Sunday, December 13, 2009

On the Problems Of Automated Trading System Design... Why a Broader Vision is Necessary

I have always been interested in the design of automated trading systems and the creation of long term profitable trading strategies. When I was beginning my journey in this area it became apparent that I was missing a key component to achieve success in designing and progamming strategies that could be profitable in the long term. During my whole learning process I have been able to uncover most of the things that were hindering my approaches to profitable algorithmic trading and I can now say that this has led to the achievement of several of the likely profitable trading systems I have programmed up until now such as the Watukushay expert advisors, the GBP/JPY daily breakout system, the god's gift ATR (partially programmed by me) and other similar expert advisors. Today's post will be dedicated to the explanation of one of the most important truths I have realized through my journey in automated trading, the need for a holistic approach which involves a lot of creativity and observation in order to overcome the short comings one faces when going from theory to practice.

When I first started to program trading systems I quickly realized that the theoretical usage of many indicators and the actual implementation of them in a trading strategy is a completely different matter. Take the stochastic oscillator as an example. There are several ways to trade this indicators, crossovers of the %K and %D indicator lines, crossovers from above 80 or 20 to below 80 or above 20 which indicate sell and buy signals respectively and divergence criteria which are very discretionary and not at all straightforward. If you try to implement any of the first two criteria into a trading system you will find several problems. There is no guidance as when or where to exit trades (except on opposite signals) and there is no indication of how you should set your stops or what lot size to trade. It is just like having one piece of a 1000 piece jigsaw puzzle and trying to figure out the whole picture, it is very hard to get there.

Now what many people try at first (including me) is to setup a system with the entry signals and a fixed TP, SL and lot size. Try this and you will find that with ANY system you will NEVER achieve long term profitable results. This is because you are not looking at the big picture and you are just taking into account the most irrelevant aspect of trades, the entry. Knowing when to exit trades and how to trade is much more important and this is often a fact neglected by people new to automated system design.

If you introduce an adjustable SL, TP and lot size that changes against market conditions you still have a big problem. You are neglecting the characteristics of the specific instrument you are trading. You will find out that a stochastic oscillator based system with the above entry criteria will be very unprofitable even with an adjustable SL, TP and lot size, mainly because you are not evaluating if your entry criteria for the indicator fits the instrument you are trading. Forex currency pairs are very different from each other. Different trading pairs have different characteristics and a deep and careful analysis of indicator behavior against each pair is absolutely necessary to arrive at a long term profitable system. Also there is the fact that most indicators do not work in the market with the criteria they are "supposed to", I have found out that it is almost always the case that, after a deep analysis, the long term profitable results are generated by a "counter-intuitive" or complimentary use of many different aspect of an indicator or group of indicators that make mathematical sense together. You will find out when Watukushay FE is released (on Dec the 24th) that it uses a counter-intutive approach to the RSI indicator to trade the market profitably.

My recommendation is then pretty simple. If you are truly interested in the development of profitable automated trading systems you will need to have a deep understanding of the mathematical meaning of each indicator you want to use and how its movements can lead to the location of inefficiencies on different currency pairs. A system that may work very well on the GBP/USD may not work at all on the USD/CHF because of this very reason. If you would like to have more information on how I design, trade and profit from long term profitable trading systems please consider buying my ebook on automated trading or subscribing to my weekly newsletter to receive updates and check the live and demo accounts I am running with several expert advisors. I hope you enjoyed the article !

No comments:

Daniel Fernandez's Expert Author Email Alerts
Sign up to receive email alerts of Daniel Fernandez’s latest articles from!

Email Address: