The detection of important price levels - so called support and resistance - is one of the most difficult things to do in algorithmic trading since their detection by the human eye seems to be extremely discretional, something that just pops out at an experienced trader when he or she looks at the chart. Through the past few years several attempts have been done to detect important support and resistance levels but most of them fail due to the fact that intermediate levels -which are not important - are often detected and used by computers when their real relevance is minimal. The question then arises : Can we algorithmically detect support and resistance levels in a reliable way ? Moreover, can we actually make a computer "know" the importance of each level ? Can we then develop a profitable system based on this detection ? On today's post I want to give my first set of answers to these questions showing you my first attempt at the computational detection of S&R levels and the achievements this technique has had up until now when used as part of a mechanical trading system.
On previous posts I had talked about how we could approach S&R detection in forex trading by using the fractal indicator (not the default but one that doesn't repaint) and performing a historical evaluation of the accumulation of fractals in certain zones, assigning a particular importance to each of these levels. However this approach seems to be a little bit complex so I decided to implement a much simpler approach in order to first evaluate the concept behind this way of "detecting" S&R levels. What I did was simply to use the High and Low levels of past candles counting their presence amongst different zones and assigning a "percentage value" to each price level pertaining to the "population" of historical candle high/low levels around this area. Areas that are heavily populated by hourly highs and lows and prone to be important daily support and resistance areas owing to the fact that price tends to "hug them" failing to move "straight through them".
The significance of these levels is then easily evaluated by the percentage population of each price level and we can easily measure if price is moving towards a zone with a high population or outside a zone with a high population. When price is trading within a highly populated area it will have a tendency to remain there while when price trades in a "population scarse zone" it will tend to go to a zone where more trading has happened in the past. As it is shown on the image above, we could create a system by entering a position after we "breakout" of a zone that has a high population value. It is good here to note that we don't use levels but zones using a given amount of space to gather a given high/low population, this is done due to the fact that in forex trading support levels tend to be "spread" over a given area instead of pined to a given level (this is a consequence of not having a central exchange).
What is the result of implementing this S&R "tactic" on an expert advisor ? Eventhough the results of my first analysis are not absolutely incredible, it does reveal that the above given way of measuring S&R level and entering trades has some merit and - at least - a positive mathematical expectancy and some potential for profitability. The below image shows you the equity curve of an S&R system implementing the above explained criteria for entering breakouts of S&R levels (10 year backtesting period Jan-2000, Jan-2010).
The system does most of the time what I intended it to do, entering trades after breakouts of S&R zones like it is shown on the image below. However the problem now becomes to know which period needs to be used to evaluate the S&R levels and what percentages are going to be required for any given S&R level to be considered important on different timeframes. As you may see, the population is an absolute number and we need to do a deep historical analysis to see what this number should exactly be to call a level "important or unimportant". Nonetheless I consider this step a very important achievement since there is now a way to measure the population of a given price level and some mathematical criteria to establish its importance. I will continue to develop and improve this S&R technique and there will most certainly be an EA with it in the future :o).
If you would like to learn more about automated trading and how you too can learn to use, design and implement your own likely long term profitable systems based on sound trading tactics please consider buying my ebook on automated trading or joining Asirikuy to receive all ebook purchase benefits, weekly updates, check the live accounts I am running with several expert advisors and get in the road towards long term success in the forex market using automated trading systems. I hope you enjoyed the article !