Friday, February 27, 2009

Backtesting God's Gift, Consistent Profits Over Several Pairs

A few days ago I introduced my new version of the God's Gift expert advisor through a blog post (my newsletter subscribers received a detailed email explaining that release in detail as well as the 8and8 expert advisor). Now I want to talk a little bit about the optimization jobs I have been doing with this expert and how the God's Gift expert advisor proves to be profitable amongst several different forex pairs.

In order to know if the ea was profitable or not, I started by back testing it in the GBP/USD pair using the settings I had known to be profitable form last year's forward testing (12 months). According to the values of stop loss and trailing stop used, I realized that they would be the equivalent of about 30% of the ATR for the stop loss and 15% of the ATR for the trailing stop, however, I realized that the trailing stop was much more profitable if it was changed from 15% to 70% of the ATR, also decreasing the risk to reward ratio of the expert advisor. The back test graph I got was the one shown below.
Now that I knew that the expert could trade the GBP/USD profitably, I wanted to know if this time I could get it to work with other forex pairs. My previous efforts to do this had failed dramatically as I could not find any other pair that was profitable since 2004 or before with God's Gift 7c besides the GBP/USD.

The first pair I started to test was the EUR/USD. The same settings failed to work straight from 2004 to 2009, but I did see that the settings used for worked perfectly when the pair was in high correlation with the GBP/USD. The solution was to optimize the stochastic parameters of the ea in order to find a better fit for the EUR/USD. In my mind, the stochastic parameters, which determine when to enter a trade, are what defines the "psychology" of the pair. After a lot of effort, I got the backtesting curve you see below (this one is from 2000 to 2009).
Again, I decided to retest the expert advisor with another pair, the USD/JPY, trying to still refit the stochastic values against the pair's trading style. As if the expert advisor could magically adjust to every market condition, I found a set of parameters that worked the USD/JPY perfectly from 2000 to 2009. The image is shown below.
As you can see, the new expert advisor does really good in backtest (and we know this correlates very good with forward and live testing according to last year's results) and does not have any of the "black holes", so common in today's commercial expert advisors. This expert has good linear equity growth, with loses, fixed stop loss values and trailing stop values you can see and set. The expert has a very good risk to reward ratio as all the backtests show that the average win is almost double the maximum loss.

If you would like to learn more about this new version of the God's Gift expert advisor I coded and the results we had with this and other expert advisors last year (as well as this one) 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: