The time frame says nothing about the period you look back,
Well, it says everything about the period you can look back.
I agree. When I see a long term backtested System on, for example, the 5 min TF, and trades are open for the majority of the day or even many days … my conclusion, generally, is that the System is not trading the 5 min TF.
That seems obvious to me, but possibly I don’t understand what you mean to say. Anyway :
When I make s system for a 1s TF it is totally obvious that I can’t even use a 2s BackTest ion it. Let alone 5 minutes. This seems the other way around from what you are saying, but that is obvious just the same (I think). Trading at the broker works in real time, which can be boiled down to “per tick”. If you have a 1H TF which your strategy acts upon, it works on that one tick-change at the full hour (each full hour); all in between, the strategy can’t act upon. Additionally, if you made the strategy to work per the full hour, you really can’t expect that it works even remotely the same when run live and than per 1 second.
Maybe it depends on how the details in your code are, but mine is full of anticipation of the most real time possible (1 second). This is why it can’t work with another TF. The clear downside : I can’t backtest for more than 1 month (1M seconds of live-market data).
.
