|
PureBytes Links
Trading Reference Links
|
This was an experimental method to begin the prediction of next bar
Low. Although it is obvious, I will add verbatim here that I would be
more than glad to see a better proposal.
We are lucky to have all the comparison tools in AFL structure.
Any improvements/suggestions/hints/proposals appreciated.
Dimitris Tsokakis
--- In amibroker@xxxxxxxxxxxxxxx, "Dimitris Tsokakis" <TSOKAKIS@xxxx>
wrote:
> The Question : To anticipate the next bar Low.
> The main Hypothesis : Next bar RSI() will not be lower than min(RSIA
(H,14),min(RSIA(L,14),RSIA(C,14)))
> Test period=the last 200 bars
> Errors:
> Error05<0.5%
> Error1<1%
> Error2<2%
> Statistics:
> Longterm05, the percentage of error<0.5% for the last 200 bars
> Longterm1, the percentage of error<1% for the last 200 bars
> Longterm2, for error<2%
> The first attempt to answer:
> Paste in your Indicator builder [or Automatic analysis] window the
>
> // Anticipate next bar Low, by D. Tsokakis, June 2003
> R1=RSIA(H,14);R2=RSIA(L,14);R3=RSIA(C,14);
> MinR=Min(R1,Min(R2,R3));
> FACTOR=0.025;
> CoL=1-FACTOR;
> Value=MinR*CoL;
> WildPer = 14;ExpPer = 2 * WildPer - 1;
> AUC = EMA( Max( C - Ref( C, -1 ), 0 ), ExpPer );
> ADC = EMA( Max( Ref( C, -1 ) - C, 0 ), ExpPer );
> x = (WildPer - 1) * ( ADC * Value / (100-Value) - AUC);
> RSIsupport = IIf( x >= 0, C + x, C + x * (100-Value)/Value );
> IntradaySupport=Ref(RSIsupport,-1);
> NextBarSupport=RSIsupport;
> Plot(IIf(Cum(1)>LastValue(Cum(1))-10,LastValue(IntradaySupport),-
1e10),"",4,8);
> Plot(IIf(Cum(1)>LastValue(Cum(1))-3,LastValue(NextBarSupport),-
1e10),"",7,8);
> Plot(C,"",1,64);
> App1=1;Approx1=abs(-1+L/IntradaySupport)<=App1/100;test=200;
> Longterm1=0.5*Ref(Sum(Approx1,test),-1);
> App2=2;Approx2=abs(-1+L/IntradaySupport)<=App2/100;
> Longterm2=0.5*Ref(Sum(Approx2,test),-1);
> App05=0.5;Approx05=abs(-1+L/IntradaySupport)<=App05/100;
> Longterm05=0.5*Ref(Sum(Approx05,test),-1);
> Title=Name()+" next bar Low prediction with error < "+
> WriteVal(App2,1.0)+" % was reliable the "+
> WriteVal(Longterm2,1.0)+" % of the last "+
> WriteVal(test,1.0)+" bars [ "+
> WriteVal(Longterm1,1.0)+" % with error <"+
> WriteVal(App1,1.0)+" %], [ "+
> WriteVal(Longterm05,1.0)+" % with error <"+
> WriteVal(App05,1.1)+" %]" ;
> Filter=1;
> AddColumn(Longterm05,"Longterm ERROR<0.5%",1.0);
> AddColumn(Longterm1," Longterm ERROR<1%",1.0);
> AddColumn(Longterm2," Longterm ERROR<2% ",1.0);
> AddColumn(MA(H/L,200),"Average spread");
>
> Some first notes now :
> *The 10-bar red line is today support, already known from yesterday.
> *The 3-bar yellow line is the prediction for the next bar Low.
> *As soon as you download intraday EOD for the new day,
> a. the yellow line will automatically become red and will be
the same for the whole session. It is the intraday support.
> b. a new yellow line will appear. It is the next [unknown] bar
support, it is variable as new OHLC come and will have
> its final price by the end of the session.After the first
half of the session, this yellow line [with a reliability
confirmation]
> is very interesting, especially when you see high volume
positions at this level. Sometimes you will have the impression
> that some other people also "know" your secret yellow line.
> *Highly volatile stocks [MA(H/L,200)>1.08 !!] are not that reliable
and should be handled with care.
> *Change the filter condition to
> Filter=Approx05;
> or
> Filter=Approx1;
> or
> Filter=Approx2;
> explore for the n=1 last quotations
> to see the stocks with actual/predicted Low less than 0.55, 1% and
2% respectively. for the June16 N100 session we saw
> 86 stocks with 2% approximation, 68 stocks with 1% and 31 stocks
with just 0.5%. Not bad at all for a positive biased day.
> *FACTOR may vary between 0.2 to 0.5 with similar results.
> *Red and yellow line at tha same level may denote accumulation at
this level and [probably] a turning point...
>
> The attached gif explains why Copper is my favorite. It was 90%
reliable, ie for 180 of the last 200 bars the error between
> predicted and actual Low was less than 2%. For 126/200 bars the
error was less than 1%. [for 80/200 bars the error was less
> than 0.5%]
> This is the first, short description of this experimental method.
> Improvements will hopefully follow.
>
> References :
> 1. G. Siligardos, Reverse Engineering RSI, TASC June 2003
>
http://www.traders.com/Documentation/FEEDbk_docs/Archive/062003/Abstra
cts_new/Siligardos/siligardos.html
> 2. Tomasz Janeczko, TASC June 2003
>
http://www.traders.com/Documentation/FEEDbk_docs/Archive/062003/Trader
sTips/TradersTips.html#amibroker2
> 3. Tomasz Janeczko
> http://www.amibroker.com/members/traders/06-2003.html
> 3.Two older messages
> http://groups.yahoo.com/group/amibroker/message/2751
> and
> http://groups.yahoo.com/group/amibroker/message/2758
> The list began this discussion two years ago [June21, 2001]!!
> 4.The analytic AFL for RSI()
> http://groups.yahoo.com/group/amibroker/message/5461
> 5. My unsuccessful efforts to create RSI Keltner channels though
the RSIHigh, RSILow
> http://groups.yahoo.com/group/amibroker/message/5476
------------------------ Yahoo! Groups Sponsor ---------------------~-->
Get A Free Psychic Reading! Your Online Answer To Life's Important Questions.
http://us.click.yahoo.com/Lj3uPC/Me7FAA/ySSFAA/GHeqlB/TM
---------------------------------------------------------------------~->
Send BUG REPORTS to bugs@xxxxxxxxxxxxx
Send SUGGESTIONS to suggest@xxxxxxxxxxxxx
-----------------------------------------
Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx
(Web page: http://groups.yahoo.com/group/amiquote/messages/)
--------------------------------------------
Check group FAQ at: http://groups.yahoo.com/group/amibroker/files/groupfaq.html
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
|