Building a Stable EA Together (Looking for Backtest ForwardTest)

Hi traders,

I’ve been working on a couple of EA variations over the past weeks, focused on one main objective:
Consistency with controlled drawdown.
I’m currently testing two versions:

1. JG-3
2. JG-4

Both are still in the development phase, and instead of keeping everything closed, I’d like to open this up to the community for independent testing and feedback.

What I’m looking for.
If you have time, I’d really appreciate your help to:

1. Run backtests (any pair / timeframe you prefer)
2. Try forward testing on demo or small accounts
3. Share:
   1. Results (profit, DD, consistency)
   2. Screenshots or reports
   3. Any observations (good or bad)

Notes.

1. No aggressive settings are intended
2. The focus is on sustainability rather than high risk returns
3. Feel free to tweak settings if needed and share your version

My current direction.
I’m personally aiming for:

1. Stable monthly growth
2. Low to moderate drawdown
3. Long-term survivability

(Not interested in “blow fast, grow fast” systems)

Why I’m sharing this.
Simple:
I believe collective testing is stronger than isolated testing.
Different brokers, different conditions, different perspectives = better system.
If the results are promising, I’ll keep updating this thread with:

1. Performance progress
2. Adjustments
3. Live observations

Thanks in advance to anyone willing to contribute
Let’s see where this goes.
— abeprabowo

Hi everyone,
Really appreciate the interest so far — I noticed the EA has already been downloaded several times, and some of you have even reached out to me directly. Thank you for that
Just to clarify, this EA is not for sale at this stage.
The main purpose of sharing it here is to collect feedback, ideas, and real testing results from fellow traders in this community. I truly believe that building something robust requires collaboration and different perspectives.
Also, I forgot to mention an important detail earlier:
This EA currently performs best on XAUUSD (Gold) with M1 timeframe
If you’re testing it, I’d really appreciate if you can share:

1. Your broker & spread
2. VPS or local machine
3. Forward test results (even small ones)
4. Any unusual behavior or improvement ideas

My goal here is simple:
build something stable, consistent, and long-term viable — not just another over-optimized EA.
Looking forward to hearing your insights and working together with you all.

Cheers

Bróker: Roboforex demo
spread:2 pip

Roboforex demo
spread 20 point

what is the strategy?give some introduction please

Looks nice. But note that the profit starts after 325 trades. The test is running over 9 months. It could mean that we have to sit watching nothing for 6 months. Not necessary true, cause I cannot se which date the trade no# 325 is.

Just a note. Keep developing and testing. This EA have probabilities.

Appreciate your observation — that’s a very valid point.
Yes, the equity curve does show that profitability starts after a certain number of trades, and this is something I’m paying close attention to.
At the moment, I’m trying to understand:

1. Whether this is due to specific market phases (range vs trend)
2. Or if the entry logic still needs filtering to reduce early drawdown

That’s also why I mentioned the need for longer forward testing (around 3 months), to capture different conditions and avoid conclusions based on one phase only.
Your point about “waiting period” is important — ideally, the system should not rely on long inactive or drawdown phases before performing.
Definitely something to improve in the next iteration. Thanks for highlighting it

Thanks for your question.
The EA is based on a semi-martingale approach, but with controlled risk using a defined stop loss. The idea is not pure recovery, but to manage entries and exposure more carefully.
At this stage, the EA is still under development. The goal is to forward test it for at least 3 months to see how it behaves across:

1. Different market conditions
2. High-impact news events
3. Ongoing global uncertainty

The system runs continuously (24H), mainly to capture intraday opportunities, especially on XAUUSD M1.
Sharing this demo here is mainly to gather feedback and insights from other traders. There’s still a lot to improve, and I believe collaboration can help refine the logic further.
Also to be transparent — this is still considered high risk and not ready for real account deployment yet.
Appreciate your interest

Thank you for taking the time to test and share these results — really appreciate it
Very useful to see both scenarios with different spread conditions (2 pip vs 20 points). This helps a lot in understanding how sensitive the EA is to execution and cost.
From your results, it seems:

1. The system is able to recover and generate profit after certain phases
2. But the equity curve still shows dependency on specific market conditions

This is something I’m currently trying to improve, especially:

1. Reducing early drawdown phase
2. Making performance more consistent across different spread environments

Your test confirms that spread and execution play a significant role, especially on XAUUSD M1, which is quite sensitive.
If you have time, it would be very helpful to also observe:

1. Behavior during high-impact news
2. Differences between sessions (Asia vs London vs NY)

Thanks again for sharing — this kind of input is exactly what helps move the development forward

Another issue is: 325 trades without any profit, then the cost is 325 * Commission on a real account. So that is a loss.

