Request for Feedback on EA Performance (8 Years of Backtesting)

Hello, fellow traders,
I’ve recently developed an Expert Advisor (EA) using AI and have backtested it on historical data spanning from 2017 to 2025. The results have been interesting, but as I’ve been away from active forex trading for a few years, I’d love to get some honest feedback from the community regarding its performance.

The strategy employs trend-following logic with dynamic grid trading, and I’ve also incorporated a pullback mechanism into the decision-making process. It is based on structure breaks and momentum shifts.

Key questions:

1. Based on these backtest results, do you think the EA is viable for long-term trading?
2. How do you view the drawdowns, risk/reward ratios, and profitability?
3. Given that I haven’t been trading actively for the last few years, how do these results compare to current market conditions and other EAs you’ve tested?
4. Are there any adjustments you would recommend to improve risk management or increase the win rate?

I truly value the feedback from this community and would really appreciate your honest opinions on this EA’s performance over the 8 years of backtesting.
Thanks so much for your time and insights!

Here are the backtesting results from 2017 to 2025

Backtest Summary

1. History Quality: 99%
2. Bars: 52,180
3. Ticks: 2,111,356,17
4. Symbols: 1
5. Initial Deposit: 10,000.00
6. Total Net Profit: 14,899.62
7. Gross Profit: 24,713.59
8. Gross Loss: -9,813.97
9. Balance Drawdown Absolute: 2,71
10. Balance Drawdown Maximal: 1,286.46 (7.29%)
11. Balance Drawdown Relative: 7.33% (1,090.11)
12. Equity Drawdown Absolute: 88.05
13. Equity Drawdown Maximal: 2,534.47 (9.32%)
14. Equity Drawdown Relative: 9.32% (2,534.47)
15. Profit Factor: 2.52
16. Recovery Factor: 5.88
17. AHPR: 1.0040 (0.40%)
18. GHPR: 1.0039 (0.39%)
19. Expected Payoff: 63.67
20. Sharpe Ratio: 5.35
21. LR Correlation: 0.96
22. LR Standard Error: 1,251.33
23. Margin Level: 523.24%
24. Z-Score: -7.15 (99.74%)
25. OnTester Result: 0

Trade Statistics

1. Total Trades: 234
2. Total Deals: 405
3. Short Trades (won %): 130 (77.69%)
4. Long Trades (won %): 104 (62.50%)
5. Profit Trades (% of total): 166 (70.94%)
6. Loss Trades (% of total): 68 (29.06%)
7. Largest profit trade: 840.73
8. Largest loss trade: -369.68
9. Average profit trade: 148.88
10. Average loss trade: -144.32
11. Maximum consecutive wins ($): 25 (2,237.53)
12. Maximum consecutive losses ($): 6 (-1,286.46)
13. Max Consecutive Profit (count): 6 (1,286.46)
14. Max Consecutive Loss (count): -1,286.46 (6)
15. Average consecutive wins: 6
16. Average consecutive losses: 3

If you can provide the full backtest report (export it) then I could assess the data a bit more.
My strongest recommendation is to AVOID changing the core logic or trying to "improve" the win rate. You have a proven, 8-year model. Tinkering with it will invalidate the backtest and likely lead to curve-fitting. Instead, focus on verification and implementation strategy.

 

1. Forward Testing (Crucial): Before committing significant capital, run this EA on a live or demo account for at least 3-6 months. This is the single most important next step. You need to see if its performance in the current live market aligns with the backtest.
2. Monte Carlo Simulation: Because of the low trade count, this is a perfect candidate for a Monte Carlo analysis. This analysis shuffles the order of your 234 trades thousands of times to simulate different possible equity curves. It will give you a much better understanding of the worst-case drawdown scenarios and the probability of reaching certain profit targets.
3. Risk Sizing: The risk seems well-managed. The only "adjustment" should be your position sizing. You could run it at half the risk of the backtest to be even safer, especially during the initial forward-testing phase. Do not increase the risk until you have extensive live proof of performance.

Summary: Pros vs. Cons
Excellent Low Drawdown (<10%)
Very Good Profit Factor (2.52)
High Win Rate (~71%)
Proven over 8 years & multiple market conditions
Strong Recovery Factor (5.88) & Z-Score
Balance and Equity lines are very close - No risky open positions.

