Fisycs Capital – taking emotion out of the trading process

e-Forex talks with Alexandre Vigier and Arnaud Amsellem founding partners of Fisycs Capital, a systematic investment management firm based in Paris.

First Published: e-Forex Magazine 42 / Tradertalk / January, 2011

Alexandre Vigier and Arnaud Amsellem
Alexandre Vigier and Arnaud Amsellem

When was Fisycs formed and what type of investment objectives was it designed to undertake?

AV: Fisycs Capital is an independent entity established in November 2009 and solely owned by its founding partners. The company received its authorization from the AMF (Autorité des Marchés Financiers) in France in October 2010 and is now a fully-fledged asset manager.

As a money manager our goal is to provide investors with above average return with as low as possible volatility. This is achieved through not focusing on any asset class in particular, we prefer instead to investigate different strategies and keep only what provide good enough return to risk ratio. Investing only liquid asset classes guarantees that our clients can meet their financial constraints at any time. Finally our systematic approach rationalises the investment process and remove any emotion from the trading decision. Both of us experimented this approach for many years as Quantitative Portfolio Managers for different institutions and it proved to be efficient. We decided to set up our own shop to push this logic even further and building the entire company around those concepts of Quantification, Diversification and Liquidity.

Who are the key people involved in the firm and what are their main day to day responsibilities?  

AA: Alexandre and myself as founding partners and owners of the company are making the final decision on all major issues. As a systematic asset manager we are heavy users of IT and data and this takes a good portion of our time. We are defining the research agenda together and distribute the tasks. In addition we do most marketing presentations together. However, Alexandre having an extensive knowledge and background in trading technologies is the best person to decide on trading and IT infrastructure. He actually spread his time equally between IT issues,research which entails coding up ideas in prototyping languages, interpreting the results and implementing the models in live environments, and trading.

I spread my time between the operational tasks and the research. Even if we try to externalise as much as possible anything that is not directly business related, we still have a fair share of operational issues to deal with. I take care of most of them. In addition, I do the research as well. In my case it means a lot of coding especially back test calibration and result reviews.

You have more than 12 years experience researching, developing, implementing and managing quantitative equity strategies. What do you like about working with FX as an asset class?

AV: The ample liquidity of the FX market makes it a playing field of choice for anyone who is interested in quantitative investing. Coming from that kind of background it is just a natural extension of what I was doing previously. However there are several differences that make the FX market even more attractive. Firstly FX offers diversification relative to the strategies (or asset classes) we already offer. Secondly there is a large offer of ready to use (i.e. clean) data. Gathering usable results is quick as opposed to equity where one needs to invest heavily in data and databases to extract any usable information from the market. Finally the FX market is very simple in the way it works and the trading technology offer is more advanced than for equity for example. 


AA: This makes the full R&D cycle much shorter and efficient: the feed back from the market can be gathered almost right from the start. A typical example is MT4. There is a myriad of brokers offering the ability to trade via MT4. Even if it is not suited for professional money managers, opening an account (or even demo account) and monitor portfolios is made possible with little programming and with very limited capital.

Fisycs focuses exclusively on ultra liquid markets. Was that decision influenced by your investors?

AA: Fisycs Capital has been set up as a multi assets shop and we intend to keep it that way for the foreseeable future. We strongly believe in diversification and investing in all liquid assets is a major step in that direction. Our decision to enter the FX market was motivated by two different factors. Firstly, Alexandre and I both have been Quantitative Portfolio Managers for years we therefore have a strong bias toward systematic investing. As a result, the FX market offering such a deep liquidity is a must have for us. Secondly, investors acted as a catalyser for our FX strategies.  

AV: We started to research FX Intraday strategies initially because they were offering maximum diversification compared to the programs we already traded. But as we started to talk to investors about it we felt a strong interest and spot FX grew quickly from a research project to one of the major asset class for our company. Today we are dedicating roughly 50% of our time to this asset class.

