Free Expert Advisors circulate everywhere: MQL5 forums, Telegram groups, YouTube channels, marketplaces. Some are abandoned commercial products, some are serious amateur work, and some are simply bait.
Being free is not a flaw in itself. It merely removes the two usual safeguards — the seller's commercial reputation and the support that comes with a purchase — and shifts the entire verification burden onto you. Here is the checklist.
Start with the business model
Nobody gives away a profitable strategy without something in return. Identifying what that something is comes first, because it determines the level of risk.
Broker referrals. The robot is free provided you open an account through an affiliate link. The distributor earns a rebate on every lot traded. The direct consequence: their income depends on your volume, not on your performance. Robots distributed under this model almost systematically overtrade.
The product teaser. The free version is restricted: a single instrument, a fixed lot size, features switched off. This is an honest model when it is disclosed. What to check: what exactly is restricted, and whether that restriction distorts the results you will observe.
A demonstration of skill. A developer releases a simple robot to establish credibility. This is often the best category: the code is generally clean and the strategy is openly presented as basic.
Data collection. The robot is free because your email address and your account credentials are the product. Any request for your master password falls into this category.
No identifiable return. This is the most concerning case. If you can see no affiliation, no restriction and no identifiable author, the return still exists — you simply have not found it.
Security: the non-negotiable checks
An Expert Advisor is a program running on the terminal connected to your trading account. Three absolute rules:
Never hand over your master password. No EA needs it to operate. An investor password is enough for read-only access. A request for the master password is an attempt to take control of the account.
Be wary of an .ex4/.ex5 file with no source. These files are compiled, and therefore unreadable. You are running code whose contents you do not know. That is acceptable from an identifiable, accountable publisher; it is not acceptable from a pseudonym on a messaging channel.
Always test on a demo account first. Without exception, including for a robot recommended by a source you trust.
Use a separate terminal. A dedicated MetaTrader installation for testing, distinct from the one managing your capital, contains any incident.
Six things to check in the settings
Once the robot is installed on a demo account, open the settings window before you launch anything.
1. The default lot size. Many free EAs ship with a fixed lot calibrated for a 10,000-unit account. Applied to an account of 500, the effective leverage becomes absurd. Check whether the size is fixed or calculated as a percentage of equity.
2. Whether there is a stop loss. Look for a parameter such as StopLoss and check its value. A value of 0 means no loss is capped.
3. Multiplication parameters. Any parameter named Multiplier, Martingale, LotFactor, Recovery, GridStep or MaxTrades points to a grid or martingale strategy. A multiplier above 1 means positions grow after a loss.
4. The maximum number of simultaneous positions. A high value on the same instrument, in the same direction, describes exposure that accumulates.
5. The maximum permitted drawdown. Its presence is a good sign: the developer has provided a stop mechanism. Its absence means nothing will halt the robot.
6. Time and session parameters. A robot designed for a specific session but left running continuously will operate in conditions it was never intended for.
Thirty days of testing, then a verdict in figures
A month on demo, with the amount you actually intend to commit and at the broker you will use, produces a statement you can work with. Export it and examine: maximum drawdown reached, profit factor, payoff, consistency of lot sizes, and any clusters of positions opened simultaneously.
This is the stage at which most free EAs disqualify themselves — not because they lose, but because the way they win reveals a hidden risk of ruin.
Rather than calculating these figures by hand, import the statement into Judgebot: real drawdown, profit factor, recovery factor and payoff are recalculated from your trades, martingale and grid behaviour is detected, and the robot receives an A/B/C/D grade along with its weak points.
👉 Analyse a statement for free
When a free EA is worth it
A free robot is reasonably usable when it meets all of these conditions: an identifiable and contactable author, a strategy described in plain terms, a stop loss that is present and configurable, position sizing indexed to equity, no multiplication parameter after a loss, and thirty days of demo results consistent with the description.
That describes a minority of what circulates. It is also a minority that genuinely exists.
In summary
Free costs nothing to acquire and can cost the entire account to run. Identify the business model, never hand over your master password, inspect the settings before the first launch, test for thirty days on demo, then judge on the figures in the statement rather than on the promise. A free EA that clears these steps is worth more than a paid EA that clears none of them.