At the suggestion of Jashfx, I have run an optimization to see what I would get if I vary the BE and BE+Pips settings on the 38/38 plan proposed by DesertEagle. In this test I only used the most recent longer-run that I have on my best data, i.e. BST 2010.
Method: FMT 4.3 trading at 5:15 GMT = 6:15 UK only. Spread is fixed at 2.5 pips and S/L and T/P are also fixed at 38/38. Data is at 99%, but the opening time on Mondays is different from Alpari UK.
Asking for: Best combination of BE and BE+Pips in a range starting at 20/0 to 30/10, in in one-pip increments, showing total pips earned from 29 March 2010 to 30 Oct 2010.
The test found 121 combinations in the series. To display this graphically, I printed out the graph and wrote the outcome on some of the data points to show the highlights, and scanned the graph back into my system. The attached spreadsheet gives the detailed results, along with that graph. There is no significant difference in the better settings, and we have several that are good.
Mark has laid out the basics of optimization on his site, so there is really no need to go further. We should not pick one-off jagged peaks, but rather something within a range that works reasonably well. The plateaued areas could be good examples, since several BE+pips work well in a sequence of BE values, as noted on the chart. Another way to do it is to look at the spreadsheet and pick the top 10 performers, see if the settings are close and pick something within the range rather than the very top. This is a way to avoid over-optimizing. Some statistics gurus may want to chime in on this also, but we try to keep this simple.
I have sorted the spreadsheet from highest income to lowest so that it's obvious right away which settings flowed to the top.
Meanwhile, I have my Alpari UK demo doing this for the whole period Jan 2010 to present, to see if this comes up with anything different, at 90% accuracy. This will take a few hours. We'll then see if anything else needs to be done for more recent months. I'm also running the same data as this test on 5% compounding to see what happens.