I have found in my back testing that you have a max DD which for me be like three losing days max in a row for day trading and that comes up once every three years and in between I have mini DD's of one losing day of Max, so if I was to implement a new signal even though I would have backtested the crap out of it, I would wait till I had losing day and next day start using it.
Now through the years I have bought many systems from venders, I have found six that do fairly well except they will have up to 50% drawdown before making new equity highs, so again, did much backtesting, they programmed and waits for 80% drawdown of the "mean" before it can take signals and continues till 80% of the "mean" of equity curve going up, so the overall drawdowns are acceptable. Besides I came up with idea after drinking much wine, this works for me, but it was difficult to start in beginning cause it goes against what seems logical. We all want to get on board a trend as it feels good to do so, but markets not logical-they move without any news "just because", and I have overcome emotions and most trades look like they make no sense....
Any trade is 50/50.
It's true that often markets move for no apparent reason. Even without any volume ! But they move. And we search for an explanation EX POST and we find a rational, a pattern, most of the time spurious.
It's good to drink wine =P Especially Red one. My favorite "alcohol". And we could talk days and nights about the insufficient reason of trades outcomes. Is it a fundamental uncertainty ? Or an Epistemological one ? Look at your backtest , it's an epistemological uncertainty. Your trades aren't splitted this way around loosers and winners (50/50). If we can describe and anticipate or even recognize conditions that are favorable with our signals. Then we can increase the fitness, accuracy, of our system, trades. IMO.
Ok. But I wasn't talking of denaturalization of the strategy by adding, removing signals. It's just that I'll be less sensitive to signals. Like if there are louder ones than other. I'll take into account only strong signals.
