A different way to access how profitable the strategy is.
Often a strategy was very profitable only to find that is not during the demo testing, this I think is the way we track strategy by using the normal way to track profit because it emphasis high profitability in the past trades, hence the current performance need not be even profitable because the past profit can cushioned the poor current performance.
We need to figure a mechanism to give emphasis to strategy that is doing very well in the near past while still profitable in the distant past for it to be "robust" in the future.
This is a concept as I still have not figure out how to assign a measurement so the mechanism that prefer better profitability in the current past over that of the distant pass, maybe we section the testing period over multiple section and measure the profit over each period and give higher weight to strategy that perform better in the current past over that in the distant pass as big past profit allow for bad current performance and confirm it with the normal way profit is track.
May be one way is to assigned the profit for the last trade to be zero and the traded lot size to be say at 100 lot and calculate the equity in reverse order of the trades, what the capital would have to be at the first trade, the lower the capital the better the better the strategy ( for profitable strategy, the initial capital is suspected to be very negative figure).The effect of the distant past needs to be negated ie if the strategy call of addition of positions, in the reverse calculation it should do the reverse and calls for a reduction as we step backward in timeline. This will negate the effect of high profit in the pass to cover poor current performance.
This calculation is to be use for the strategy generation only, the normal way of profit calculation is to be used display after the optimization.