Hello members of the DFC!
This is my first article after the return from holiday, and as I have not recovered from all the willingness to look at programming strategies, in this article I will not teach how to develop trading strategies, but I will present a study on the strategy that I am currently using in Strategy Contest.
Given that the performance of my strategy used during July and August in Strategy Contest was very similar and very positive, I decided to
make a study of 2 variables that I would like to improve in my strategy. So, I gathered all the information about closed orders during August from the Strategy Position Report and decided to work it into our good old friend Excel.
Therefore, taking into account that what I want is to improve the strategy for Strategy Contest, I decided to analyze the next points:
- The time that each order was open: in order to try to reduce the orders that are open more than 12 hours, a half day. As in the competition our strategy have to open at least twenty orders, are not convenient to have orders that remain open for a long time. Since I've seen strategies to stay out of the TOP-10 because they did not open the necessary orders, I think this point is very important.
- The maximum Drawdown: of each order to avoid the risk we take a margin call during the contest. We should pay particular attention to the open orders over the weekend, where Dukascopy low the leverage from 1:100 to 1:30.
I think that if I can control and eliminate orders that cause very large deviations of these two variables, my strategy will continue to have the same good results in the future.
This image is a graph in Excel where is represented the duration, in minutes, of closed orders for my strategy during the month of August. As you can see, if we remove the orders that exceed more than 12h open, a half day, the distribution of the time that they are open is fairly homogeneous.
And how the strategy works in 5 minutes TIMEFRAME, if the trades opened not take too long to close, more opportunities it will be to open new orders.
As you can see in this picture, the same happens with the drawdown. If I can remove the orders with more than 10% drawdown, the distribution is fairly homogeneous.
In my next version of this strategy will try to filter the orders that may have more than a 20% drawdown, and then maybe 10%.
By analyzing the graph beside, concerning pips of each open order from the strategy was to lose before closing in profit, the first amendment to do will be reduce the StopLoss from 200 to 120 pips. As well as a backtest at least one year.
I think that if I can control and eliminate orders that cause very large deviations of these two variables, my strategy will continue to have the same good results in the future. So, during the month of September the strategy that is still running is the same as in July and August. But if I have time to make these changes during the month of September in October I will have a new strategy that hopefully get better results.
When I decided to write this article, it was because I looked back on the course of a year of strategy contest and noticed it had been a year fruitless for me. And why? Were the other strategies so good and my so bad? No, the problem was really me, and each month a strategy that did not win, I changed strategy or changed the code without first completely analyze what went wrong. This time I decided to test and analyze the same strategy in order to find out what can be improved. So my advice to all members who are developing strategies before giving up and leaving for another, try to really see what can be improved.