Yeah it’s difficult to choose the results of a losing WF period, but without the optimum results the loss could be worse?
I guess that this is one of the problems that I have with regularly re-optimising. You just had a section of data that you traded with some set variable values from a previously optimised section of data and you made a loss and then you optimise that data and it tells you that you should have been using totally different values but you would still have made a loss, just a smaller loss but then lets say that you add that section of data to the previous section of data and optimise that – you would have made a profit with totally different values but is it more than the two data samples traded with different values?
Which optimised values do you persuade yourself will work best in the upcoming data – the values that made a profit but maybe less than the values that made a profit and then a loss?! You are just curve fitting recent history and hoping that curve fit works going forward or you are saying that you didn’t like the data you curve fitted to and so adding more data till you find a curve fit that made a profit and an equity curve that you like.