How would you describe your investment philosophy?

AV: Our investment philosophy is articulated around three major ideas:

Systematic

•  Our investment approach is entirely quantitative

•  Research and Development are the corners stone of our investment process

•  We are constantly looking for improvement in our strategies

Diversified

•  Asset Classes - We invest in all liquid markets regardless of the asset class

•  Regions - We cover North America, Japan and Europe

•  Investment horizon - Positions are kept between a few minutes and a few months

•  Models - Many systems come into play in the final portfolios

•  Positions size - Many small size positions in the portfolios at any time

Liquid

•  We focus exclusively on high liquidity markets to guarantee maximum transparency

•  Our investors can meet their financial constraints at any time

What range of strategies does Fisycs offer?

AA: We offer both intraday and lower frequencies (i.e. daily) strategies. We do not focus on any particular type of strategy, we prefer instead to identify an idea and test it extensively. Ideas are generally coming from academic articles or market stylized facts that we are trying to translate into investable processes. If initial back tests prove to be encouraging we are moving the strategy to paper portfolio then live trading.

As of today we are trading strategies with holding periods ranging from a few minutes to a few months with no particular bias toward persistent or anti persistent strategies or any particular cross. The number and nature of crosses traded is tailored to clients constraints and requirements. We always start with very liquid crosses and add less liquid ones to gain diversification.

We also started to investigate latency dependant strategies recently but we are not ready yet to open it to investors.

Fisycs Capital – taking emotion out of the trading process

How diversified do you consider your investment activities to be?

AV: We are paying a particular attention to diversification as it is a key concept of our investment philosophy. As a company Fisycs Capital invest not only in the Forex market but in futures and cash equity as well. We perceive diversification as a multidimensional component of our investment process. We identified five major axis in that component.

•  Asset Classes - We invest in all liquid markets regardless of the asset class

•  Regions - We cover North America, Japan and Europe

•  Investment horizon - Positions are kept between a few minutes and a few months

•  Models - Many systems come into play in the final portfolios

•  Positions size - Many small size positions are preferred to a concentrated portfolio

Regarding the FX market in particular the three major axis of diversification are: crosses, trading frequency and models. Our goal as quantitative portfolio managers is not only to come up with the best possible models but to find the best possible balance between those three axes as well.

Your investment approach is entirely quantitative. What do you consider to be the key strengths and weaknesses of adhering to that?

AA: Different people will come up with different answers but for us there are three major advantages to approach investing in a systematic way..

Metric of investment performance: By setting proper hypothesis, one produces back tested results that should be very close to live trading. We identify strength and weaknesses of strategies much quicker.

It removes any emotion from the trading process. The machine does the job regardless of the state of mind of the trader.

Scalability: most if not all quantitative approaches, being based on statistics, requires a lot of data which is best suited to liquid markets. It translates into investment strategies that are by construction highly scalable. As of today, we estimate we can manage up to $1 billion on our FX strategies without having to change anything.

Replicability: once a strategy is identified for a given instrument, it usually does not take a lot of effort to apply it to others. It potentially increases the level of diversification and the capacity with little marginal work.  

AV: There is however one weakness: The black box perception: having spent years explaining quantitative investment to investors, we sometimes struggled to get the message across. It takes a lot of effort to translate a sophisticated investment process relying on advanced techniques borrowed from different scientific fields into plain English.

In what ways do you leverage your research agenda to help improve  the design of new investment strategies and the ongoing enhancement of your existing investment processes?

AV: Our research agenda is driven by the needs as defined by the already existing material. We define two general themes. First we constantly reassess the adequacy of the behaviour of our existing strategies, the ones already invested. Sometimes, mostly often, we do not need to come back on them, no red light having being fired. Then we define our research targets to be as different as possible from what we already have, in order to guarantee that we will ultimately come up with a non correlated source of alpha. As far as we experienced this, a good common sense at the very beginning of the research process is the key. We do not try to data mine huge amounts of data (the best way to find spurious relationships) we prefer instead to follow financially sound ideas.

