Strategy Tester : optimized results different when replayed

New comment
 

Hi there !

On my last post I was wondering how to work with OHLC, founding no solution, I now use for the EA ticks. When I optimize, results appears, but when replayed the results differs. What that could be due to please ?


thanks.

 
Icham Aidibe:

Hi there !

On my last post I was wondering how to work with OHLC, founding no solution, I now use for the EA ticks. When I optimize, results appears, but when replayed the results differs. What that could be due to please ?


thanks.

Always different between test and real result.
 
updated history file.
 

In every tick mode MT4 generates .fxt file (pseudo ticks) to make a test. I said pseudo ticks because those are not real ticks, that means that from time to time it generates new tick file (eg when you update testing period)

So after that even if you go back to prior date from - date to then you will have "new" ticks. If you are testing ea like grid, especialy where multiple positions are closed in the same time, one or few ticks makes

a big difference. The same issue you will find when trading breakouts with breakout buffor etc.

The solution that I use for this, is to use M1 chart, and read all values on closed bars on higher timeframes.

I described it here: http://sharpquant.com/general/ea-testing-why-open-prices/

I hope it helps :)

EA testing – why “open prices”?
EA testing – why “open prices”?
  • 2015.07.13
  • Kale
  • sharpquant.com
Metatrader platform provides us with few different testing models. There is a little difference between MT4 and MT5 testing models. If you are interested in detailed information about each model, including info about “how ticks are generated” I recommend you to visit this link: Personaly I’m using only 2 models, and sometimes I mix them...
 
Bartlomiej Gorski:

In every tick mode MT4 generates .fxt file (pseudo ticks) to make a test. I said pseudo ticks because those are not real ticks, that means that from time to time it generates new tick file (eg when you update testing period)

So after that even if you go back to prior date from - date to then you will have "new" ticks. If you are testing ea like grid, especialy where multiple positions are closed in the same time, one or few ticks makes

a big difference. The same issue you will find when trading breakouts with breakout buffor etc.

The solution that I use for this, is to use M1 chart, and read all values on closed bars on higher timeframes.

I described it here: http://sharpquant.com/general/ea-testing-why-open-prices/

I hope it helps :)

Okay so trading ticks is better with M1. Thanks dude!
 

You know there's ticks between the Open ---> Close. You won't occult it, the new opening candle price may be shifted from a session to another.
I have the same issue, I need real ticks to improve strategies but checking every change on my code by every tick made it long long long, & now expensive ! I hope a solution will be found to answer to that.

New comment