Are free EAs on MQL5 actually dangerous to use?

Most free EAs on MQL5 are not malicious — the primary risks are performance (most underperform or fail silently) and abandonment (no updates when market conditions change). True malicious EAs — code designed to steal credentials or trade against you — are rare but exist. The MQL5 marketplace does code review, but it is not comprehensive. The practical advice is to only run EAs on a dedicated MT5 account with limited funds and 2FA enabled. Never give an EA the ability to withdraw funds (disable wire transfer permissions in your broker account settings). This protects against worst-case scenarios while still allowing you to test performance.

What does "cherry-picked backtest period" mean and how do I spot it?

A cherry-picked backtest period shows only the historical window where the EA performed best, hiding periods where it failed. To spot it: (1) check if the backtest period starts after a major market regime change — many free EAs show backtests starting in 2020 because pre-2020 data would show worse results; (2) look for the "out-of-sample" section — a credible backtest optimises on one date range and then tests on a completely separate, unseen date range; (3) note the drawdown — very low drawdown backtests are a red flag because real XAUUSD trading always encounters periods of adverse conditions; (4) check if the same parameter values were used throughout or if they were re-optimised mid-period.

At what price point does a paid EA become worth it?

The price of a paid EA is not strongly correlated with its quality. A $30 EA can outperform a $500 EA and frequently does. What matters is the evidence package: verified live track record, out-of-sample backtest, active development, support. Evaluate the evidence, not the price tag. As a rough framework: under $50 — could be a quality EA or a reskin of a free EA with a price sticker; $50–$200 — mid-range where quality varies enormously; $200–$500 — should come with comprehensive documentation, verified live track record, and active support; over $500 — requires exceptional evidence to justify, most traders are better served by mid-range EAs with strong evidence packages.

How do I test any EA (free or paid) before putting real money on it?

The standard protocol: first, run the EA's official backtest in MT5 Strategy Tester using tick data (download from your broker) with realistic spread settings (15–20 pips for XAUUSD, not the default 0). Compare results to the developer's published backtest — significant differences indicate data quality issues. Second, run on a demo account for minimum 30 trading days — not calendar days, but days with active trading. Third, if results match expectations, start live at 0.01 lot (minimum position) for another 30 days before scaling to your intended lot size. This protocol catches both poor strategy performance and execution issues that demo cannot simulate.

Why do some developers offer free versions of paid EAs?

The best case: a genuine limited version that demonstrates the strategy works, with the full version offering additional features (more instruments, more modes, higher frequency). This is a legitimate business model — the free version is a functional product, not a loss leader. The worst case: a deliberately hobbled "free" version with parameters set to lose money, designed to make you frustrated enough to buy the paid version under the impression it must be better. How to tell the difference: if the free version has user-configurable parameters and a verifiable (even limited) track record, it is likely genuine. If the free version is a black box with fixed parameters and no track record, it is likely a marketing tool.

What are the most common red flags specific to paid EAs that buyers miss?

Five red flags that buyers frequently overlook in paid EAs: (1) No live track record, only backtest — any EA developer serious about their product should be running it live; (2) "100% win rate" or "no losing trades" claims — these indicate either fictional backtests or survivorship bias in reporting; (3) Strategy described only as "proprietary algorithm" with no plain-language explanation — if you can't understand at a high level what the strategy does, you can't diagnose when it stops working; (4) No refund policy on a software product — legitimate developers are confident enough in their product to offer 30-day money-back; (5) Required ongoing subscription to signals or updates — many EAs become dependency traps where the EA stops working if you cancel a monthly fee.

How many active, quality XAUUSD EAs are actually available on MQL5?

Out of approximately 18,000+ EAs listed on MQL5 Marketplace, the number with all of the following attributes is very small: verified live track record of 6+ months, positive expectancy visible in both backtest and live results, active development within the last 3 months, and reasonable documentation. Based on community analysis, this number is likely under 200 EAs — roughly 1% of the listed total. The vast majority are either abandoned, untested, or performing below expectation. This is not unique to MQL5 — the signal-to-noise ratio in automated trading software marketplaces is universally poor. The filtering criteria in our trustworthiness scorer above is the fastest way to cut through the noise.

EA Marketplace · Buyer's Guide

Free vs Paid XAUUSD Expert Advisors: What You're Actually Getting

Quick Answer

Of approximately 18,000+ EAs on MQL5 Marketplace, fewer than 200 have verified live track records, active development, and complete documentation — roughly 1% of all listings. Free EAs are predominantly educational examples, abandoned projects, or lead-magnets. Paid EAs range from $20 reskins to $500 professionally developed strategies. Price is not a reliable indicator of quality. The 8-point trustworthiness framework below works equally well for evaluating both free and paid EAs.

