As the creator of the "decrease factor" I have given it a lot of thought and used it in some different tests. I think it does do a lot of curve fitting. But it usually improves things. I think A 990505 might be the best most of the time. One series of trades where your B 995005 works best is when, you get one win followed by one loss back and forth. In such a situation A will get you 5 % on each win and 99 percent on each loss, whereas B will get you 50% on each win and 99 percent on each loss. I had a currency where I saw that pattern. I have been contemplating changing the decrease factor to work with the last 4 trades. In a choppy market running one win one loss one after the other you would get the same factor either 5 percent with A or 50% with B. Again it is all curve fitting. If the decrease factor works with the last 4 trades instead of the last 3 only, you will be out of the action more.

The other improvement on the decrease factor would be to distinguish the buys from the sells, and base the decrease factor from buys on the number of successes for buys, and the same for sells.

My cable modem was fried by an electrical storm, that hit a special fan switch, the kids Wii, our VCR or DVD. So I ran some optimizations, while I was out of the water for 4 days, on the max trades parameter on mode with 5.7.3, and I found that it intensified the amount of money gained or lost.

I think as a last step, tune the decrease factor and number of trades by the currency. No one setting is perfect, neither mine or yours.