Portfolio Risks: Taking Into Account Instrument and System Correlation

In what is commonly known as modern portfolio theory, the goal of the asset manager is to diversify investments in a way in which losses in one asset will be offset by profits in another while – in the long term – all the different investments will yield at least an overall positive net return balance. The idea sounds very nice and it is in fact known as the only “holy grail” in finance but there are some very important hidden dangers behind portfolio diversification which only become apparent when things get truly bad. Within this post I will talk about instrument and system correlation as it relates to algorithmic Forex trading and how portfolios must be developed to take into account possible events that may change the way in which they interact with one another.

When you want to be as certain as possible about your future survival when using variable return investments diversification becomes very important since you want to make sure that you’re not exposed to any particular type of market occurrence. If for example there is a war in the middle east, a breakup of the European Union or a hurricane in the Caribbean you want to make sure that your investments have the highest chance to survive this type of event (or even end it with a profit). In order to do this, asset managers usually build portfolios which have a varied composition in which many different classes of investments are brought together to minimize risk.

In Forex trading we apply a similar technique to diversify our investments in a way in which we can effectively minimize risk as much as possible. In algorithmic trading this is achieved by using an arsenal of mechanical trading strategies which trade different instruments or use different strategies to tackle different aspects of market behavior. However the problem with diversification in general is the assumptions of correlations which may or may not sustain themselves in the long term.

One of the biggest problems of modern portfolio theory is the fact that the assumptions of correlation tends to fail whenever there is some systemic risk in the market. This means that instruments tend to correlate themselves whenever something “bad” happens so the time when you need your diversification protection the most is the time when your diversification protection fails you. This is what happened in 2008 when the first world economic crisis caused a massive correlation of instruments which broke the previously established “order of things”.

Certainly the big advantage we have in this regard in Forex algorithmic trading portfolio development is that we do not develop portfolios  based on instrument correlation but based on the correlation of the profit and draw down periods of mechanical strategies. In this way we ensure that our strategies are as uncorrelated as possible so that if any changes happen within the correlation of instruments our strategies will most likely offset each other and reduce the damage this produces. Furthermore, massive instrument correlations in Forex trading are usually tremendously directional, giving some of the most profitable trading periods when using trend following systems (especially long term trend followers).

However we are still exposed to an event which causes massive system correlation, something which can cause very large draw downs on algorithmic trading portfolios. This can be totally independent of instrument correlation as it simply relates with the way in which strategies interact with one another. In order to prevent massive system correlation from happening in a destructive manner it becomes important to ensure that systems do not share very important sets of logic components and to make sure risk is adequately distributed between the different trading strategies (risk is equalized amongst all systems).

Another important way to diminish the probability of system correlation is to ensure that systems aim for different levels of profitability and possibly trade on different time frames. By using systems that trade on one time frame with systems that trade on another and further varying this with systems that aim for very different market movement lengths we get an even further degree of diversification as the possible circumstances which would cause all strategies to go into draw down become even more exotic and – in the worst case – we would slowly lose money as a consequence of spread costs (since all systems would have lost their edges and outcomes would be random).

Another great thing to do is to gear systems such that massive instrument correlations breed a lot of profitability, protection ourselves from potential market risks caused by systemic risk events. For example building portfolios which favor directional trading tend to favor this type of event and therefore protect accounts against massive instrument correlations. When the market “goes bad” this type of correlation always happens since money tends to flow in a very directional way when people are in panic. Building a portfolio based on strategies that fail when systemic instrument correlation appears is a bad idea since these movements are generally very aggressive and they usually take such systems into extreme levels of loss while directional portfolios are tremendously favored.

In the end portfolio building in algorithmic trading is all about putting together strategies which share common goals but differ in specific aspects, it is about putting together systems that can work great when the “starts align” but that do not fall apart as a group when conditions are not the best for all of them. Overall algorithmic Forex trading can bring a much more effective diversification than what regular asset diversification can do, ensuring that massive correlations bring profit whenever bad events unfold. If you would like to learn more about my work in automated trading and how you too can build portfolios based on system correlations please consider joining Asirikuy.com, a website filled with educational videos, trading systems, development and a sound, honest and transparent approach towards automated trading in general . I hope you enjoyed this article ! :o)

Print Friendly, PDF & Email
You can leave a response, or trackback from your own site.

One Response to “Portfolio Risks: Taking Into Account Instrument and System Correlation”

  1. erick says:

    Portfolio construction is the next level up from “long-term profitable systems”.

    Instead of combining LTPS into optimum portfolios, could you use a genetic approach to first create a set of designer LTPS that would best fit an optimum portfolio?

    And then use a genetic algo to create the optimum portfolio?

Leave a Reply

Subscribe to RSS Feed Follow me on Twitter!
Show Buttons
Hide Buttons