Back test optimizing - which model type?

New comment
 
Hello Traders

I have been testing these EA's forever am trying to ascertain which model type is the most accurate for interpretting the data. I went through the previous posts and the help files of MT4 so please do not suggest this again thanks.
I understnad that fractal interpolation of least timeframe would work best if having 5 min or 15 min system (or higher) since it can evaluate the finer 1 minute candles below.
MT4 help files however give a 'warning' that "if there are no lower time frames (I have 1 minute systems) then this method may be inaccurate".
Meanwhile, the Open prices only method is pretty stodgey also being rather a crude average of what may or may not have happened in that candle duration.
So what should I do? Test on the Open prices method and see if results then match Fractal method or the other way around?
Which is to be accurately interpreted of fairly realistic 1 minute movement?
I have built on fractal....then it gives not so great (greatly reduced) on Open prices.
I have built on open....then it gives OK on fractal but not as smoothed.
I have done this to see if i can get a 'happy median' in between on believable data.

Thanks, I look forward to your input

FxNorth
 
I shift 1 bar for everything so that I know its using the last full bar instead of (trying to use) tick data for the current incomplete bar. In this manner the results are the same regardless of the model chosen. This is how I solved the questionable results due to a lack of tick data.
 
Thanks Bryant,

I appreciate it.

Kind regards

FxNorth
 
I shift 1 bar for everything so that I know its using the last full bar instead of (trying to use) tick data for the current incomplete bar. In this manner the results are the same regardless of the model chosen. This is how I solved the questionable results due to a lack of tick data.



Bryant, can you give an example how you code your EA to Shift 1 Bar in Backtesting.

Thanks,
Wackena
 
You got to have tickchart for beliveable results. There are huge difference between real ticks, and simulated ticks, since M1 is smallest. Therefore all M1 bars are simulated. If any movements were made in 1 minute, you may got the same movement several times within an M1 bar (because of platform trying to distribute the volume somehow) instead of the real nature of price movement. This causes LOTS of faulty orders...
New comment
Reason: