Who wants a strategy? Lots and for free) - page 55

 
zfs писал(а) >>

Where do these gaps outside the optimisation period come from?

They follow the optimisation period...

 
Figar0 >> :

They follow the period of optimisation...

Isn't the optimization period equal to the whole period. Or maybe it's a sample from this period... And the parameter is the number of samples...

 
And the quality of strategies has indeed deteriorated, at least visually, the likelihood of a strategy with good characteristics being issued has decreased...
 
zfs писал(а) >>
And the quality of strategies has indeed deteriorated, at least visually, the probability of issuing strategies with good characteristics has decreased...

Maybe it's not the quality of the strategies that has deteriorated, but the quality of the tester/optimiser that has improved:)

 
Figar0 >> :

Maybe it's not the quality of the strategies that has deteriorated, but the quality of the tester/optimiser that has improved:)

Maybe. I was just testing with a certain stop timeframe step, and the result I got in the new version was inferior in some respects to the previous ones, although I expected better result and much better. Although it is not an indicator, of course...

 

I can't program the indicators from the FSB. I have an idea that there is a problem with averaging. I never thought I would encounter a problem with Oscillator of MACD.

for(int i=0; i<limit; i++)
MacdBuffer1[i]=iMA(NULL,0,FastEMA1,0,MASmooth,BasePrice,i)-iMA(NULL,0,SlowEMA1,0,MASmooth,BasePrice,i);

for(i=0; i<limit; i++)
MacdBuffer2[i]=iMA(NULL,0,FastEMA2,0,MASmooth,BasePrice,i)-iMA(NULL,0,SlowEMA2,0,MASmooth,BasePrice,i);

for(i=0; i<limit; i++)
OscBuffer[i]=MacdBuffer1[i]-MacdBuffer2[i];

This is all the code in principle. However, the data in FSB do not match. Can anyone help decompose the iMA construct or can someone share the codes.

Or maybe the developer Miroslav will clarify the calculation of averaging parameters in his program in more detail.

HELP!!!

 

Hello,

Oscillator of MACD = MACD1 - MACD2

The MACD in FSB is the same as MACD in MT.

Test first both MACD1 and MACD2 (Use single MACD). It hasn't be any difference. In the other case correct MACD first.

 


USDJPY 1h












Close Close Close Close







19.03.2008 16:00 1 93,85


19.03.2008 17:00 2 93,91


19.03.2008 18:00 3 94,53


19.03.2008 19:00 4 94,5


19.03.2008 20:00 5 94,51 94,51

19.03.2008 21:00 6 94,5 94,5

19.03.2008 22:00 7 94,5 94,5 94,5
19.03.2008 23:00 8 94,52 94,52 94,52 94,52
20.03.2008 0:00 9 94,5 94,5 94,5 94,5
20.03.2008 1:00 10 94,63 94,63 94,63 94,63


Simple 94,395 94,52667 94,5375 94,55



Slow2 Slow1 Fast2 Fast1










MACD1 0,023333




MACD2 0,1425











Osc -0,11917











By FSB MACD1 0,0211




MACD2 0,2949




Osc -0,2738










by MT MACD1 0,035




MACD2 -0,019




MyOsc 0,023




Difference 0,054
 

I couldn't resist - I went over it... Please tell me what I'm doing wrong... 3 variants have different results. MACD1(Simple,Close,3-fast,6-slow) - MACD2(Simple,Close,4-fast,10-slow).

You can also see from everything that I can't even subtract 2 indicators programmatically.

 
My manual values are the same as the averages projected in MT. The question is where do the MACD values come from, because MACDs are like FastMA-SlowMA.