AT: What do you say to scepticism about the viability of complex AI systems in financial markets?
Marco: The debate between simple and complex can be framed in many different ways. One way of looking at this is that simple systems tend to not have the characteristics of having [an] adaptive and intelligent nature that we believe is critical for predictive systems. Most systems involved in systematic trading tend to identify trends as opposed to predict price behaviour and are based on simple rules that have been pre-determined by the portfolio manager.
Those simple systems tend to work very well in trending markets. The problem is that if the markets are not trending, for example when they are choppy or bound within a tight trading range, the signals are not as clear, and the systems get caught out. This also happens when the markets reverse very quickly when a trend ends.
AT: What is the difference between trend following and your own adaptive systems?
Marco: What we have been working on and been developing and using are systems that are intelligent in the sense that they are able to change their own rules by themselves, as a function of the changes in the market. If, for example, they see a market that is choppy, they behave more like an oscillating system that is more reactive to short-term changes, not stuck in a single (long or short) position as if there was a strong trend. On the other hand, if they detect trends in the market our systems can recognise this and keep their positions longer, so basically our systems are designed to continuously learn and adapt to changing market conditions based on what they see; and the way they do this is by changing their own rules at each prediction point, so every day. Our systems are also short-term focused. They focus on predicting price behaviour over the next 24 hours and have an average holding period less than four days, whilst trend-following systems hold positions for several weeks or months. Our systems are also market-specific, unlike most trend-following systems.
AT: What is the mathematics/science underpinning your technologies?
Marco: It is very well established in industry outside of the world of finance so, for example, it's similar to the type of technology that is used in weather forecasting or by the online search engines. If you think of Google when you start to type your name in, Google is giving you a prediction of what you are likely to be searching for, taking into account what you searched for most recently and also what the online search community that best matches your profile has been searching. These types of industrial applications use adaptive machine learning technologies similar to the ones we have developed for financial markets.
[The technologies] all have in common this feature of being able to self-generate the predictive rules at each prediction point, as opposed to being based on fixed or pre-determined rules, which is what the simple systems do.
Some of these technologies work better than others in financial markets. So - just one example - neural networks were a real fad several years ago. A lot of people were talking about neural networks as very promising, and they proved to be a bit of a disaster.
Essentially, if you want to know whether you need to go long/short and by how much, [neural networks provide] a very detailed map of the past that can be very accurate in predicting the future, if exactly the same conditions persist in the future. But if you get something that deviates from the past, which is the norm in financial markets, the systems can get confused and what they predict often amounts to wild guesses, with little or no value - could be right, could be wrong - but their predictive accuracy is not reliable.
Our systems use machine learning technologies that are more reliable as universal approximators. Rather than memorising the past, they attempt to model it in more general forms, capturing only its most essential characteristics, and in this sense they act much more like a compass or a smart sat-nav system in predicting likely future behaviour over a broad area, than a very detailed paper-based, static street-map focused on a narrow strip of territory. For this reason they tend to be much more reliable than neural networks at inferring likely future price behaviour based on unseen data, especially in changing market conditions.
AT: Your systems predict short-term movements. Will you be continuing in this direction?
Marco: Yes. Financial markets are not only very complex but also non-stationary, meaning the conditions that describe the market today change over time quite quickly; and so if I am making a prediction which is two weeks out, the base upon which those predictions are made will have changed significantly by the time I get to the two weeks. So generally, the shorter the forecast horizon, the higher the accuracy. It is a little bit like weather forecasting, where predictions from one day to the next tend to be more accurate than predictions from one month to the next.
Our systems work on a maximum forecast horizon of 24 hours, also on an hourly basis and half hourly basis intraday.
AT: Can you tell me about your performance?
Marco: We have generated positive returns since inception in June 2009, strongly outperforming all the various indices like the Newedge CTA Index and the Newedge Short Term Traders Index, and we are right at the top in terms of risk adjusted performance among all CTAs. If you then look at which have been the best performing markets for us - because we have these market specific systems so we can look at them in isolation - I would say the commodities markets have been very strong. Our commodities strategies have comfortably outperformed all commodities programs we are able to track globally. As commodities are traditionally among the most difficult markets to trade, we are particularly pleased about that.
AT: When you look at what is coming next in this field, is there anything that gets you particularly excited?
Marco: I think that there are always opportunities to improve the computational efficiency of your systems - how quickly they can produce their prediction -based on technical improvements on the existing infrastructure of your systems. There are things, for example, like rewriting part of your programmes in a software language that is more efficient so that it can be processed more quickly by the machines, so that you can get an answer more quickly, even without making any additional changes in software functionality.
And in terms of new functionality, this self-learning, adaptive characteristic of the systems, which allows them to self-generate the most appropriate rules, input variables and system parameters at each prediction point, can be refined and expanded more and more.
Anything that moves in this direction, which is effectively what our whole development programme has been about from the start, we think is pretty exciting. What we are trying to do is limit as much as possible the human subjective bias that is associated with constraining the systems to operate within a specific, narrow set of pre-defined rule choices. We have already achieved this to a significant degree, but we are continuously working on extending this objective.
AT: What do you think about academic research in this area, or about what other firms are doing?
Marco: We don't know that many players that are doing what we are doing. We just don't. If someone points them out we will be grateful. We think there is work going on in artificial intelligence/machine learning, but most of that is happening in a few proprietary trading firms, which don't say much at all. There is also some interesting work in academia but most of it is theoretical with limited practical application in terms of trading strategies. In terms of mainstream industry, many of the underlying predictive technologies we are using have already been very successfully applied. You will find examples in weather forecasting, in security applications like credit card fraud detection, in areas like critical illness medical diagnosis, in speech and object recognition, in online information retrieval systems, such as those used by Oracle and Google. But there are significant barriers to entry to be able to get it right in financial markets. It is difficult to translate some of these technologies to the realities of financial markets. For one thing, markets are very noisy. They are not only complex but also very noisy. So, for example, the data-processing components that seek to remove noise are very important.
These are some of the reasons why actually we set up our company in the first place.
AT: Over a year ago, you said that Titian was "100% transparent on our positions because we believe it is impossible to reverse engineer a software that is made of 500,000 of code in C++". Is that where you still are?
Marco: All correct, except we have more lines of code, like 600,000 now. Also, the reverse-engineering point is reinforced by the fact that the rules, input variables and parameters of our systems change every day.
What is interesting is that if you ask the same question to a trend follower, they are typically very reticent of giving end-of-day positions of the individual markets they trade, because there is this possibility of reverse engineering the rules which are often quite simple. In the end the proof is in the pudding. The fact that we are showing that we don't care about you seeing the individual position is testament that we mean what we say. We can show it to you and investors.
AT: What kind of computer programming language are you using?
Marco: We use C++ and other developmental languages. Some languages have different efficiencies, but you can basically translate them to a generic C++ equivalent.