Buford, please can you tell us a little about your background and some of the trading and investment management activities you have been engaged in during the course of your career.
I started my career as a commodity trader, and evolved to become a very active market-maker in London Metal Exchange metals. Market-making involves taking proprietary positions (at least it used to!) and I had an affinity with computers, so I started creating and back-testing models to gain an edge in my proprietary trading, incorporating lessons learned along the way. As time went on, more of my revenue was being generated by model trading than dealing, and I became less of a dealer and more of a prop trader. My asset universe grew to encompass all futures and FX markets, and eventually I started managing client investments alongside my own. That evolved into a distinct asset management business focused on model-driven trading, with a team of quant traders and structurers, and over $2bio in assets in seven different strategies, working within a bank infrastructure. Last year, I left the banking world, and along with some former team members, set up Stelrox. We have focused on FX from the start, where we see the greatest opportunity over the coming 3-5 years.
What prompted C-View Ltd and Stelrox Capital Management to get together and form a strategic alliance and what core attributes do each side contribute to it?
Starting an investment management business in the current regulatory environment is a daunting task. Building a platform from scratch requires months of arduous and expensive work to establish regulatory registration, prime brokerage and dealing relationships, operational infrastructure etc. To trade even a small account requires a large and robust infrastructure, and, once built, infrastructure capacity can be much larger than what’s actually used. We recognised this from the start, and rather than build from new, we looked to collaborate with a company that had the necessary infrastructure already in place, and that was interested to monetise their excess operational capacity. C-View was a great fit for us, and vice versa, as they have a top class infrastructure built specifically for FX. We have collaborated on many levels, not just for infrastructure, as Stelrox has brought some systematic expertise that can also be applied to areas other than trading. We also collaborate with marketing, where we have found that clients are often interested to invest in both discretionary and systematic strategies, rather than one or the other.
You are Managing Partner at Stelrox. Who are the other key people involved in the firm and what are their main day to day responsibilities?
Stelrox has a total of four partners: myself, Charlie Genge, Richard Garlick, and Suzy Scott. Charlie is our main platform developer, and between the two of us, we handle systems development and strategy trading. Richard is our CFO, and is very qualified, having been head of finance and product control for the trading division at Lloyds Bank- Stelrox is lucky to be his ‘retirement project’! Suzy is a former bank salesperson, and focuses on marketing for us. All of us tend to be involved in all aspects of the business, as there is plenty of work to go around! We depend on C-View’s staff for trade processing and operations: a case-in-point showing how it has made sense for us to collaborate.
What sub-strategies go to make up the C-View Stelrox Systematic Currency Strategy and what are they each designed to do?
As I mentioned earlier, we have chosen to focus on the FX sector. We have a portfolio of three FX sub-strategies, all of which have evolved over years of trading, and have low correlation to each other. Our objective is to focus on relative value as the main driver of returns, which is managed with an acute focus on macro-economic risk. Two of our three sub-strategies focus on relative value trading, split between G10 and emerging markets respectively. We diversify this approach further with our third sub-strategy, FX trend-following, which can not only add value to the portfolio independently, but also act as a hedge when relative value strategies are underperforming. Each of the three sub-strategies are independently successful, but are most effective when combined into a portfolio, which gives us another layer of diversification, and a robust returns profile.
Why do think the launch of the Stelrox strategy came at a good time for investors and who are you offering it to?
Our strategy performs best in an investment environment with slow to medium growth, with clear macro-economic divergence, and in particular, with rising interest rates in some parts of the world and flat or falling rates in others. This appears to be the environment that’s currently developing. A rising rate environment is generally detrimental to bonds and a headwind for equities, but a boon for currency strategies like ours. We think we are extremely well placed to help investors diversify their investments at this point in the investment cycle.
The overall investment environment in 2015 was a complex one. Did the Stelrox strategy perform well and meet your expectations during the course of the year?
At our launch in Q4 2014, the first US interest rate rise was generally expected within 4-5 months, which would have been the start of macro policy divergence. Fast forward a year, and the divergence still hasn’t happened. So no, 2015 did not meet expectations- arguably we started up a few months too early. But that time has given us time to bed down all of our systems in a relatively quiet environment. We are now very optimistic for 2016, especially given recent developments.
What primarily drives the Stelrox investment process?
I would say it’s primarily a simple, logical, sensible trading process- scaled up through the use of computers. We try to systemise our experience, by programming in the profitable approaches we’ve learned, and programming out known weaknesses. That means having effective risk management and good diversification, not just robust trading models.
When broken down into individual components, our process is remarkably straightforward and simple- it just becomes more complex when we scale it up to a portfolio level containing hundreds of positions across dozens of currency pairs.
What do you see as the benefits of investing in a systematic currency strategy?
The beauty of FX is that currencies can be paired in hundreds of different ways to produce hundreds of different return streams, which enables a very high degree of diversification. We monitor dozens of cross rates, and for each currency pair, we will have 10-15 data streams that we monitor: things like credit spreads, volatility surfaces, equity volatilities, etc. So computers give us the ability to scale our observations across a large and diversified portfolio, though each individual component might be remarkably simple. At any given point, we may hold positions in 60 or 70 currency pairs. An individual trader simply cannot monitor positions in that size of portfolio, whereas computers can monitor hundreds of inputs, to calculate risks and identify opportunities on a constant, real-time basis.
