Hello everyone.
I am trying to apply following rule of estimating position size:
(f*q*E)/(m*ATR),
where f - fraction of current equity, q - currency pair quotation, E -
current equity, m - multiplier of volatility measure (ATR).
Assuming that SetPositionSize(...,spsPercentOfEquity) would be adequate,
I prepered following formula (here only parts):
//Backtest for AUDJPY//
a=45000/2.6729;//initial Equity = 16865,65 //
spread=0.025;
SetOption("InitialEquity",a);
SetOption("PriceBoundChecking",False);
SetOption("CommissionMode",2);
SetOption("CommissionAmount",0);
SetOption("AccountMargin",1);
/*RoundLotSize=0.01*/
TickSize=0.001;
PointValue=1;
//here comes trading rules...//
//among them there is a line with a following variable:
start[i]=BuyPrice[i];//start is a trade open price//
StaticVarSet("kurs_"+Name(),kurs);
//finally there is a MM part of the code:
f=Optimize("capital",1,0,100,0.25);
SetPositionSize((f*start)/(e*ATR(20)),2);
Unfortunately backtest results (positions' sizes) are weird. For example
in last trade variables are as follows:
f=1, start=61.525, e=2.25 (for profitable system it should be much
higher, however, such a value helps to show the embarassing differences)
, ATR(20)=0,1255 and Equity(=Cash)=3493,15.
For such values position for the last transaction should be equal to
2741,08. However, backtester set it to 7169,57.
I decided to use CBT to achieve some stability as far as Equity() is
concerned. This is my code:
SetCustomBacktestProc("");
if(Status("action")==actionPortfolio)
{
bo=GetBacktesterObject();
bo.PreProcess();
for(i=0;i<BarCount;i++)
{
for(sig=bo.GetFirstSignal(i);sig;sig=bo.GetNextSignal(i))
{
kurs2=StaticVarGet("kurs_"+sig.Symbol);
kurs3=kurs2[i];
pozycja=(f/100*kurs3*bo.Equity)/(e*ATR(20));
sig.PosSize=pozycja;
}
bo.ProcessTradeSignals(i);
}
bo.PostProcess();
}
Again bad results...
What is especially confusing when bo.Equity is higher than initial
equity (that is when system is profitable) everything is good. I really
don't know where the problem is.
Do you have any ideas? Please help.
Tomasz
