Another week, another challenge,

I had a bunch of erratic outcomes in backtesting so I stripped down the code and ...

Using the PRC2 from the downloads area, and NOW the only condition in the strategy is the Math.Abs portion of the code

My pseudo code [idea to test]

IF the (Absolute Value of the ((PRC2.SqHI2 minus HIGH) is GREATER THAN
( the Absolute Value of the ((LOW minus PRC2.SqLOW2 ))
Then Go LONG ...

ie there is more potential to go up than there is to go down if that is true. The reason I am using Abs is to cover cases when the HI might already be greater than the Sqh2 .... reversing logic to go shorts ... on the screen shot you will see that the trigger bars are (in this example) in position to take SHORT, but 1 went LONG instead.. 33.3333% of trades going the wrong way do not a happy tester make!

Likely it is my mis use of the tool, but I dont know any better lol

Thanks especially for a fix ... been hitting on this all weekend trying to find the problem
Jon

 

Code

if (Position.MarketPosition == MarketPosition.Flat  && (Math.Abs(PRC2(PRC2Degree, PRC2Period, 1.62, 2.00, PRC2HATEMAPeriod, PRC2TEMAPeriod).Sqh2[0] - Highs[0][0])) >  (Math.Abs(Lows[0][0] - PRC2(PRC2Degree, PRC2Period, 1.62, 2.00, PRC2HATEMAPeriod, PRC2TEMAPeriod).Sql2[0]))  ) // go long

I believe Math.Abs simply ensures the return is a positive number. Is that what you want?

For example Math.Abs(-24312) returns 24312, and Math.Abs(24312) returns 24312.

prc is probl the most heavily repaint (recalculation) indicator.

MXASJ 

I believe Math.Abs simply ensures the return is a positive number. Is that what you want? For example Math.Abs(-24312) returns 24312, and Math.Abs(24312) returns 24312.

Cory,

Thats the plan ... it is ... just the way the results come out it has no pattern that I can determine what the error comes from. The screen shot shows two very similar, consecutive entries, but one is long and the other short.

Jon

cory 

prc is probl the most heavily repaint (recalculation) indicator.

Cory,

I appreciate the analysis.

I have looked at the code, I cant really see where it recalculates ... but it is definitely too complex for my wetware.

Any thoughts on a version that doesnt recalculate, or let it recalculate but only allow changes to the current bar? Something like a 'window' to stop changes to already plotted data?

Jon