The executive summary introducing your currency strategy talks about exploitable FX price phenomena. What is that and why is it important?
The main phenomenon we are referring to is the FX forward discount anomaly, also known as the forward rate bias anomaly. It’s the main driver of FX carry profits. There have been scores of academic research papers on the topic- all debate why the phenomenon exists, but all agree it yields outsized profit probability. Norges Bank wrote a great discussion note last year* that summarised around 70 studies on the topic. As mentioned earlier, we use this as a major input our in our relative value models, though we also have a multitude of other data streams that feature as well. We think 2016 shows exceptional promise for strategies that exploit this phenomenon, as outsized returns often coincide with the early stages of rate divergence.
What frameworks have you developed to enable you to leverage your research in the most effective way in order to fine tune your existing trading models, develop new investment ideas and implement more robust risk management methodologies?
We have some basic guidelines when approaching trading systems development, which have evolved over the years, but retain a lot of consistency. First, we try to keep things simple- I have a strong view that simple is usually better. With trading systems, this means we look to develop systems with low numbers of parameters, and running the same parameters across multiple markets. This helps us to avoid over fitting the data set, and we use several other techniques to limit this effect as well. We also try to develop systems with an eye on our operational capabilities, and scalability. For example, we won’t work on a system that our infrastructure can’t support, or that has a small capacity. We try to achieve scalable, diversified models that have consistent returns with low volatility, then combine these in a portfolio that further diversifies risk.
What are the main currencies you usually trade?
We trade only deliverable currencies, no NDFs, but in that set of the market, we are able to create literally hundreds of currency pairs. We then focus on uncorrelated subsets to achieve diversification, which is important for a systematic approach. G10 gets a lot of our attention, but there are quite a few deliverable emergings too. Why no NDF’s? A couple of reasons- we avoid the basis/benchmark risk that can occasionally be an issue; also, we avoid some of the most volatile currencies. There is still plenty of juice in the deliverable emerging market currencies, and much better liquidity.
How did you go about building your trading desk infrastructure and what issues have influenced the trading technology you use?
We get pretty excited about our infrastructure, which is entirely cloud based. One of many advantages is that we can spin up a super computer in the cloud for a research project, run it for a few hours, and then shut it down- and we pay for it by the hour. This gives us the ability to match or exceed the computing power we had in our former bank’s grid super computer, for a fraction of the cost, with more reliability. We’ve built our platform using open source computer resources that not only keep costs down, but also enable us to leverage our analytical capabilities well beyond what we were able to do using bank proscribed systems.
For operational infrastructure, which includes electronic execution platforms, clearing, trade processing, risk oversight, etc., we use C-View’s excellent infrastructure, which is designed particularly for FX trading. We use their electronic price aggregation systems for low-latency execution, and benefit from their wide range of liquidity providers, which has been built up over their 18 years in business. Being able to plug into the C-View infrastructure as a start-up saved us considerable time, administrative hassle, and expense, and also makes us more attractive to prospective investors who want to see robust operational infrastructure. From that standpoint, we have considerable advantages over other start-ups.
Once you have developed ideas for a new strategy how do you model and test it?
Different models may require different testing techniques, but a common approach is to back-test an idea over years of data, and across multiple currency pairs, to see if the model is robust in different environments. With back testing, there is always a danger of ‘curve fitting’ (maximising profits with the benefit of hindsight), so we have some interesting methods of ‘de-optimising’ models which result in more robust models. As mentioned earlier, we are able to use quite a lot of computing power to crunch through millions of scenarios.
Is combating latency particularly important for your trade execution performance and, if so, what steps have you taken to improve your connectivity pathways and IT architectures to reduce it?
I think anyone who trades with poor latency in today’s world is giving away ticks (at best!) and with the current state of technology, low latency architecture is within reach of most players, either directly or indirectly through intermediaries.
In our case we use intermediaries that have servers based in the primary data centres, like LD4, which solve a lot of the latency vulnerabilities for us. At the end of the day, we are low frequency, long term traders, so our current solution is cost effective for us.
Do you think the value proposition of investing in a currency manager like Stelrox has strengthened over the past few years and if so why?
I do- in fact I think it has strengthened considerably over the past few weeks. I think FX is a sector that will benefit greatly from divergence in macro-economic policy, and in a rising US interest rate environment it may possibly be the best sector to invest in. We are emerging from an extended period of homogenous macro-economic policies, and low rates globally, that have produced disappointing results for the FX sector, and investors have been well rewarded for concentrating on equities and bonds. I think that period is behind us now, and we are beginning a major new phase in the asset cycle, in which FX will present extremely attractive prospects. We have chosen to focus on FX at Stelrox because of the opportunity we see over the 3-5 year horizon, as the phase develops.
Looking ahead, where will you and your team be focusing your attention to explore new currency investment opportunities for Stelrox?
One of our day-to-day pursuits is research. We look at different time horizons, different models, different portfolios, etc- there are limitless combinations. We try to focus our attention on areas where we can achieve scale. Our models are built with a degree of self-evolution programmed in, so we implement changes to existing models only when we discover something that makes a significant improvement, but we keep looking.
We also look at new approaches, to look for ways to diversify our portfolio further. At the moment we are researching shorter term systems, which, if our efforts are successful, may become a fourth sub-strategy at some point in the future.