Short answer
A warning sign is not proof of a scam. Many honest vendors launch with a young signal, an introductory price tier or only a few reviews. A warning sign is a point you check at the source before buying: on the signal itself, on the tracker, in the marketplace rules. If several come together and none can be cleared, the answer is simple: do not buy.
The checklist below lists eight warning signs we keep finding in our EA reviews, and how to check each one yourself. The examples come from our reviews and describe evidence, not people. We do not accuse any vendor named here of fraud or deception. Each example describes a single piece of evidence on a given date that you can check yourself. As everywhere on AlgoVerdict, we do not show returns or drawdown figures.
The eight warning signs at a glance
| Warning sign | What you check | Where |
|---|---|---|
| Backtests only | Is there a live account with real money? | Product page, signal link |
| Myfxbook without track verification | “Track record” tick on the system page | Myfxbook profile |
| Price ladder with time pressure | Announced prices against existing evidence | Product page |
| “No stop-loss needed” | Does the EA set a stop on every position? | Demo in the strategy tester |
| Signal starts after the sales launch | Signal start date against publication date | Signal page, product page |
| Growth before the signal start | Monthly history before and after the start | Signal page |
| Deleted or replaced signals | Do all linked signals still work? | Links on the product page |
| Reviews in the first days | Dates of the reviews | Reviews tab |
1. Backtests only, no live account
A backtest shows how a strategy would have traded past prices, with settings the vendor has often tuned to exactly those prices. Spread, slippage and execution at the broker are missing. A smooth curve from the strategy tester is therefore not evidence but a sales picture.
How to check: look on the product page for a linked live account, ideally an MQL5 signal or a verified tracker. If there is none, treat the EA as unproven. In our grading, vendor backtests do not count at all; an EA without live evidence gets a low value in the verifiability sub-grade. Why backtest and live drift so far apart is covered in why EAs fail live and backtesting and forward testing.
2. Myfxbook without a verified track record
According to its own help page (read 19 Nov 2026), Myfxbook checks two things: whether the history matches the broker’s (track record) and whether whoever added the account also has trading rights (trading privileges). A screenshot with the word “verified” replaces neither.
How to check: open the profile yourself, not the screenshot. Look for both ticks, real or demo account, start date, deposits and withdrawals, and leverage. An example from our reviews: for Goldwave Ultra EA, a product image showed “Myfxbook verified”, but no Myfxbook account was linked (as of 17 Nov 2026). For Aurum AI, the developer points to a cent account on Myfxbook for the time before April 2026; a cent account would not be allowed as an MQL5 signal. We have not checked this account. What the ticks mean and what they do not is explained in our Myfxbook report.
3. Price ladders and time pressure
Many vendors in the MQL5 Market launch cheaply and announce price tiers, often with a note that only a few copies are left at the current price. That is allowed and not a scam in itself. But it creates pressure to buy exactly when the evidence is thinnest.
How to check: put the announced final price next to the evidence that exists today. One example: Prop Firm Gold EA advertises “LIMITED COPIES AT CURRENT PRICE!” and names a higher final price; Aikon MT5 lists price tiers up to a planned final price (both read 17 Nov 2026). A later purchase at a higher price with a longer track is often the better purchase. Where the vendor offers a rental, it is the cheaper test; more in how to judge an MQL5 Market EA.
4. “No stop-loss needed”
A vendor who promises that an EA needs no stop-loss because it “sits out” losses or closes them as a basket usually describes a grid or martingale. Such systems win often and small and lose rarely, but then a lot. Other marketing can hide the same design: Gold Matrix EA was described as a momentum strategy; the trade log of our test account shows positions added against the move, in other words a grid with martingale.
How to check: run the free demo in the strategy tester in visual mode and watch whether every position gets a stop-loss and whether the EA adds positions against the price. The product page of BTC Anchor, for example, describes it openly: it states that the EA works without a stop-loss per position and that a total loss is possible (read 17 Nov 2026). Why a grid still needs its own risk limit is covered in grid and martingale EAs.
5. The signal starts only after the sales launch
If the signal starts after the sales launch, the vendor sold the EA before any live evidence existed. That can be harmless, but it means the first buyers bought on the basis of a backtest.
How to check: compare the start date on the signal page with the publication date on the product page. For Pac Man BTC, the only public signal began ten days after the sales launch (as of 17 Nov 2026).
6. Growth from before the signal start
In our reviews we come across the opposite case more often, and it is harder to see. Under the MQL5 signal rules, the entire trading history of the account is requested when a signal is set up (section III.5, read 19 Nov 2026). An account that traded weeks earlier brings that growth along, even if it comes from a period nobody could follow live.
How to check: open the monthly history on the signal page and compare it with the date from which the account is listed as a signal. For Quantum Valhalla, all of the reported growth came before the signal start; for Kings Eye, almost all of it (both read 17 Nov 2026). For your judgement, count only the time since the signal start.
7. Vanished or replaced signals
A signal that disappears takes away the evidence you could have measured the EA against. Under the MQL5 rules, a provider can have a signal without subscribers removed immediately on request (section III.14), and the administration can remove any signal (section I.15, both read 19 Nov 2026). Why a signal is gone is then stated nowhere.
How to check: click every signal link on the product page. For Harmonizer EA, one of the two linked signals can no longer be found on MQL5; for Gold Scalper Pro, the only published live account is no longer listed. In neither case does any source give a reason (as of 17 Nov 2026). Regardless of that: if a new product advertises the signal of a predecessor, check whether this EA really trades there.
8. Reviews that cluster in the first days
In the MQL5 Market, only users who bought the product or downloaded a full free version may review it (Market rules, section III.4, read 19 Nov 2026). That keeps strangers out but says nothing about timing. Reviews from the first week describe the first week.
How to check: sort the reviews by date. For Hedgehog MT5, all reviews date from the first ten days after the sales launch, for Allylla Pullback Bot all from the first week, for Kyora MT5 all from the first two weeks (as of 17 Nov 2026). We do not accuse any of these vendors of fake reviews; we only say that early enthusiasm is not a long track. Buyer reviews do not count towards our grade.
How we check ourselves
Every EA on AlgoVerdict gets a grade from 0 to 10 by the same criteria and caps. For the live track, an account in our own EA park counts most, an account published by the vendor counts, backtests and screenshots do not. Young evidence leads to a grade we mark as “provisional”. The criteria are in our methodology, the calculation with raw data in the calculation page. Every EA we review is in the EA overview. Disclosure: we sell EAs in the MQL5 Market ourselves and so compete with the vendors named here. We do not grade our own products; the checks on this page apply to them just the same.
Checklist before you pay
- Is there a live account with real money, and can you open it yourself?
- Does a Myfxbook account have both ticks, and is it a real account?
- When did the signal start, when did sales start, and where does the growth come from?
- Do all linked signals still work?
- Does the EA set a stop-loss on every position in the demo?
- From which period do the reviews come?
- Is there a rental or a small test size before you pay the full price?
How to compare licences and sales channels is covered in how to buy a forex EA. This page is not investment advice; it shows which evidence you can check before a purchase.