18,000+

EAs listed on MQL5 Marketplace

~85%

have no verified live track record

14 mo

Average time since last update for top-rated free EAs

<200

Meet all 4 quality criteria simultaneously

Anatomy of a Free EA vs a Paid EA

Free EA Anatomy

What you typically get — click each row for detail

Paid EA Anatomy

What you should get — click each row for detail

EA Trustworthiness Scorer

Score Any EA — Free or Paid

Check all criteria that the EA meets. Works for both free and paid EAs.

Verified live track record (Myfxbook/cTrader) — minimum 6 months

Out-of-sample validation shown in backtest (not just in-sample)

Strategy logic explained in plain language (not just "AI-powered")

Active development — update in last 90 days

Support channel available and responsive

Developer trades it live with own money, publicly verifiable

No hidden upsells, monthly fees, or "signal subscription" required

Independently tested or reviewed by third party

Score: 0/14

Avoid or Investigate Further

Too many evidence gaps. Do not trade real money until these criteria are addressed.

The Free EA Ecosystem: What MQL5 Free EAs Actually Are

The MQL5 Marketplace operates on a self-publishing model with a code review process that checks for basic functionality and adherence to MQL5 API standards — it does not evaluate trading logic, backtest validity, or strategy effectiveness. Any MQL5 developer can publish a free EA. This creates a marketplace where the volume of listings far exceeds the quality of any individual listing.

Free EAs on MQL5 fall into roughly four categories. First: educational examples — simple moving average crossovers, basic RSI strategies — published as code demonstrations, not as tradeable systems. Second: abandoned projects — EAs that may have worked during a specific market period, published by developers who lost interest or moved to paid products. Third: lead magnets — deliberately simplified versions of a paid EA, designed to get users interested in upgrading. Fourth: genuine free strategies — rare, usually published by developers who have a different monetisation model (consulting, courses, or simply altruism).

The 6 most important evaluation criteria are covered in our EA selection framework guide, which approaches the question from the angle of common failure modes and works backward to what separates working EAs from non-working ones.

The selection bias problem is critical: the best strategies are monetised. When a developer creates an EA that generates consistent real returns, the rational move is to sell it, run it as a signal service, or use it exclusively for personal trading. EAs that are genuinely profitable are almost never given away for free — they are too valuable. Free EAs are, by selection, the ones that are not valuable enough to monetise.

What Development Actually Costs — And Why This Explains Pricing

A professionally developed XAUUSD EA with a verified track record represents a substantial investment of time and capital. Strategy research typically involves 50–200 hours of backtesting, optimisation, and walk-forward validation before a strategy is considered viable. Live testing adds another 3–6 months of real-money trading to build a verifiable track record. Documentation, support infrastructure, and ongoing maintenance add further ongoing costs.

$20–$80

Budget paid EAs

Usually reskins of community strategies with minor parameter adjustments. May work, may not. Evidence package is typically weak.

$100–$300

Mid-range professional

The most common price range for genuinely developed EAs. Quality varies enormously — evidence package is the differentiator, not price.

$500+

Premium / Custom

Requires exceptional evidence justification. Some are genuinely premium products; others are overpriced at this level without supporting live results.

Red Flags Unique to Free EAs

For scalping bot pricing specifically, see our free vs paid scalping bots guide, which focuses on the cost-of-ownership calculation for high-frequency strategies where commission drag makes the free vs paid decision more nuanced.

⚠

Look-ahead bias in backtest

Free EA code often uses future data to confirm signals, producing impossibly smooth equity curves. In MT5, this appears as a backtest with very few losing trades and a near-linear equity curve. Any real strategy on XAUUSD will have visible drawdown periods.

⚠

No changelog or version history

A developer who has maintained an EA over time will have a changelog showing what was fixed and improved. No changelog = either brand new (insufficient track record) or the EA has never been updated (abandoned).

⚠

Fewer than 5 downloads in 6 months

Very low download numbers on a free EA often indicate the EA was tested by its own author and found to not work, so it was never promoted. Real community-tested EAs accumulate downloads quickly.

⚠

Default parameters that obviously over-trade

Some free EAs are configured with default parameters that generate hundreds of trades per day, making the commission cost structure invisible in the backtest. Check what the default lot size and frequency are before anything else.

How to Test Any EA Safely: The 60-Day Protocol

Demo testing protocol before going live is covered in detail in our 60-day demo guide. The short version: run on demo for 30 trading days with realistic settings, then start live at 0.01 lot for a further 30 days before committing meaningful capital.

The MT5 Strategy Tester is your first line of defence. Run every EA through it with: tick data quality (download real tick data from your broker, not built-in MT5 data), real spread settings (15–20 pips for XAUUSD ECN, not the default), and commission settings (your broker's actual commission per lot). A strategy that looks profitable with default settings but fails with realistic settings has a built-in assumption that actual trading conditions will not support.

The 60-day demo period is not just about performance — it is about learning the EA's behaviour. When does it trade? How does it react to news events? Does it hold positions over weekends? What is the maximum drawdown you observed? You need to understand the EA before trusting it with real money, and that understanding takes time.