Backtestable?

[Shakone]

I was thinking about my strategy as I do from time to time, and wondering why it works and so on.

And I had the thought about whether backtestable strrategies can work. My strategy is not possible to backtest (if that sounds strange to you, then you're not reactive to what happens on the day).

On the one hand, it is possible that a successful strategy in the past might be worthwhile, but from my perspective, if it was successful, it would be exploited and then ruined.

So a question for those more knowledgeable and more experienced than I...Should people be aiming for a strategy that is actually impossible to backtest?

 

[Rossini]

No, strategies that can be back tested can work, as can strategies that are very difficult or imposable to back test.

 

[£10kLoser]

What you're really saying is that a successful trading method changes according to conditions. I fully support that concept and the idea of creating a system by backtesting is futile.

After all, hundreds of thousands of people have been doing that for years and got nowhere (or at least 99.999% have got nowhere). This crops up regularly and the fact is that backtesting is basically curve fitting to the past. Even a totally random chart will have patterns (or it wouldn't be totally random) but to assume that those patterns will continue tomorrow has cost traders millions and millions of dollars...