Optimizing the forex trading system parameters: EUR/USD

User Rating: / 2
Written by Forex Automaton   
Friday, 24 April 2009 14:28
Article Index
Optimizing the forex trading system parameters: EUR/USD
Optimizing the forecasting parameter
Optimizing the stop-loss parameter
Optimizing the entry and exit parameters
Summary of progress
All Pages

I continue with the series of forex trading system optimization reports for the individual currency pairs, traded on the day scale, which began with AUD/JPY. In the algorithm, the forecast signal whose nature will not be disclosed is fed into money management framework driven by three adjustable parameters. In the current test setting, the adjustable parameters form a multidimensional space populated by figures of merit, as obtained by Monte Carlo simulation of independent trading histories. The interpretation of the data set leads to a quantitative discussion of relative merits of various trading styles. Two features of the data set make this material unique: first, the data incorporate the real inefficiencies of the real market, as opposed to more academic simulations with martingales. Second, the money management styles are being tested in a setting where the forecasting part is entrusted to the generally successful "artificial intelligence" component.

Analysis approach and the data set

It can not be overemphasized that for the analysis to be of any value, the algorithm may not trade the data used to train its decision making. A run of the program included simulations of trading histories of over 13,000 independent "virtual traders" (forex robots), each of them being an incarnation of the same algorithm, differing by the setting of the adjustable knobs. This report uses the EUR/USD day scale data covering the time interval from August 20, 2002 to September 1, 2008. The key concepts of conditional projection distributions and profile histograms have been explained in the previous report.

There are two key differences with the approach used in the AUD/JPY report. First, the comparison with the results from the same algorithm trading simulated markets is abandoned as the issue is believed to be exhausted. Second, instead of one performance quality condition, we track several of them in hope that the possible systematic dependences will be more visible that way.

Last Updated ( Monday, 04 January 2010 12:36 )