Disliked{quote} Hi Dave, I wonder if it would help the cause to implement a "2-D" optimization graph similar to that produced during MT4 optimizations. The benefit of the graph is that it illustrates visually whether the backtest result of any particular set of parameters was simply random or is likely to be repeated in live trading. See the 2 attached image files. Each square represents an optimization run based on a different parameter setting as the algorithm "steps" through the process. The dark green squares represent the parameter settings that produced...Ignored
The one problem about the multi dimensional optimizations is precisely what the article alludes to, and that is what factors need to have more emphasis. For example, having decided that Break Even at 10 seems too short, do we then abandon Break Even as a consideration and leave the default of 30 and work on other areas or do we adopt an idea of trading in a basket of currencies each with their own behaviour.
When I achieved a better optimization result on the GBP leaving the default at 30 and ticking some other boxes (eg. Percentage of Equity), there were better optimization results. However when I tried the same on AUD, they were worse when I un-ticked Set Stop Loss To Break Even.
If it were just 2 or 3 variables being optimized the 2d presentation carries weight but too many assumptions with an 8d optimization but I will have a look at some type of variance report.