What trading platforms do you use and what factors influenced your choice?

AA: Being a fully systematic house we needed a robust framework capable of handling a large amount of data. We were actually looking for the following features: Very good brokers connectivity, extensive testing capability, flexible programming language, possible link to back/middle office and, to a lesser extent, low latency.

Fisycs Capital – taking emotion out of the trading process

After reviewing what was available, we decided to use several products : On the one hand, SmartQuant and QuantFactory from QuantHouse. For the later (institutional version of the software), we developed FIX connectivity from scratch  with the major players in the FX space. On the other hand, Dukascopy, which is a high performance platform. In addition it provides a Java based programming language (JForex).

How did you go about building your trading IT infrastructure and what steps did you take to improve the operational management of the business?

AV: As a start-up we need flexibility but as a Quantitative Asset Manager we are constantly crunching numbers and we therefore need a lot of computer power. Cloud computing offers the best of both worlds: Computers can be bought and dropped at will with the added benefice of not having to worry about the maintenance. We distinguish three types of machines. The ones we have on our desks are mainly used for communication, monitoring and light coding. When it comes to testing and heavy programming we have several external servers that are also used to run simulations. In case of very heavy calculation we rent computing time.

Finally there is a third type of machines used solely for production purposes. Live trading activities are hosted on those dedicated servers with very low failure rate. This was actually an important criteria when we decided on external providers.

What back-testing methodologies do you employ to confirm that a strategy is relevant to your long term trading goals and performance criteria?

AA: Our experience thought us there is never enough testing when it comes to systematic investing. Unfortunately time is scarce so one need to come up with a research methodology that is not only rigorous but time efficient as well. Years of experimentation allowed us to come up with a three steps R&D cycle that has been extensively tested and proved to be efficient and robust.

Prototyping

Starting from the same dataset we implement the strategy in two different programming languages usually R and C++. It avoids leaving any major bug in the code. Until we reach the exact same result with both implementations (i.e. identical trade signals) we keep on carrying tests.

Paper Portfolio

Once we are happy with back test results we set up paper portfolios and track the system behaviour until we get fully confident with what we observe. 

Live trading

Once we are comfortable with the paper portfolio we move to live trading which entitles recoding the strategy in our trading platform in C# or Java depending on the client choice of platform. Usually starting with very small trade size that we increase gradually over time. The feed back from the market is an important source of changes and improvements.

Fisycs Capital – taking emotion out of the trading process

What key risk management frameworks have you put in place and how do you adjust your investment horizons?

AV: Risk management is embedded in the trading itself. Every strategy is designed with strict maximum drawdown constraints and we attach to every single position a stop loss and a take profit. Putting multiple strategies at play in a portfolio produces well bounded volatility that we are constantly monitoring. Deviation from a normal range is subject to close scrutiny.

AA: As part of our maximum transparency policy, we are also offering to clients the ability to push what we call the “panic button”. In case of extreme event, the client can interrupt trading at anytime should he requires it. Obviously there has to be a very good reason to do so. Finally for strategies running on high frequencies, we do not keep any overnight position. 

For all intraday strategies we do not keep any position opened over the week end where the liquidity is scarce. There is no explicit reference to horizon in our investment strategies. However, as a rule of thumb, the holding period goes from a few minutes to a few months depending on strategies.

Looking ahead, where will you be looking for new investment opportunities and what new strategies are you considering employing?

AV: We are constantly researching new investment strategies with one underlying concept in mind: diversification. We only start investigating a strategy if it potentially offers low correlation to existing systems.

In the FX market one major axis of diversification is trading frequency (i.e. holding period). We currently trade all but latency dependant frequencies and obviously this is high up on the research agenda. We started to put the necessary pieces together and when time comes we will start presenting it to investors.

We are also investigating other asset classes as long as they offer the level of liquidity compatible with the transparency we want to offer to our clients.