[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[EquisMetaStock Group] Re: Rules Management for Trading System



PureBytes Links

Trading Reference Links


Hi MG Ferreira:
  For Kalman Filter's weighting factor, in term of selecting periods 
for evaluation and out of sample.  If I evaluate 4 indcators' 
performance using the periods between 2000 and 2003 and adjust each 
weighting factors based on their own ROIs, then each indicator has 
been trained/optimized during this periods [2000 - 2003] for getting 
the system's steady state.  However, If I load any data within the 
evaluation period by opening the chart, such as opening and loading 
data between 2002 to present [2002 and 2003 overlapping the period 
for evaluation] then will each indicator re-trained again during this 
period by using the same information and cause the problem of curve 
fitting?  Do you have any suggestion on how to select the length of 
period for evaluation and out-of-sample?

p1 = 20% [2000 - 2003] then w1 = 1000
p2 = 20% [2000 - 2003] then w2 = 1000
p3 = 40% [2000 - 2003] then w3 = 2000
p4 = 20% [2000 - 2003] then w4 = 1000

Should I open the chart and load the data starting after 2003 to 
avoid this problem?

================================================================

On the other hands, if I combine 4 different indicators together, 
such as

INDICATOR 1 - RSI
Sl = +(RSI(14)>30)*(REF(RSI(14),-1)<=30);
Ss = -(RSI(14)<70)*(REF(RSI(14),-1)>=70);
S  = Ss+Sl+(Ss=0)*(Sl=0)*REF(S,-1);
Vl = RSI(14)-30;
Vs = RSI(14)-70;
T  = Vs*(S<0)+Vl*(S>0);
Tfin1 = T/Max(ABS(T));

INDICATOR 2 - Moving Average
Sl = +(C-MOV(C,40,S));
Ss = -(MOV(C,40,S)-C);
S  = Ss+Sl+(Ss=0)*(Sl=0)*REF(S,-1);
Vl = RSI(14)-30;
Vs = RSI(14)-70;
T  = Vs*(S<0)+Vl*(S>0);
Tfin2 = T/Max(ABS(T));
...

Peroformance
p1 = 20% then assign w1 = 1000
p2 = 30% then assign w2 = 1500
...

For indicator 2, should I use 14 for moving average rather than 40 
for avoid the potential problem of out of sync? or will penalty 
function solve this problem as show below?

dr = C-REF(C,-1);
p1 = (dr*Tfin1>0)-(dr*Tfin1<0);
p2 = (dr*Tfin2>0)-(dr*Tfin2<0);

pc = 0.01
w1 = w1*(1+p1*pc);
w2 = w2*(1+p2*pc);
...

w = w1 + w2 + w3 +w4;

T = (w1*Tfin1 + w2*Tfin2 + w3*Tfin3 + w4*Tfin4) / w / 4;

If yes, then will it be no problem by combining different indicators 
with different periods together as long as each one multiply each own 
weighting factor? such as

Indicator 1 using 10 periods * w1 +
Indicator 2 using 30 periods * w2 +
Indicator 3 using 50 periods * w3 +
Indicator 4 using 60 periods * w4

Do you have any advise?
Thank you
Eric





------------------------ Yahoo! Groups Sponsor --------------------~--> 
What would our lives be like without music, dance, and theater?
Donate or volunteer in the arts today at Network for Good!
http://us.click.yahoo.com/Tcy2bD/SOnJAA/cosFAA/BefplB/TM
--------------------------------------------------------------------~-> 

 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/equismetastock/

<*> To unsubscribe from this group, send an email to:
    equismetastock-unsubscribe@xxxxxxxxxxxxxxx

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/