Forex backtests: How to really analyze an MT5 trading robot (and avoid the pitfalls)?

The Forex backtest is often presented as the ultimate weapon for evaluating an MT5 Expert Advisor.
In reality, it's a powerful tool... But also terribly misleading if you don't know how to interpret it.

In this article, you will learn:

How to correctly analyze a backtest
Which indicators really count
How to spot an over-optimized robot
How to test an EA for prop firms (FTMO or other)
And why some robots explode in real life despite a "perfect" history

This article will give you a clear advantage over 95% of individual traders.

What makes a good Forex backtest?

A backtest is a simulation of the past. Sounds simple ... but the reality is more subtle.

A good backtest should be :

Realistic
Reproducible
Consistent with your money management
Robust under different market conditions
Insensitive to a specific pair or period
Compatible with prop firm rules

If your robot has a "perfect" backtest, but only on EURUSD 2022 ... You haven't tested anything!!!

Stop "perfect" backtests: Here's how to spot them

There are 5 signals of over-optimization:

1️⃣ Too regular an earnings curve

An equity curve that rises like a straight line → Generally distorted.

The market is never this clean.

2️⃣ An abnormal success rate (>75%)

Truly robust strategies run between 45% and 65%, while remaining profitable thanks to the ratio.

3️⃣ An irrationally low drawdown

Un DD très bas (<3%) = Optimisation excessive.

Markets breathe, retrace, sting.
A robot that doesn't breathe = unrealistic tuning

4️⃣ A profitable EA only on one pair

Symptom number 1 of an over-tuned robot.

A solid strategy wins :

On EURUSD
On GBPUSD
And on several horizons

5️⃣ Ultra-specific parameters (minute optimization)

Example:

EMA 47
Stop ATR 1.38
Trailing 0.92 ATR
Etc

It works on the past, not the future...

The 7 really important indicators

And no, it's neither gross profit nor the earnings curve...

1. Drawdown (DD)

This is the most important parameter, especially for prop firms.
DD max acceptable:

< 8% sur un compte classique
< 4% pour FTMO (sinon c’est mort)

2. Profit Factor

Concrete measure of robustness.

➡️ PF ≥ 1.4 = Good
➡️ PF ≥ 1.7 = Very good
➡️ PF ≥ 2.0 = Often a sign of over-optimization

3. Expectancy

The most ignored metric ... and yet crucial.

Formula:
(average gain × success rate) - (average loss × failure rate)

If positive → viable strategy!

4. Trade recurrence

An EA that takes :

3 trades per year = Useless
800 trades per month = Dangerous

Healthy zone: 5 to 40 trades/month per pair.

5. Multi-pair robustness

A robot must stand on at least :

EURUSD
GBPUSD

If it only works on USDJPY → Bad news .

6. Multi-year robustness

The test must cover :

Volatile periods
Flat periods
Directional markets
Chaotic markets

7. Sharpe / Sortino ratio

More advanced, but ideal for comparing two robots.

How do I correctly backtest an MT5 robot?

This is the pro-level protocol used by institutional traders.

Step 1 - Using the "real tick" mode

No simulated ticks
No noisy data
No extrapolation

Step 2 - Test on 3 horizons

1 year (FTMO stress test)
3 years (robustness)
10 years if available (structural strength)

Step 3 - Check risk consistency

Your EA must follow the MM exactly, including :

Risk % real
Correct position size
SL & TP aligned with ATR

Step 4 - Adding variable spread

Test only in low spread = rigged test.

Step 5 - Simulate correlation

Example:
EURUSD long + GBPUSD long = Double risk on USD.

A serious EA must avoid this ! Such is the case with Titan Breakout

How do you analyze an EA specifically for prop firms?

All impose the same rules:

1. Daily drawdown

The most difficult test.
An EA must show very low volatility.

2. Max DD overall

If your EA shows -12%, it's unusable for a prop firm.

3. The ratio of winning trades to losing trades

Prop firms prefer :

Few trades
A logic of its own
A stable strategy

An ultra-frequent scalper → Generally refused

4. Average trade duration

"Breakout H1", "trend following" or "ATR" strategies pass easily.
Scalpers news → Never.

Why do some robots fail despite excellent backtesting?

Common reasons :

Real spreads explode in volatility
Robot has no anti-correlation filter
MM is badly calibrated
The EQ curve was rigged by optimization
No intelligent SL management
No intrabar break-even
Insufficient real-world responsiveness

A clean bot on paper ... Can become a nightmare in real life

How can you ensure the reliability of your backtest before going live?

Strict FTMO checklist :

Backtest ≥ 12 months
DD < 4%
PF > 1.4
Expectancy > 0
Multi-pair OK
Multi-year trend respected
Ratio R:R ≥ 1.5
Break-even mastered
Trailing stop logic
Consistent EMA/ADX/RSI filters
Max spread limited
Active anti-correlation
Global mutex to prevent simultaneous inputs

An EA that passes this checklist has a 90% chance of surviving in prop firm. And that's clearly the case with Titan Breakout, when the user has set it up and tested it properly!

Advanced analysis of a Forex backtest: Technical elements that traders often forget

To go further in the analysis of an MT5 backtest, it is essential to integrate a more global vision of a trading robot's behavior in the face of different market structures. A truly robust Forex Expert Advisor needs to demonstrate its ability to adapt to a variety of environments: strong trends, extended ranges, extreme volatility during economic announcements, slack periods or less active Asian sessions.

Many traders focus on the final result of the backtest, when the key lies in the structural details. The analysis should include:

Robot reaction to changes in volatility (measured by ATR)
The consistency of dynamic stop loss
Signal quality (EMA50 / EMA200, ADX, RSI)
The robot's ability to avoid dangerous trades in times of widening spreads
Behavior in the event of slippage
Trailing stop stability over several years

A high-performance robot doesn't just have to have a good final payoff: above all, it must have consistent logic, realistic execution quality and the ability to manage risk professionally.

Multi-timeframe analysis and consistency with modern Forex strategies

A professional backtest should also validate the consistency of the strategy with a multi-timeframe analysis. Successful prop firm traders know that H1 signals must be confirmed by the D1 trend to limit bad positions.

For an MT5 EA, this means :

A directional filter based on the price's position relative to the EMA50 and the EMA200
Validation of the overall trend over several time units
Consistency with liquidity zones
Risk-reward control on all trades

Prop firms like FTMO impose rigorous discipline. An EA that ignores the upper trend or contradicts the overall market structure will not pass the challenge phases, even if it seems to perform well in the backtest.

Multi-pair robustness: an essential criterion for a reliable MT5 robot

A Forex trading robot that performs too well on a single pair (EURUSD, GBPUSD or XAUUSD) is often a sign of overfitting. To validate the soundness of an Expert Advisor, it is essential to observe its ability to adapt:

Major pairs (EURUSD, GBPUSD, USDJPY),
To minor pairs,
At different market periods (2020, 2021, 2022, 2023, 2024).

A multi-pair backtest shows whether the strategy is based on a true market principle, or simply on adjusting the parameters to a specific period.

Conclusion: A good backtest guarantees nothing... But a bad backtest guarantees failure.

A solid backtest is not a promise of success, but it is an indispensable filter.

Smart traders:

Don't look for a perfect robot
Looking for a coherent robot
Looking for a robust strategy
Looking for a logic that can be verified over several years

A realistic backtest + Strict money management + A disciplined robot like Titan Breakout:
➡️ The real recipe for success in prop firm!

Forex backtests: How do you really analyse an MT5 trading robot?