I have been solely focusing on foreign exchange for the last four years after initiating my career in the financial business in 1999. The last two years specifically have been dedicated to the development of models for the Saxo Bank Strategy Team along with Johan Ditz Lemche, where we as a team continue to strive towards increased long-term profitability; focusing both on the fundamental and the technical aspect.
My education was based on classical macro-economics. This laid the foundations for a healthy interest in the FX Markets, my intrigue and passion for the FX Industry is to be attributed to Saxo Bank’s historic focus on FX. Although this has now changed and we now dedicate ourselves to all asset classes, I felt that this was the asset class where I could add the most value - and have since grown to love it.
Though I have dabbled in Equities Models, I find FX much more linear, and “consequent”. When compared to Equities, FX Models have no exchange risk, lower level of fluctuations on entry points, and it is of course so much easier to back test. This may not have been the case 10 years ago, when the FX market was almost a closed game for inter-bank dealing, but thanks to the advent of online platforms, (like ours), we can easily create intra-day models for our clients that needn’t counter for the draw backs of heavy commissions and sporadic liquidity.
You are currently responsible for the FX strategies at Saxo Bank. How would you describe the strategies you design? E.g. trend following, pattern based, reversal, scalping, artificial intelligence etc? Or do you use a blend of strategy types for diversification?
I would not classify my strategies as belonging to one specific type or class. The FX market moves in different cycles. We consequently use volatility in all our models to help us understand the state of the market. Depending on the latter we apply the set of “tools” that we deem is best suited for the current conditions. E.g. we use breakout, reversals and trend trading, on a short to medium term horizon to increase our hit ratio; for short term volatility squeezes followed by breakouts we have recently created intra-day signals which I have been able to back test over the past 2 to 3 years. Finally we determine the quality and strength of our signals against the overall market conditions by incorporating everything into our fundamental model.
When developing strategies, how (approximately) would you normally expect to allocate your time among building entry signals, exit signals and money management strategies?
I started off by trying to create as many possible signals as I could, spending approximately 90% of my time developing new strategies, but this has changed dramatically over the past 2 years. I have grown to believe that building entry signals should be the least of your worries, it’s the money management that matters if you are going to be profitable in the long run. I estimate that 80-90% of the time is taken up by the money management part, with particular relevance to the size of my positions versus the profitability of the model. I actually think the biggest misconception when designing models is best summarized by:… “If I just have a good entry level, I will make money”, .. this is certainly not my experience.
Do you favour any particular data time frames in your strategies, or do you diversify across a range of them?
Yes, we have four models that we update everyday on New York close and then I update our intra-day signals on a continuous basis, which I run from early European trading to the first 2 hours of the NY trading session. I think it is important that if you choose to trade on intra-day signals you do it when both liquidity and volatility is high, otherwise the money management techniques used simply do not add up in the long run.
What historical data do you use in developing your strategies – data from Saxo’s own platform or from another source?
For our daily models we link to the Bloomberg Feed, which we find to be very reliable on a day-to-day basis. When developing intra-day strategies in FX it is a different story as it is not easy to find a reliable tick by tick feed. But we have been so fortunate to be granted access to Saxo Bank’s own intra-day feed. I find it to be faithful enough to sustain proper back-testing. Given that a significant number of Saxo Bank’s clients may trade your strategies, how do you model the impact of their trading activity on strategy performance?
We have encountered no major problems affecting our strategy performance. We do have a lot of clients trading our strategies, but we only trade 10 to 15 of the major currency pairs, where liquidity is high. It is of course something that we monitor closely as we move along and more clients decide to join our recommendations.
What tools do you use in the strategy development process? A generic trading/strategy development platform, or do you use specialist statistical/quantitative tools as well?
When looking at either developing new trading strategies or trying to optimize our money management techniques and position sizes, we primarily use statistical probability theory and mean reversion in true ranges as trading tools. These combos have yielded for us the most reliable trading techniques and of course the possibilities are endless.
How frequently do your strategies trade and does this vary widely from strategy to strategy?
The frequency varies a lot from model to model. As a comparative example consider that our reversal model for can remain on hold for 3-5 months before giving a signal whereas our intra-day break out model could give anything from 2-5 signals on any given day. But it will vary greatly from one trading day to another as our volatility filter is the back bone of all our models. We have had days with absolutely no trades and days (November 2006 when EURUSD finally broke through the much talked about 1.2980 level), which sparked volatility across the board. I remember we closed one day in that period with 8 positions, (where the importance of position sizing really came through). Without revealing any proprietary secrets, do you currently see any particular area(s) of research as being potentially the most promising for new FX trading strategies?
I think that the increase in liquidity and the lack of volatility is here to stay. I am not suggesting that we will continue to see these historical low vols, but the FX Game is changing again, and FX traders should prepare themselves for that. A good example is the current EURUSD range. If we look back 500 days, it has gone from an average of 91 pips to 67 pips a day; similar shift can be seen for all the other majors. So the longer term trend/volatility breakouts are becoming less profitable in the majors, whereas emerging markets still show sings of profitability trading in these sorts of patterns. To cater for this we have developed, and will expand upon some very interesting short term models based on certain time periods of the day, which I think will become promising strategies looking ahead.
From your interaction with Saxo clients through the chat function, do you feel that the average sophistication of retail traders as regards trading strategies (and especially automated strategies) has increased significantly in recent years?
From our chat function in the Saxo Trader 2 clients do seem to hold on to trading only on specific signals, but good signals are just not going to cut it. The information available to retail traders has definitely increased significantly in terms of both quality and quantity. Most online platforms provide some sort of automated strategies and most traders are also able to test the validity of said or prop strategies themselves (without having as a pre-requisite a degree in computer science). I see a trend for more and more “retailers” to start focusing on signal trading. I am still worried that too much emphasis will be put on the signals and too little on the money management.
Do you find that it is easier to build successful trading strategies for less actively traded pairs? Or do relative lack of liquidity and wider dealing spreads outweigh any potential gains from being able to capture more inefficiencies?
It all depends on your time frame and your activity. Obviously the shorter your time frame the more difficult it will be considering spread and liquidity. When building strategies for such currency pairs, you need to look at the bigger picture and longer time frames. Once again, emerging market currencies prove to be the prototypical example. I do not think you can or should choose. Liquidity and spreads should be treated as two variables which require integration in any successful trading model.
Have you been surprised by the speed with which electronic FX trading has evolved over the last few years and where do you think it’s had the most impact on your own trading activities?
Not really, FX trading is by far the most accessible asset class for retail investors. It is open 24 hours a day, has extremely low commissions, and its leveraging abilities make it attractive and appealing to our gambling instinct. I am conscious that most investors do not like "that word"; but in my mind, any sort of trading is gambling. The only thing that changes is the name of the game, whether the game is open or a closed to all participants, and the "cards" being used. Trading and strategizing is all about maximizing your odds. On my personal trading the greatest impact has been the lowering of spreads in the majors. It helps and nurtures the creative side of modelling strategies as it is gives me more options to exploit the market and increases my odds of making money.