Download MetaTrader 5
To add comments, please log in or register
Do you need a programmer? Freelance service will find the developer you need!
Luciano Ola
996
Luciano Ola 2014.04.11 07:45 

I've just finished backtesting my EA with dukascopy 99.9% tick modeling, the results are quite different from the lower modeling quality results.

I have double checked the dukascopy data with the mt4 charts which come from another broker and there are now price explosions from dukascopy, the charts appear to match up.

The ostensible issue is the significance of the lot size used (as I begin trading with the minimum lotsize) I have ensured that the parameters used during the dukascopy EA match that of the broker used in previous tests.

Here are my results:

Previous back-test

Test length 8 years Time frame is D1

Dukascopy back-test

Test Length ~5 years Time frame is D1

I am very confused as I did not intend for the EA to be this aggressive, and I understand the risks associated with an EA that is this aggressive, but I am already using roughly the minimum lot size.

Is it possible that the previous broker has changed it's required margin?

Btw on the second chart I had reduced the aggression of the EA (by making the Lotsiz three times smaller), so that it had a third of the aggression of the initial chart, as I wanted to reduce the drawdown.

Luciano Ola
996
Luciano Ola 2014.04.11 15:29  
Any suggestions would be greatly appreciated.
Luciano Ola
996
Luciano Ola 2014.04.11 15:51  

I would appear the spread I used for the first one is 5 times larger than the highest spread that broker has ever had, and the second is roughly 10 times smaller than then non weekend spreads. I'm not sure if this may have caused the discrepancies but it this does seem to be likely. This is quite odd because the spread from the broker historical spread is not the same as the spread provided when I don't use the dukascopy data (and I am connected directly to the broker's server).

Wasn't the spread it was the tick value.

Luciano Ola
996
Luciano Ola 2014.04.11 19:22  
Wasn't the spread it was the tick value.
Daniel Petrovai
440
Daniel Petrovai 2014.04.12 03:01  
MetaNt:

I've just finished backtesting my EA with dukascopy 99.9% tick modeling, the results are quite different from the lower modeling quality results.

I have double checked the dukascopy data with the mt4 charts which come from another broker and there are now price explosions from dukascopy, the charts appear to match up.

The ostensible issue is the significance of the lot size used (as I begin trading with the minimum lotsize) I have ensured that the parameters used during the dukascopy EA match that of the broker used in previous tests.

Here are my results:

Previous back-test

Test length 8 years Time frame is D1

Dukascopy back-test

Test Length ~5 years Time frame is D1

I am very confused as I did not intend for the EA to be this aggressive, and I understand the risks associated with an EA that is this aggressive, but I am already using roughly the minimum lot size.

Is it possible that the previous broker has changed it's required margin?

Btw on the second chart I had reduced the aggression of the EA (by making the Lotsiz three times smaller), so that it had a third of the aggression of the initial chart, as I wanted to reduce the drawdown.

Quite difficult to get an informed opinion when you don't make the code available.

You mentioned different brokers. You do understand that different brokers have different server locations and different start/end of day times.

Depending on what your EA is taking into account, results may be different from broker to broker.

The open, close, low,high, volume per bar on D1 based chart may be different and so could be your results.

So, unless you post the code, I doubt that you can get more detailed and more precise help.

Cheers

/
To add comments, please log in or register