Using the command on top of each graph, one can vary the number of bars observed (200 for instance). However, launcing a backtest from “the earliest date displaied” the baktest is not carried on the 200 bars chosen, but it is always started at the same past date whatever is the number of bars one has chosen to display.
Is there a way to limit the backtest to the number of bars (say 200) one has chosen to observe?
This is exactly the problem, Nicholas: it doesn’t.
Now: 20 september 2016 1:00 PM; asset: South Africa 40 at 3 minutes, 200 units; the graph as I see it starts correctly 200 bars before on 20 september 2016 3:18 AM. But after the backtest the vertical black line on the equity line is reported on 19 september 2016 1:03 AM and the equity line itself starts on 19 september 2016 8:50 AM. I do not understand.
If a new run is launched back from the beginning with 1000 units instead of 200, the same phenomenon happens: the equity curve starts 24 hours in advance from the first date and hour displaied on the graph.
Let’s summarize a bit: to make my statistics I need to tell the system: “optimize only on the last 200 bars, no one more, no one less”. Maybe PRT does not allow such a standard operation, but it seems to me quite unlikely.
and subtract the end barindex number by the start one and you’ll find how much bars are tested. Then change your start date accordingly to how much bars you want to be tested by your strategy. Hope that helps.
Why not? Got 200 bars in 1 hour timeframe from August 31st to now. See example attached.
I don’t feel it’s a bug, I think this is something that is not implement. I don’t see anywhere a bar quantity selection to contain a backtest within this value.
.
.
