Using your data scramble-method implies the following assumptions:
daily returns are absolutely independent of each other.
intraday returns are not independet of each other, at least it seems
you wouldn't mix the OHLC-returns of one bar.
Furthermore, there have been studies that "show" that intraday return generation prozesses
at least at high frequenca level is not totally random.
So just mixing bars results neighter in a totally randomized data series nor in a
history-dependency at all time frames.
Might matter or not, depending on what one wants to test for.
@koganam: I think I got it.
I will use your sugested method and will import a self made data history with every High far above the highest High and every Low far below the Lowest Low of the data history to test for. Closes of that imported data will also always be far above the data range of the data to test for.
Now, choosing that imported data as the primary instrument and
constructing the random data to test on by user defined dataseries
I can use entry stop orders to force the entry price I want and entries will always take place since no cancelation conditions will be met. Same for exits.
That way it should work.
Comment