Home / Blog / How to evaluate a gold trading robot: 10 questions to ask…

How to evaluate a gold trading robot: 10 questions to ask before you buy

In short: To choose a gold robot, get written answers to ten questions before paying: worst case, full-depth lots, any stop, live record, terms. A vague answer is a warning.

Evaluate a gold EA: 10 questions to ask before you buy a gold robot; a vague answer is a warning
10 questions to ask before you buy a gold robot; a vague answer is a warning

Ask ten questions and get the answers in writing: the worst case, the account it needs, its lots at full depth, any stop, who verifies the live record, how the tests were made, who is behind it, the terms, a demo route and the update history. A vague answer is also an answer.

The checklist fits any gold Expert Advisor (EA), ours included; a box near the end shows how Midas answers. Figures are arithmetic of settings, calculated on 8 October 2026 with the grid calculator formulas, or come from cited pages. They are not forecasts.

On this page (6)

What kind of robot is it?

Name the type first. A single-trade robot opens one order with a stop-loss, so each loss is capped. A grid or martingale robot adds orders when price moves against it, usually in growing lots, and closes the basket on a bounce. Nothing caps its loss except the balance, so the size of its deepest order is the number that matters. Questions 3 and 4 weigh most for grids.

How do you choose a gold trading robot? 10 questions to ask before you buy

Choose a gold robot by getting written answers to ten questions: the worst case, the account it needs, its lots at full depth, any stop, who verifies the live record, how the tests were made, who is behind it, the terms, a demo route and the update history. In our view four matter most: questions 1, 3, 4 and 5.

Each question has a clear answer and a vague one. Ask in writing, so the answer can be quoted.

1. What does it do, and what is the worst case?

Clear: entry, adding, exit and the worst case in plain sentences ("if gold keeps falling, the broker's stop-out closes the positions and nearly all of the deposit is lost"). Vague: "AI", "adapts to the market", no worst case. The CFTC states that AI "can't predict the future or sudden market changes" (CFTC, 25 January 2024).

2. What account does it need?

Clear: a written list: platform, hedging or netting, cent or standard, symbol, leverage. A netting account holds one position per symbol, a hedging account several (MQL5 reference), and a robot that holds opposite positions needs the second. Vague: "works with any broker". Check that the broker accepts clients from your country. See hedging vs netting and what a cent account is.

3. How big does it get at full depth?

Ask for two numbers: total lots when every order is open, and dollars per $1 of gold price at that size. A bridge shows its load limit on a sign; a grid has one too, rarely on the sales page.

Example: cent account (1 lot moves $1 per $1 of gold; multiply by 100 on a standard account), a buy grid whose first order opens at a gold price of $4,170 and price then falls, start lot 0.01, step $1.95, leverage 1:2000, stop-out at a 40% margin level, balance $1,000.

Multiplier, ordersLots at full depth$ per $1 of goldTimes first orderGrid depth, $
×2, 11 orders20.4720.472,04719.50
×2, 9 orders5.115.1151115.60
×1.5, 11 orders1.671.6716719.50
×1.0, 11 orders0.110.111119.50

Same 11 orders, same step: multiplier 1.0 to 2 turns 0.11 lot into 20.47, about 186 times as much; two fewer orders cut the doubling grid to a quarter. On $1,000, stop-out sits $65.6 below the first order for 11 orders and $208.5 for 9. Vague: "depends on the market", "risk-managed". Method: the arithmetic of stop-out.

4. Is there any stop?

Clear: each protection with what it does and does not do (loss limit, order cap, hedge), what happens after the last order, and that the broker's stop-out is separate and cannot be switched off by the robot. Vague: "smart risk management", or only a take-profit.

5. What live record exists, and who verifies it?

Clear: a statement you can open yourself, on the broker's server with a read-only investor password or on a third-party monitoring page, showing deposits, withdrawals and every trade. Vague: screenshots, a video of a balance, "verified" with no verifier. The CFTC's advisory describes a case in which demo accounts were used to fake customer balances. Method: how to verify an EA's track record.

6. How were the backtests made?

Clear: a tester report with period, symbol, tick mode, deposit, lot rule, spread, commission and delay settings, plus the .set file. MetaTrader 5's help describes "Every tick based on real ticks" as real tick data accumulated by brokers and "1 minute OHLC" as four emulated prices per bar (MetaTrader 5 help). With four prices per minute, a grid's path inside the minute is a guess. Vague: "tested on ten years" beside a curve with no settings. A test is also not a live result.

7. Who is behind it, and how do you reach them?

Clear: a named company or team, written support, and a plain statement of what is asked of you. A key tied to your account number is one thing; some sellers ask for your trading login and password, which the National Bank of Slovakia says leaves the account exposed to people outside your control (NBS warning on forex robots). Vague: anonymous sellers, chat handles, pressure to move money.

8. What are the price, refund and license terms?

Clear: price, period, number of accounts, what happens on expiry (are open positions closed or left?), and a written refund rule or an honest "no policy yet". Vague: countdown timers, bonuses for bringing friends (the AMF describes robot offers paid through recruiting commissions), terms only in chat.

9. Can you test on demo, and what is your own loss limit?

Clear: a demo route, and an admission that live spread and slippage are often worse. Then your part: write down the amount you can lose entirely before you fund anything. If Question 3 puts stop-out $65.6 away on $1,000, can you accept losing nearly all of the $1,000? Vague: "start bigger for better results".

10. Is there a change log, and how do updates arrive?

Clear: dated versions, what changed in each, how to update, a version number visible on the panel. A changed default changes Question 3: a ceiling raised from 9 to 11 orders turns 5.11 lots into 20.47. Vague: "constantly improved", silent updates.

How do you score the answers on one page?