Thanks again for your input — really appreciate the detailed observation.
You’re right, and this is something I’m currently trying to understand deeper. The behavior where the system needs a certain phase before becoming profitable is not ideal, and I agree it needs improvement.
At the moment, I’m analyzing whether this comes from:

1. Market condition dependency (range vs trend phases)
2. Or entry logic that still allows too many early positions

The goal is to make the EA more consistent from the beginning, not relying on recovery after a drawdown phase.
Also, your point about waiting time is very valid — in real trading, sitting through long inactive or negative periods is not practical.
I’ll continue refining this part and share updates as I progress.
Really appreciate your insights — this kind of feedback is very valuable

20 points in gold is 2 pips, am I wrong?
I tested both EAs with the same spread.

Good question—and this is actually where many EAs go wrong.
From my experience testing different systems, the real issue is not just how we convert points to pips, but how the EA interprets them internally across different brokers.
For gold (XAUUSD), even if 20 points = 2 pips on paper, inconsistencies can come from:
Different digit formats (2 vs 3 decimals)
Broker-specific contract specifications
Spread calculation during execution (especially in volatile sessions)
That’s why in my EAs, I always normalize everything based on Point and Digits, instead of assuming fixed pip values. It makes the strategy much more robust across brokers.
Curious—did you notice any difference in results between the two EAs beyond the spread?

I see a big difference in the achieved return and the relative draw down. The result of J4 is less than half of that of J3, and the relative drawdown is much higher.
I put J3 on a one-minute gold chart in the morning hours (GMT+1), but it hasn't traded yet today.

Thank you for your observation and for testing.
From my side, both JG-3 and JG-4 are running normally and have been opening trades as expected. I’ve attached examples where multiple positions were executed on XAUUSD M1 from both systems.
One possible reason for the difference could be:

1. Market conditions at the exact time (entry criteria not triggered)
2. Spread or broker execution differences
3. Slight variation in data feed between accounts

Since the EA is rule-based, it will only open trades when all conditions are met. Because of that, it’s normal that in some setups it may appear inactive for a period of time.
If possible, you may also want to check:

1. AutoTrading status
2. Experts / Journal tab for any messages

Really appreciate your testing — this kind of comparison helps a lot to understand how the EA behaves across different environments

Quick update on the ongoing development and testing:
Over the recent sessions, both JG-3 and JG-4 have been actively running on XAUUSD M1.
So far, both systems are showing stable behavior in terms of trade execution and are able to adapt across different market conditions. JG-4 in particular appears slightly more responsive during higher volatility phases.
One important observation:

1. The systems are capable of generating consistent small gains
2. However, performance is still sensitive during strong directional moves, which can lead to temporary drawdown

This confirms that the core logic is working, but risk control and entry timing still need further refinement.
Currently focusing on:

1. Improving entry filtering during high volatility
2. Reducing early exposure in trending conditions
3. Enhancing overall stability across different sessions

The testing will continue over the coming weeks to capture more market scenarios, including news events and varying volatility conditions.
As always, really appreciate the feedback and comparisons from different setups — it helps a lot in improving the system further

The EAs have expired. What should I do, will there be new ones?

Hi @pusztafie,
Yes, the JG-3 and JG-4 EAs were time-limited, so they’ve now expired as part of the normal cycle.
Going forward, I’m no longer focusing on distributing standalone EAs. The main reason is that fixed EAs tend to struggle with changing market conditions, especially recently.
Right now, the strategy is being run in a more adaptive way, where execution and risk are adjusted continuously rather than relying on static parameters. The performance has been significantly more stable compared to the old versions.
I’ll keep sharing updates here as things progress

Hi @Aleks007,
Thanks for sharing the backtest
The curve looks clean, but just a small note — results like this are often very sensitive to modeling quality, spread settings, and especially parameter optimization. It’s quite easy for a strategy to look stable in backtest but behave very differently in live conditions.
From experience, the biggest challenge with this kind of setup is how it handles changing market structure over time, not just how smooth the historical curve looks.
That’s actually one of the reasons I moved away from fixed EA logic and started focusing more on adaptive execution and real-time risk management.
Curious to see how it performs on forward test as well