Very Low Trade Count (234 in 8 years) - Risk of statistical fluke.
Long Stagnation Periods - Requires extreme patience.
Modest Annual Return (~12% CAGR) - Not a "get rich quick" system.
History Quality is 99%, not 99.9% - Minor but potential for inaccuracy.


Final Verdict
You have what appears to be a solid, robust, and safe EA. Its main strength is capital preservation. Its main weakness is the low trade frequency, which raises questions about statistical significance and requires immense patience.

Action Plan:

1. Do not modify the EA's logic.
2. Run a forward test on a live/demo account for 3-6 months.
3. Perform a Monte Carlo analysis if you have the tools.
4. If forward testing confirms the backtest results, you can consider it a viable part of a diversified automated trading portfolio.

Based on your backtest results the EA is viable for long-term trading in the Strategy Tester. However, the test results do not allow any conclusions to be drawn about behaviour in a real market. All kind of things will happen. Other things, not related to your strategy.

The figures may look good at first glance, but they conceal the fact that your test period is too long for the number of trades. This is a common misconception. A backtest does not become more meaningful the longer it is. The number of trades/period is more important and in your case this number is very low: approx. 1 trade per week. In reality, this means that even smallest disturbances/market changes have a large (too large) influence on your final trading results.

A yearly profit rate of 15-20% is easily achievable using automation, as long as you manage to run the strategy successfully for long time periods. In the last 15 years I haven't seen any EA being able to run a few weeks without errors, not to speak of years. That's because code quality discussed/shared in public places of the internet is generally not sufficient for long-time operation in a live environment. Think about network issues, broker disconnections. Happens all the time in a real market. Also prohibits forward testing using demo accounts. If someone advises forward testing using a demo account you immediately know that person never really traded automated strategies.

At the current stage I wouldn't think about adjustments. The strategy setup seems to be profitable over the long run in the Tester. I would work on solutions for the addressed issues. Further I would try to get better statistics, especially regarding the recovery periods between drawdowns/new equity highs. For long-term tests this times are usually huge (as in your case). I would like to know exact numbers, so I can ask myself: Am I psychological capable to endure a recovery time of a few months?

A guess: 99% of traders are not.

Solution: shorter tests with increased number of trades. Causes a whole bunch of different issues but makes it better reproducable for a few weeks/months. Also makes you look for completely different types of setups.

Another misconception common among retail traders. If your statement was true, then professional traders would avoid such a scenario, right? However, reality shows the opposite. Prop-shops, hedge funds, independent high-profile traders all do exactly the opposite. The whole business idea of day-trading futures is short term trading. Why is that? Because you need many trades to ensure stability and statistical significance of your results. You can't achieve this with 1 trade per week.

The main enemy is most of the time lack of an edge. In other words: traders assume they have a strategy but in reality they are gambling. If you do not have an edge then no time horizon matters, in the long run you will always lose - for one reason or another. But if you really have a defined edge then those traders try to exploit the market to the maximum, knowing that nothing lasts forever. This prefers many shorter over fewer longer trades.

There is a good reason why many bucket shops prohibit scalping (extrem short-term). If your assumption was right, then all the bucket shops would happily accept retail traders that want to scalp. Again, the opposite is true.

My custom EA was completed recently, and this is its performance over a 6.5-year backtest. The spread was extremely high during the first 3 years, which prevented the robot from trading. To fix this, I increased the spread filter to 500. The strategy itself is not sensitive to spreads, latency, or slippage. There is absolutely no martingale, no grid, and no dangerous money management built into this EA—just a pure, straightforward logic with high SL and TP settings, plus a built-in lot size multiplier for controlled compounding, and that's it.
Nasdaq M5
aggressive mode.

normal mode. ( The system delivers institutional-grade performance in Normal Mode, featuring an incredible 7.09 Sharpe Ratio and a massive 17.33 Recovery Factor over a 6.5-year period )



aggressive mode.
Tested with 100% real ticks, 10 ms latency, and variable spread settings over a 1.5-year backtest.



Do you guys think it's stupid to list this EA on MQL5 for $1,000?

Go ahead for $1500, Mate.
Scam QQ(Quantum Queen) can list for S1,999