|
PureBytes Links
Trading Reference Links
|
Yes it
can be done. Really what you are about to do is signal averaging
minute data wrt an event, the event could be the start of the day, or a
particular weekday, or a trading signal. You can do a
search for "signal averaging" in the archives, i am sure i posted some code some
years ago and i think there is an DLL in Yahoo files under my name. I am
including some code below that averages up to four test values (you could use
foreign() to access different stock data here).
I am
sorry but i have no time to step you through the code... I haven't used it for a
long time. Run it in an indicator on Minute data and you should see a display as
show below, the Red and Blue are all daily charts Signal Averaged wrt the
Open of the day, i.e. they are what it looks like if you overlay all the days in
your DB and calculate the average for each minute bar's TestValue. The X-axis
labeling is not used, the vertical lines are half-hour
markers.
I hope
the code will give you some ideas.
Good
luck,
herman
Pd= Param("Pd",10,1,10,1);
NewDay = DateNum() !=
Ref(DateNum(),-1);
DailyBarNum = Nz(BarsSince(NewDay));
MaxDailyBarNum = LastValue(Highest(DailyBarNum));
NewDayNum = Cum(NewDay);
MaxDayNum = LastValue(Highest(NewDayNum));
FirstNewDayBarNum = LastValue(ValueWhen(NewDayNum == 0,
BarIndex()));
BarNumAtNewDay = ValueWhen(NewDay, BarIndex());
StartLastWholeDay = LastValue(ValueWhen(NewDayNum == (MaxDayNum -1) AND NewDay, BarIndex()));
ValidData = Flip(NewdayNum == 1,
NewDayNum == (MaxDayNum-1));
StartDisplay = StartLastWholeDay == BarIndex();
EndDisplay = BarsSince(StartDisplay) == 390;
DisplayRange = Flip(StartDisplay, EndDisplay);
TestValue1 = abs(C-O)/C*100;
Testvalue2 = (H-L)/C* 100;
TestValue3 = Null; //abs(H-O)/C*100;
Testvalue4 = Null; //abs(O-L)/C*100;
Result1 = Result2 = Result3 = Result4 = 0;
for (b=0; b<BarCount-1 ; b++)
{
ResultIndex = Nz(StartLastWholeDay + DailyBarNum[b]) ;
Result1[ ResultIndex ] = Result1[ ResultIndex ] + ValidData[b] *
TestValue1[b];
Result2[ ResultIndex ] = Result2[ ResultIndex ] + ValidData[b] *
TestValue2[b];
Result3[ ResultIndex ] = Result3[ ResultIndex ] + ValidData[b] *
TestValue3[b];
Result4[ ResultIndex ] = Result4[ ResultIndex ] + ValidData[b] *
TestValue4[b];
}
MaxDisplay = LastValue(Highest(ValueWhen(DisplayRange,Result1)));
Plot (NOT DisplayRange ,"",1,styleOwnScale|styleArea|styleNoLabel);
Plot (Result1,"Result1",6,1|4|styleNoLabel);
Plot (Result2,"Result2",4,1|4|styleNoLabel);
Plot (Result3,"Result3",8,1|4|styleNoLabel);
Plot (Result4,"Result4",7,1|4|styleNoLabel);
Plot (NewDay ,"VD",1,2|styleOwnScale|styleNoLabel);
Plot (Cum(DisplayRange)%30==0,"",0,2|styleOwnScale|styleNoLabel);
GraphXSpace = 5;
Here is a
challenge:
I have seen normalized charts where several securities start
at the
same starting point on the left and vary from that point
on.
What I would like to do the same thing but I want to use one
security
and normalize the last XX trading days on a chart.
For
example, if this is Friday, I would like to see 4 normalized
lines, of the
same secutiry. One for Monday, one for Tues, one for
Wednesday, and
one for Thursday. Each line would start at 9:30am and
stop at
4:30pm. This would be done on real time data down to 1
minute.
I would eventually like to enhance this to include only the
last XX
Mondays, or only the last XX Tuesdays, etc.
It would be
nice to have a parameter input so we can select a number
for "XX"
days.
My goal is to study how the market trends from day to day and
find
patterns.
I not experienced enought to figure out how to do
this. I actually
bet no one can do this!! Prove me
wrong.
Rick :)
Check
AmiBroker web page at:
http://www.amibroker.com/
Check
group FAQ at: http://groups.yahoo.com/group/amibroker/files/groupfaq.html
Check AmiBroker web page at:
http://www.amibroker.com/
Check group FAQ at: http://groups.yahoo.com/group/amibroker/files/groupfaq.html
Yahoo! Groups Links
|