Tips for robustness

[FxWRLD]

Hey guys,

I've been using FX Dreema for a couple of weeks and really loving it. I've finally managed to create a successful EA but it only works on certain currency pairs and certain timeframes (i.e. USDJPY, 5m, 01/01/2022-07/03/2023).

I realise that I may need to edit some of my conditions to make it successful for different currency pairs however when I change the date range to 01/01/2021 for USDJPY it loses money.

Does anyone have any tips I could incorporate into my EA or any advice when it comes to making it more robust and be able to handle different market conditions?

Thanks in advance

[seb 0]

@FxWRLD Unfortunately there is no universal tip to make an EA "more robust", any improvement depends on the strategy itself.
If your EA only makes profit on a few markets\timeframes\periods it is probably because it is not profitable at all. These winning periods might only be due to luck\conjecture.
Still you can look at the mql5 website, "articles" section, you will find a lot of "tips", even thought its not always easy...

[FxWRLD]

@seb-0 thanks for the honest answer.

I think I've found part of the problem anyway - for most of the year when I backtest in 2021 the spread of USDJPY is 5. Bot seems to perofrm well when spread is lower. Not sure if its bad data in backtesting or if spread actualyl was that high for most of the year

[seb 0]

@FxWRLD The spread is expressed in points, 5 points=0.5pip. It doesn't seem that high. How much was the spread for 2022 ?
Can I see the training and testing results ?

[FxWRLD]

The spread seems to 0.2 for 2022. How best to share the results with you? Just screenshots or can I extract them from MT5 in some way?

[seb 0]

@FxWRLD
A screenshot would be enough.

[Alan.paredes2]

@FxWRLD no cometas el error de ponerte a ajustar los parámetros del bot , siempre vas a encotrar un conjunto de parámetros en el que el bot es rentable y eso suele confundir a muchas personas , puedes llegar a pensar que elbot es realmente rentable en mi experiencia los bot que son rentables sin importar la configuración suelen ser bot con un control en la cantidad de ordenes que abren, un control en la cantidad de lotes que tienen abiertos a la ves y con algún mecanismo para cerrar todas la órdenes a la vez cuando estas estén en ganancia o cuando estén cerca de una perdida razonable y tengan mucho tiempo de estar abiertas

[FxWRLD]

I've finetuned entries and exits a bit but here the latest.

So for 2022 until March 2023:

And from 2021 until March 2023:

So you can see that in 2021 it goes heavily into drawdown in comparison to 2022

[FxWRLD]

@Alan-paredes2 thank you thats very good insights. I do limit the orders my bot can place and have systematic exits in place. Although since finetuning and getting better results I've removed a previous exit condition but I get what you're saying.

[roar]

All I can say that your EA seems to work during higher daily volatility. But volatility cannot be really forecasted, so I cant give any answers there.

[FxWRLD]

@roar thanks that makes sense. So possibly use it on strongly trending pairs or have some sort of volatility condition on my entries. What would be the best indicator for that ATR?

[seb 0]

@FxWRLD
You can try adding an ATR filter but as roar said volatility can hardly be foretasted, a high volatility on a recent past doesn't mean a high volatility on a close future.

[FxWRLD]

Yeah I get you. Thanks for all the input and insights its been a great help