asset classes where we have a non-human trading element and human execution. We benchmark by using a machine learning technique, which incorporates slippage data over different time frames and allocates larger percentages to either human or fully-automated trading, based on historic slippage performance.
Internal, bank and EMS designed algorithms touch almost every order we execute in FX. As we all know, best execution does not come purely from the cheapest price – there can be many different variables, all of which combine to make best execution what it truly is.
In what ways can combining Machine Learning with FX trading algorithms help with achieving both superior investment returns and a more efficient execution process?
Machine learning will continue to play a huge part in new upgrades to the current bank algorithm suite. I think that the ability for machine learning to actively manage order placement with Primary and Secondary ECN venues, internal liquidity pools and process huge amounts of market data quickly may be a huge benefit to algorithmic trading in the coming years. More broadly, machine learning has been incorporated across many parts of our business now, whether this is in portfolio construction, strategy construction, physical trade execution or pre and post-trade cost analysis.
How do you see the future of bank algorithms within the FX market and how do you see your use of them changing over time.
I believe bank algorithms have a huge part to play in the FX market over coming years. Banks have and continue to invest significant sums of money into this space and the technology race is still in its relatively early stages – in my view, the FX market will probably start to look a little more like the equity market over the coming years.
I think we are all waiting for that one algorithm that is a cut above the rest and maybe, as machine learning continues to be brought in we are not too far away from that. You can already identify some have some very smart features; others need some developments. Banks welcome feedback on individual execution experience and I continue to work with the sell-side to improve features within algorithmic execution. I do believe the sell side will shortly be able to offer SOR functionality for the buy-side to gain DMA access into certain NDF currency pairs. I would also encourage EMS systems to continue to expand into the algorithmic space.
*As at 31 March 2018. All investment management and advisory services are offered through the investment “engines” of Man AHL, Man Numeric, Man GLG, Man FRM and Man Global Private Markets (GPM).