Score each 0 (no answer), 1 (verbal or partial), 2 (written and checkable). Our rule of thumb, not a standard: a 0 on question 1, 3, 4 or 5 is a reason to wait.

#Ask forScore
1Worst case, in writing0–2
2Account matching the specification0–2
3Lots and $ per $1, recomputable0–2
4Every stop, with its limits0–2
5Third-party live record0–2
6Test conditions, .set file0–2
7Named team, written support0–2
8Written price, refund, license0–2
9Demo steps, your loss limit0–2
10Dated change log0–2

How does Midas answer these questions today?

Midas is our robot, so this is a seller's answer: check each line on its page. Status on 8 October 2026.

  • 1. An MT5 Expert Advisor for gold with a martingale-style grid, trailing exit and optional one-order lock. With too small a deposit a stop-out is possible. Size your deposit with a safety margin.
  • 2. MT5 only, XAUUSD only, hedging cent account; netting is not supported (Open a broker account).
  • 3. Defaults of 11 orders, ×2, start lot 0.01, step $1.95: 20.47 lots, $20.47 per $1 on a cent account. The Grid tab shows these and the room to stop-out (Panel settings).
  • 4. An optional lock keeps the floating loss of the locked part from growing while it is on. It does not prevent a stop-out (FAQ).
  • 5. On a live account since August 2026. No verified public track record yet and no public monitoring page yet; detailed statements come from support on request, which is the seller's own file, not independent proof (Monitoring & alerts).
  • 6. The site publishes no tester report with tick mode and costs.
  • 7. The Midas team; the site names no individuals. Support only through the Telegram bot @MidasGridBot. To issue the license key, the install guide has you send the MT5 account number, digits only (Install).
  • 8. $99 a month or $399 once, same features, one MT5 account per license. Refunds: within 90 days if the key was never activated, within 30 days if it could not be started through the seller's fault (license terms).
  • 9. The help advises running the first-run checklist on a demo account first (Install & start); the FAQ says live spread and slippage are often worse.
  • 10. New versions come from the cabinet in the bot; the homepage lists changes, including the default ceiling moving from 9 to 11 orders. No standalone changelog page yet.

Score it yourself with the table above.

Can a checklist show that a gold robot will make money, and what can still go wrong?

No. A clear answer is a claim, not proof, and a grid with perfect paperwork can still lose the whole deposit if gold travels far enough one way. ASIC (release 26-063MR, 2026) describes scammers who claim to use AI while promising fast returns, and advises checking that the provider holds an Australian Financial Services licence (ASIC 26-063MR). The AMF (2 March 2023) warns about robot subscriptions tied to recruiting commissions (AMF). Rules differ by country; check yours.

What stays open is the market. The checklist shows where the line is (Question 3), not whether price will cross it. After you buy, a broker can change leverage or the stop-out level, spreads can widen at news, and price can gap at the weekend open. Re-check your numbers, and use only money you can afford to lose (Risk, plainly). For ad claims see trading robot red flags; for terms, the glossary.

Next step: test a gold EA yourself before buying and read the gold trading bot guide.

FAQ

How do I choose a gold trading robot, and what should I check before buying?

Work through the ten questions and get answers in writing. In our view four matter most: the worst case (1), lots and dollars per $1 at full depth (3), any stop (4) and the live record with its verifier (5). Then test on demo and write down the loss you can accept. No robot removes the risk of loss.

What is the best gold EA for MT5, or the best forex robot?

There is no single best, and we recommend none. Choose by criteria: the worst case in writing, lots and dollars per $1 at full depth, any stop, a live statement you can open yourself, how the backtests were made, who is behind it, the terms and a change log. Labels such as best, AI or high win rate are not evidence. Test on demo and set a loss limit.

Is there a free gold EA for MT5 that works?

Midas, the EA we sell, is paid: $99 a month or $399 once, with the same features and one MT5 account per license, and the site lists no free version or trial. For any free EA, ask the same ten questions: who maintains it, its worst case at full depth, the account it needs, a record you can verify and its terms. Then test on demo.

Are trading bots profitable, and can you make money with a gold EA?

Nobody can honestly answer yes or no for robots in general: the rules decide, and no robot removes the risk of loss. Win rate counts how often trades close in plus, not how large the losses are, so a grid can close most baskets in plus and still lose the deposit in one deep basket. Ask for the largest loss and the lots and dollars per $1 at full depth first (Question 3).

Can I trust gold EA reviews?

Treat reviews as opinions, not records: you cannot check who wrote them or whether they were paid. The CFTC advises caution about AI hype from social media influencers, and the AMF describes robot offers paid through recruiting commissions. Prefer what you can check: a statement you open yourself and the lot arithmetic from Question 3. Broker links on this site are referral links.

Should I test a gold robot on a demo account first, and will demo results match live?

Yes, as a rehearsal, not as proof. A demo shows whether the robot installs, connects and opens orders on your account type. It does not show live spread, slippage or how a deep basket behaves in a fast move, which are often worse on a live account. Write down your own loss limit before funding anything.

This is arithmetic of settings and a summary of public regulator pages, not a forecast. Trading with leverage carries risk: if the deposit is too small, a stop-out is possible. Size your deposit with a safety margin. Nothing here is individual investment advice.

Midas · published 8 October 2026 · updated 8 October 2026 · figures calculated 8 October 2026.

Sources (retrieved 8 October 2026): CFTC, Customer Advisory: AI Won't Turn Trading Bots into Money Machines, announced in CFTC release 8854-24, 25 January 2024; ASIC, media release 26-063MR (2026); AMF, The AMF warns the public about fraudulent investment offers through trading robots, 2 March 2023; National Bank of Slovakia, NBS warning: FOREX robots; MetaTrader 5 Help, Strategy Testing; MQL5 Reference, Account Properties.