Sometimes looking at charts all day you notice recurring patterns which seem to have forecasting value. In this article I am going to test one such pattern to see how well it works as a trading set-up.
The pattern I am going to choose is one I noticed whilst researching another article. It consists of 3 down days in a row and looks similar to a 3-black crows candlestick pattern.
Below is a picture of the set-up with a box enclosing the pattern, which I have labelled “Set-up A”.
My hunch is that this set-up – labelled “Set-up A” on the chart – is a reliable indicator of more downside.
In order to test whether this is the case I decided to back-test historical incidences of the pattern. I found 55 in the last 6 years.
I only included in my sample incidences occurring after the formation of a distinctive high. By “distinctive high” I meant a high that stuck up above the two bars either side of it. The high did not have to be ‘major’ – it could be a minor high as well.
I then looked for 3 down bars occurring after the high. The bars did not have to be similar in length or of any particular shape: they simply had to be 3 consecutive down-days. They didn’t necessarily have to occur right after the distinct high, but if there were successive 3 bar set-ups after a high I would only include the first one.
In order to test whether the pattern had any validity I decided to treat it like a hypothetical trading set-up, placing a stop just above the distinctive high (A), the entry point just below the low of the 3rd down-bar (B) and the target at an equal distance from the entry as the stop was (C), so that risk:reward for each trade would be 1:1.
It was then just a case of seeing how many of the 55 incidences hit their targets before their stops. If the majority hit their targets the set-up might indeed indicate a downside bias proving my hunch, if not then it would prove the set-up was invalid .
I found that in 30 out of the 55 incidences the target was met before the stop. This meant the pattern had a success rate of 55.0%.
This seemed to back up my hunch. When I checked whether the result was statistically significant, however,it failed to meet the 5% confidence level required to confirm the theory.
I was a little bit disappointed. I had thought I had seen something on my charts – a recurring pattern which seemed to predict lower prices with unerring regularity. And yet when I had actually come to test it, I had discovered it had not in fact delivered a significant edge historically.
Had my hunch been so far off the mark? It is possible. I may have been fooled into seeing a pattern which did not exist, simply out of a desire to find the next ‘holy grail’. This can lead to the mind tricking the eyes into seeing things which don’t exist.
On the other hand it was also possible the set-up I back-tested was not exactly the same as the set-up which had at first caught my curiosity.
I reasoned that maybe I had missed a crucial condition which the set-up required to be successful and conform with the ones I had noticed when I was eyeballing the pattern in the early stages. Maybe if I refined my definition criteria – for example by stipulating that the 3 down bars be of a certain length – or only including set-ups after major tops, I might get it closer to the set-up I had initially seen as possessing so much potential.
In the next article in this series I will start to refine and test the conditions I used to define the proto-type used in this article, with the hope that I will find a trading set-up which possess a more significant edge.