How can I tell if a forex EA or robot is a scam?

Updated August 27, 2026

Ask a smaller question

“Is this a scam” is unanswerable from the outside and it puts you in the position of forming an opinion about a stranger. Swap it for something you can settle: which of these claims can I verify without trusting anybody, and do they survive?

That reframing does most of the work. A vendor whose claims are all unverifiable hasn’t necessarily done anything wrong, but you’ve got nothing to stand on either, and that’s the same place you’d be if it were fake.

Work through the following in order. Each one is cheaper than the one after it, and plenty of things stop at step two.

1. Where do the results live?

Screenshots are worth zero. Not “less”, zero. An MT4 statement is HTML, it opens in a text editor, and any number in it can be changed in about ten seconds. The account history panel is just a picture of a window.

What you want is a link to a third-party tracker that you open yourself, in your own browser, on your own device. Not an image of that page. The actual URL, which you can then poke at. If that doesn’t exist, you’re done, because everything past here needs a real track record to check.

2. Live or demo?

The single cheapest tell, and it catches a surprising amount.

Demo servers fill instantly, never requote and have no counterparty. A scalping EA working on a 0.3 pip edge looks fantastic on demo and different on a live account, and that isn’t dishonesty, it’s just what demo servers are. Trackers show account type as a field. Read it. If it says demo, the result is a statement about the platform, not the market. Watch for the sideways version too: a three week live account sitting next to a two year demo, with the demo doing the persuading.

3. Read the strategy off the trade list, not the description

This is where “smart recovery”, “adaptive grid”, “AI position management” and “hedge protection mode” go to be found out. The marketing name doesn’t matter. The behaviour does, and it’s visible in the trades.

Look for:

• Several open positions in the same direction on the same pair, added while price moved against the first one
• Lot sizes that step up in a sequence (0.01, 0.02, 0.04) rather than staying flat or scaling with equity
• Groups of trades closing at the exact same second, which is how a basket exits
• A win rate above about 90% with an average loss much larger than the average win

Any of those and you’re looking at a system that recovers losses by adding size. That isn’t automatically fraud and some people trade it knowingly, but it changes what the track record means: the closed trades are all winners because the losses are still open, and the curve is smooth right up until the sequence runs out of margin. Judge it on the worst open drawdown it ever carried, not on the win rate.

4. Refuse any number that arrives without its pair

Win rate on its own is meaningless, and so is monthly gain. Both are trivially inflated by taking bigger risk, and neither says what was risked.

Win rate goes with average win versus average loss. Gain goes with maximum drawdown. Profit factor goes with the number of trades. If a page gives you one half of each pair and not the other, that’s a choice somebody made.

“No losing months, ever” belongs in the same bucket. Over a couple of years it’s not a sign of quality, it’s a sign that losses are being carried rather than taken, or that the sample is short.

5. Look at the account underneath the percentages

Three things quietly change what a percentage means.

Deposit size. A 300% gain on a $200 account is a different claim from 300% on $50,000, mostly because $200 can be lost forty times over while you find the run that worked.

Leverage. 1:500 on a tiny balance lets a position size exist that no real allocation would.

History length. Three months is not evidence. The trade count matters as much as the calendar, too: 40 trades over a year tells you almost nothing.

The mechanics of reading all this off a published page, including the ways a real page can still mislead, are on the Myfxbook verification page.

6. Ask for the investor password

Every MT4 and MT5 account has a read-only investor password. Handing it out lets you log in and look at the account and do nothing else. No trading, no withdrawals, no risk to them.

So ask. The answer is informative either way. A vendor who gives it lets you see the account live, open positions and floating loss included, which is the thing published pages hide best. A vendor who refuses is telling you their evidence stops where their control stops. “Against our privacy policy” is not a real reason for a read-only credential on an account they’re already publishing.

7. Now look at the shop

By now you know whether there’s anything to buy. The sales page only tells you who you’re dealing with.

Countdown timers that reset when you reload. “Original price $2,997, today $97”, where the higher price never existed. “Only 20 licences left” that has said 20 for six months. Testimonials with stock photography, which a reverse image search settles in a minute. A product that exists only inside a private Telegram group. A vendor whose entire internet presence is their own sales page.

None of these prove anything about the software. All of them tell you how the seller expects the sale to go, which is quickly and without questions.

Two practical ones. Can you run it on a broker you chose, or does it insist on one specific unregulated firm (which usually means the vendor is paid on your volume). And does the refund policy exist, cover live accounts, and get honoured, which the reviews will tell you. Paying by transfer or crypto removes your ability to reverse anything.

What good actually looks like

Not “guaranteed”, not “verified” in a badge on their own site. It looks like: a live account you can open, a year or more of it, drawdown published next to gain, a plain description of the strategy including the word martingale if that’s what it is, settings you can change and that are documented, and a vendor who answers a specific question with a specific answer.

And the things nobody can tell you

Whether it’ll keep working. Whether the account you’re shown is the only one they ran. Whether the settings you get are the settings that produced the record. Whether the market it was built for still exists.

Those aren’t gaps in your research, they’re genuinely unknowable, and the response is sizing rather than more reading. If you do buy something, run it at the smallest position your broker allows for a month and compare it to the vendor’s record over those same weeks. That comparison is worth more than the entire history they showed you. There’s a ten minute version of the track record check on the fast audit page.