Trading robot red flags: 12 claims in the ad that should make you pause

In short: A robot ad deserves a pause when it asks you to believe what you cannot check: promised returns, "no losses", screenshots, a backtest or a countdown.

A trading robot's ad deserves a pause when it asks you to trust something you cannot check: a promised return, a claim of no losses, a screenshot, a backtest or a deadline. Below are twelve such claims, how each sounds, why it should slow you down, and how to test it.

This page covers marketing claims. Questions to ask before buying: How to evaluate a gold trading robot. Ways to verify a record: How to verify an EA's track record. Rules differ by country; this is orientation, not legal advice. Midas is itself a martingale-style grid for gold, and with too small a deposit a stop-out is possible, so we apply the list to ourselves below.

Why do robot ads get away with exaggeration?

Robot ads get away with exaggeration because three things make it cheap. You cannot see inside the product before paying. A screenshot costs nothing to make or to crop. And the bill for a risky design comes after the sale: a strategy can look calm for months before the one move that hurts.

Regulators say much the same. The CFTC advisory "AI Won't Turn Trading Bots into Money Machines" (announced 25 January 2024, release 8854-24) names high or assured returns as red flags of fraud and lists 100% "win" rates among the claims scammers make (CFTC, retrieved 8 October 2026). In a case announced on 25 August 2021, a federal court entered a default judgment over a fully automated forex trading system, finding that its performance history and expected profits were misrepresented and that customers were not told its creator had never opened a live account with it (CFTC release 8417-21).

None of this means every robot is a fraud. It means an ad is not evidence.

Which 12 claims should make you pause?

Twelve claims should make you pause: promised or fixed returns, "no losing trades", screenshots instead of statements, backtest-only proof, demo results shown as live, one hand-picked period, a smooth curve then a cliff, buzzwords with no mechanism, countdowns and pressure, results behind a paywall, an anonymous seller with no terms, and "works on any pair, any broker, any account".

1. Promised or fixed returns

• Sounds like: a monthly or yearly percentage stated like a price list.
• Why pause: nobody controls the market, so nobody can set your return. The joint SEC, NASAA and FINRA alert (25 January 2024) calls claims of high promised returns with little or no risk a classic warning sign of fraud (FINRA).
• Check: does the seller's own risk page admit the deposit can fall?

2. "No losing trades"

• Sounds like: a 100% win rate; "never closes at a loss".
• Why pause: a win rate counts trades, not money. A design that holds its losers can win nearly every trade and lose the account on the rare one. The CFTC lists 100% "win" rates among the claims scammers make.
• Check: ask for maximum drawdown on equity, open trades included.

3. Screenshots instead of statements

• Sounds like: a cropped profit figure, a photo of a screen, chat testimonials.
• Why pause: an image shows one moment, can be edited, and rarely shows deposits, withdrawals or open losses.
• Check: ask for a broker statement or read-only access (the verification guide linked above explains both).

4. Backtest-only proof

• Sounds like: a tester curve as the whole evidence.
• Why pause: a backtest replays past prices under assumptions. In MetaTrader 5 the 1 minute OHLC mode emulates only four prices of each minute bar, while "Every tick based on real ticks" uses ticks accumulated by brokers, so results can differ by mode (MetaQuotes). The CFTC adds that hypothetical results may not account for a trader's ability to absorb losses (CFTC trading-system advisory, page shows no date).
• Check: ask for the modelling mode, the spread and costs used, and any real-money period.

5. Demo results shown as live

• Sounds like: "live results" with no account type, or a demo balance in the picture.
• Why pause: the same CFTC advisory warns of false claims that results come from real trading when they are simulated. On a demo, spread and slippage are frequently better than on a live account, especially in fast markets, as our FAQ says.
• Check: ask for the account type and server name, and how long real money was used.

6. One hand-picked period

• Sounds like: "look at March!", or a chart that starts on a convenient date.
• Why pause: the CFTC warns that results may have been built by selecting the historical trades with the greatest returns.
• Check: ask for the record from day one, worst stretch included.

7. A smooth curve, then a cliff

• Sounds like: a near-straight equity line with tiny dips.
• Why pause: some designs collect small wins often and carry a rare large loss, so the smoothness comes from the design, not from skill. The arithmetic is in the next section.
• Check: does position size grow as price moves against the trade? Where would the largest loss stop?

8. Buzzwords with no mechanism

• Sounds like: "AI", "neural network", "institutional algorithm".
• Why pause: the CFTC states that AI technology cannot predict the future or sudden market changes. The FTC's September 2024 sweep against deceptive AI claims (its cases were about fake reviews, legal services and online-store schemes, not trading) said there is no AI exemption from the laws on the books (FTC).
• Check: ask for three sentences: when it opens, adds and closes, and how large its biggest position gets.

9. Countdowns and pressure

• Sounds like: "the price rises tonight", "only 3 licenses left".
• Why pause: the SEC, NASAA and FINRA alert lists high-pressure sales tactics by unregistered individuals and promises of quick profits among its warning signs. The Bank of Russia describes fake brokers who rush clients and say money already put in will be lost (Bank of Russia, 31 March 2022, in Russian).
• Check: wait 48 hours. A genuine offer will usually still be there.

10. Results behind a paywall

• Sounds like: "proof is in the private channel after you buy".
• Why pause: you pay before you can inspect anything. The CFTC trading-system advisory asks whether the promoter will provide independent verification of the claimed results.
• Check: ask what you can inspect before paying. The answer is evidence too.

11. Anonymous seller, no terms, no refund rules

• Sounds like: no company name, no sale terms, support that answers only before payment.
• Why pause: the CFTC bots advisory suggests researching the background of the company or trader and the age of the website's domain registration.
• Check: find the terms and refund rules first, then send support one question.

