Looking for honest feedback – small-account strategy on XAUUSD 5M (bot-based)

New comment
 

Hi everyone,

I’ve been working on a small-account trading strategy specifically for XAUUSD on the M5 timeframe.
It’s fully automated, but I’m not here to promote anything – just genuinely curious how experienced traders would approach this kind of model.

I’m testing the following principles:

  • Very tight SL and 1:10 RR

  • Entries only in volatile but structured moments

  • Focused on not blowing small accounts (starting from ~$30)

  • Scales well as balance grows

So far, backtests look promising, but I know how deceptive those can be, especially with gold.

I’d love to hear your take:

  • Is this kind of approach realistic long term?

  • What are the usual traps when designing small-account EAs?

  • Would you test something like this, and what would you look for?

Not selling anything, just looking to improve the idea and learn from people who’ve seen it all.

Thanks in advance!

 
Benhamza Oussama:
Is this kind of approach realistic long term?

To determine long term realism, do a long term backtest. My personal minimum amount of historic data for backtesting is 10 years. My best EA was tested for over 20 years. Another experienced trader in this forum believes that even 20 years is not enough.

Benhamza Oussama:
What are the usual traps when designing small-account EAs?

The usual risks for small account EA's are drawdown and consecutive losing trades. Tight stops can mitigate drawdown for a single trade, but a long string of consecutive losing trades can still blow the account.

Benhamza Oussama:
Would you test something like this, and what would you look for?

You're obviously trading Gold CFD's (contracts for difference). In CFD trading, there is no underlying centralized exchange and you're access to liquidity is limited to your specific broker-dealer's pool of participants. Therefore, you should test on Every tick based on real ticks (these ticks are provided by your specific broker-dealer).

If and when you successfully scale up, you might consider graduating from CFD's and moving on to actual Gold futures. In centralized Gold futures trading, data is provided straight from the exchange (CME) by licensed data providers. Everyone trading Gold futures gets the same trade execution priority and participates in the same liquidity pool. Unlike in CFD trading, Gold futures pricing is derived from the speculative future price of spot Gold. If you were to hold a Gold futures contract through expiration, you would likely have to take delivery of physical Gold in the State of Oklahoma in the U.S. Therefore, Gold futures are actually connected to physical Gold while Gold CFD's are not.

I say "graduate" because margin requirements are higher for Gold futures (the CME recently created Nano Gold futures having lower margin requirements, but the volume and O/I (open interest) is presently too low to support reliable trading). Also, there are structural and operational differences between CFD's and futures. Therefore, research of the futures market, and coding research, would be required in order to adapt a Gold CFD EA to Gold futures. It isn't easy.

 
Widen the Stop and go for 1:5 RR . Think about trading sessions London /New York and not running it 24/5 , Think about spreads for  Gold  and the volatile nature of it . Tight stops is the mistake of thinking you  are playing it safe when in reality it is death by a thousand cuts . 
 
Benhamza Oussama:

Hi everyone,

I’ve been working on a small-account trading strategy specifically for XAUUSD on the M5 timeframe.
It’s fully automated, but I’m not here to promote anything – just genuinely curious how experienced traders would approach this kind of model.

I’m testing the following principles:

  • Very tight SL and 1:10 RR

  • Entries only in volatile but structured moments

  • Focused on not blowing small accounts (starting from ~$30)

  • Scales well as balance grows

So far, backtests look promising, but I know how deceptive those can be, especially with gold.

I’d love to hear your take:

  • Is this kind of approach realistic long term?

  • What are the usual traps when designing small-account EAs?

  • Would you test something like this, and what would you look for?

Not selling anything, just looking to improve the idea and learn from people who’ve seen it all.

Thanks in advance!


It's impossible for anyone to give you an answer without more context. Instead, consider asking yourself a few questions:

Is your strategy based on machine learning, genetic algorithms, or is it fully manual? If it’s ML or genetic created, have you tested it on a large out of sample period to see how it handles unknown market conditions? Is your validation process solid?

If it's manual, how much historical data have you reviewed, especially during high-stress periods like the trade war, COVID, to understand how it holds up?

That’s just scratching the surface. There are plenty of deeper questions to explore (Monte Carlo simulations, etc.), but hopefully this gives you a starting point. Good luck, hope it passes all your stress tests.
 
1. Coding a strategy can be easy. It can be coded by you if you know programming or you can hire someone from freelance
2. Optimization : you can use built in optimization  mt5 tools to find best parameter for your strategy

My personal experience 
1. 1:10 Risk reward is possible using some strategies such as pending FVG / order block but I have not seen them working for long term
2. you need to confirm if your very tight sl really works. You need bid ask data and should test in last 3 years of chart to see if spread was working ok according to your very tight sl.

Traps : Traps can be only discovered by you during backtest and optimization phase. Choose a specific broker where you trade and run backtest using specific broker's bid ask chart only

Once you pass all checks and get good results you can start your EA with $30 and slowly increase amount.



New comment