PureBytes Links
Trading Reference Links
|
Hi DT,
About backtest and timed frame
you're right: there will be no "bug" if the begin of the data loaded
is always the same.
BUT if you test your system always with the last 300 bars, the
beginning of the data will be different and --------> "bug"
it is the reason why I ask for a stable time frame based on weekly
data from monday to wednesday where the close is the last close, the
high is the highest, the open the first open and the low the lowest.
the
weekbeg = dayofweek() < ref( dayofweek(), -1 );
gives "1" but.... on monday
it will be nice to know on friday the values of the timedframe in the
purpose to built a indicator based on weekly timedframe and apply it
in a system tester with others indicator based on daily data.
I want to test on daily data the association of weekly indicators and
daily indicators in the purpose to built a swing trading system (
that is already built on MSwin but can't be correctly tested)
S.C
In amibroker@xxxx, "DIMITRIS TSOKAKIS" <TSOKAKIS@xxxx> wrote:
> cum(1) is from the beginning of your data.
> So, if Friday Sept 7 has cum(1)==300, on Monday Sept 10, if you
> select again Friday, its cum(1) will ba again 300.
> And it will be always 300, except if you add data at the beginning.
> So, for Friday Sept 7, cum(1)%8==300%8==(37*8+4)%8==4.
> D. T.
> --- In amibroker@xxxx, "Stephane Carrasset" <nenapacwanfr@xxxx>
wrote:
> > Tz,
> >
> > You wrote
> > <<
> > Hi Dimitris, For example if you want to generate a "1" every 8th
> bar
> > you can now write impulse8 = ( cum(1) % 8 ) == 0; This is useful
> when
> > you try to simulate different periodicity from daily data: close8
=
> > ValueWhen( impulse8, close ); high8 = HighestSince( impulse8,
> high );
> > low8 = LowestSince( impulse8, low ); Best regards, Tomasz
Janeczko
> > >>
> >
> >
> > but the 8time frame bars ( or another lenght of time frame) are
> > calculated from the last data isn't it ?
> > and change when new data are added
> > so it can't be use in backtest
> >
> > right or wrong ? ?
|