12. "Works on any pair, any broker, any account"

• Sounds like: one robot for every instrument and account type.
• Why pause: spreads, contract size, lot step and account type (hedging or netting) differ, so a grid sized for one instrument behaves differently on another.
• Check: get the symbols, account type and deposit per lot in writing.

Why is a smooth curve followed by a cliff the hidden-risk flag?

Flag 7 is the one a chart cannot show, so it needs arithmetic. A seawall that has stood through ten calm summers has told you nothing about the storm it was built for.

The example is a martingale grid with the settings the free grid calculator uses by default, which are the Midas defaults: cent account, start lot 0.01, each order double the last, 11 orders, step $1.95, leverage 1:2000, stop-out at 40%, gold at $4,170, balance $1,000. The calculator takes every order at its exact level and ignores swap, commission and slippage, so real figures can be worse. Calculated on 8 October 2026.

Three lines to take away:

• Order 11 is 1,024 times the lot of order 1, and the basket carries 2,047 times the dollars per $1 of price: $20.47 against $0.01.
• If price never reached the level of order 6, the worst floating loss in the record stays under $1.11. With the same settings, stop-out sits $65.6 below the first order, with about $17 of the $1,000 left.
• A smooth record shows how often price turned back early, not how large the loss is when it does not.

This is arithmetic of settings, not a forecast of likelihood. Derivation: Martingale grid on gold: the arithmetic of stop-out. Grid versus martingale versus averaging: this comparison.

What does a straight answer look like?

A straight seller says, unasked: what the strategy does in one sentence, that the deposit can be lost and how, what is verified and what is not, the supported instrument and account type, and the sale terms with refunds, and sets no deadline. The list applies to us too. Where Midas stands today:

• Midas is a martingale-style grid for gold (XAUUSD) on MetaTrader 5, and with too small a deposit a stop-out is possible (Risk, plainly).
• It runs on hedging cent accounts and trades XAUUSD only. The broker must accept clients from your country.
• Midas has traded on a live account only since August 2026. There is no independently verified public track record, and a public read-only monitoring page is not available yet. Treat any result figure about Midas, ours included, as unverified.
• Refund rules are published in the license terms: within 90 days if the key was never activated, within 30 days if Midas could not be started through our fault. Trading losses are not refunded.

The third point is a gap, and we would rather list it than fill it with screenshots.

Where do I report a suspected fraud?

Report to the financial regulator that covers the seller or the broker, and keep ads, receipts and chats. Starting points (retrieved 8 October 2026):

• United States: CFTC (cftc.gov/complaint), SEC (sec.gov/tcr), FINRA (finra.org/investors/need-help/file-a-complaint), FBI IC3 (ic3.gov).
• Russia: the Bank of Russia anonymous form (cbr.ru/anonyminform), hotline 8 800 300-30-00, or the police.
• Elsewhere: your national financial regulator, and your bank or card issuer if you paid through them.

Before paying, look the seller up: Investor.gov's "Check Out Your Investment Professional" tool (US), the Bank of Russia's directory of financial organisations (cbr.ru/finorg) and its list of companies with signs of illegal activity. A software seller may not be a regulated firm, and the Bank of Russia notes that a name missing from its list does not make a company legal, so an empty result proves nothing. Reporting does not mean the money comes back.

What can go wrong?

• The checklist is a filter, not a verdict. A robot can pass all twelve and still lose, and a dishonest seller can learn the right words.
• Legitimate does not mean unable to lose. A martingale grid, ours included, can hit a stop-out on a strong enough move against it if the deposit is too small.
• A cent account makes the numbers small, not the risk.
• Never give anyone your trading password.

This is arithmetic of settings and a list of questions, 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.

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

FAQ

Are gold trading bots and forex robots legit?

Some sellers are honest and some are not, and an ad cannot tell you which. Even an honest seller's robot can lose money, so legitimate means the seller does what it says, not that the robot will not lose. Test the claims against the twelve flags above and ask for a statement you can open yourself, not a screenshot.

Do trading bots and expert advisors really work?

An EA does what its rules say; the label cannot tell you whether those rules make money. Judge a robot by its worst case, how large its biggest position gets and a statement you can open yourself, not by its ad. No robot removes the risk of loss, and the CFTC states that AI technology cannot predict the future or sudden market changes.

How can I spot a fake or scam trading bot before I pay?

Run three tests before paying: does the ad promise a return or claim no losing trades, can you open a statement yourself, and is there a deadline? Then ask in writing for the symbols, the account type and the sale terms, including refunds. If nobody answers one simple question, or the answers keep changing, stop and do not pay.

Is it risky to download a free forex robot or EA from the internet?

It can be. A free robot may come from an anonymous seller with no terms, show results only as screenshots or behind a paywall, or claim to work on any pair, any broker and any account (flags 3, 10, 11 and 12 in this article). Never give anyone your trading password. Ask for a statement you can open yourself and test on demo. We sell a paid EA ourselves, so weigh this advice accordingly.

Do AI trading bots work, and what is an AI gold trading bot?

"AI" in an ad is a label, not a mechanism. Ask for three sentences: when the robot opens, adds and closes, and how large its biggest position gets. The CFTC states that AI technology cannot predict the future or sudden market changes, and the FTC said in September 2024 that there is no AI exemption from the laws on the books. Be wary of any AI bot that promises returns.

Is a forex EA that never loses too good to be true?

Treat it that way until it is shown otherwise. A very high win rate can come from keeping losing trades open and carrying a rare, very large loss, so the win rate alone says little. Ask for maximum drawdown on equity, open trades included, and for a record that covers the worst stretch, not only the good months.