Out of curiosity I tried the original DMA with a short window of 2 days and and a long window of 10 days.
My hope was that this might be able to exploit shorter market fluctuations and generate profits earlier on (the previous one took several years until it generated profit), at the cost of higher volitatility.
As you can see, the volatility (and the max drawdown -- ouch) certainly were increased. However, the performance is pretty poor overall.
In the future it will be possible to define a range of parameters that will automatically get tested but for now those values have to be tuned by hand.