How we score Expert Advisors

Every EA Verdict uses the same eight criteria, the same weights and the same rules for what counts as proof. The method is published so you can see exactly how a score was built, and disagree with it where you have better evidence. Last updated 17 October 2026.

The eight criteria (100 points)

Each criterion is scored from 0 to 10 in steps of half a point, then multiplied by its weight and divided by 10. Evidence and risk carry the most weight because a buyer of an automated system most needs to know whether the record is real and what the worst day looks like.

1. 1. Verified evidence20 points

   Can anyone outside the vendor check the track record?

   • An independent live record (Myfxbook or broker-verified) of 12 months or more on one account
   • Whether the record belongs to this exact version of the product
   • Backtest settings that can be reproduced, and how far they match live results

2. 2. Risk and drawdown behaviour20 points

   What happens to the account on the worst day, not the average one?

   • Maximum drawdown and how long recovery took
   • Grid depth, martingale, averaging down, and whether a hard stop exists
   • Documented blow-ups and the settings and lot sizes behind them

3. 3. Profitability quality15 points

   Is the profit real, repeatable and earned for the risk taken?

   • Profit factor, expectancy and return against drawdown
   • How much of the profit comes from a few trades or months
   • Whether deposits and withdrawals distort the headline growth figure

4. 4. Sustainability15 points

   Will the edge survive a different market, or only the one it was built in?

   • Logic behind the edge, and how dependent it is on one market regime
   • Parameter robustness and the number of years tested
   • Update history and the vendor's staying power

5. 5. Transparency and trust10 points

   Does the vendor show their working and handle criticism in the open?

   • Disclosed logic, manual access before purchase, refund terms
   • Review integrity, disputes and how the vendor answers them
   • Conflicts of interest on both sides, including ours

6. 6. Practical fit10 points

   What does your account, broker and setup need to look like to run it?

   • Broker, spread, account type and leverage requirements
   • Minimum capital against the capital that survives its drawdown
   • VPS need and prop-firm compatibility

7. 7. Cost and value5 points

   What does it really cost, and is the evidence worth that price?

   • Purchase price and pricing mechanics
   • Running costs: VPS, deposit, broker conditions
   • Price against the strength of the proof offered

8. 8. Support and usability5 points

   Can an ordinary trader install it, test it first and get an answer?

   • Free demo or tester access, set files, manual
   • Vendor response time and quality
   • Install friction

Scoring anchors: what each score means

Every criterion has written anchors, so a score is a reading against a standard, not a feeling. A top score is reachable: a 10 on evidence needs a verified record of 12 months or more for the exact version on sale, on one account, with no material gap. Trust is scored on its own anchors and is not inherited from the evidence score, because a verified record cannot show whether the manual is available before purchase, how criticism is handled, or what incentives sit behind the sale.

A material gap in a record means one of the following: the record is not the version on sale; the vendor runs it with no independent confirmation; growth is concentrated in a few weeks or months; deposits or withdrawals distort the curve; the account, broker or drawdown basis is not stated.

Evidence grades: what counts as proof

Every source is graded before it is used. Grades C and D are reported but never scored as proof, and the evidence criterion cannot score above 4 out of 10 without at least one grade A source for performance.

A

Independent proofA primary document or a platform-verified record: platform documentation, a verified Myfxbook account, an MQL5 signal built from broker trade history, a regulator or ombudsman text, a contract. A record the vendor runs still counts, with the vendor's interest noted. Counts in full.

B

Checkable vendor materialPublished by the vendor but open to inspection: the product listing, set files, a manual, a signal anyone can open. Counts, with the vendor's interest noted.

C

Claim onlyA figure with no record behind it: a performance number quoted in a sales page, a screenshot, a review that links nothing. Reported, never scored as proof.

D

Conflicted or anonymousMaterial from a party that earns from the sale or the surrounding services, or from an account that cannot be identified. Reported with its conflict, never scored as proof.

Risk label

The risk label sits beside the score because a product can score well and still be dangerous. A grid or martingale design with no hard stop cannot be labelled better than High.

• Low. A loss of more than a small fraction of the account is unlikely without a clear warning sign.
• Medium. Drawdowns of a meaningful size are normal; a total loss would need unusual conditions or misuse.
• High. The design can take large losses in a single adverse move. A grid or martingale with no hard stop cannot be rated better than this.
• Extreme. Total-loss outcomes appear in verifiable records, not only in anonymous comments.

Score bands

• Strong (85+): Independent evidence, contained risk, honest vendor. Rare.
• Credible (70 to 84): Solid on most criteria with limited gaps that are disclosed.
• Mixed (55 to 69): Real strengths and real gaps. Demo first, small size after.
• Weak (40 to 54): More unknowns than answers. Demo only until the gaps close.
• Avoid for now (under 40): The evidence is too thin to risk money on. Demo only, and re-score when new proof appears.

Our rules

• Every figure is attributed to a source and the date it was opened. Prices, ratings and versions change; re-check before you buy.
• Allegations are reported as allegations, with the vendor's public reply. We do not call a product a scam without proof.
• We say what we could not verify, in a section of its own.
• Vendors cannot pay for a verdict, a score or a placement, and we take no commission from vendors for reviewing them. Where we sell a product that we review, the verdict says so in its first lines and scores it by the same method.
• ForexCopilotEA sells MetaTrader 5 Expert Advisors, including some that compete with those reviewed. We state this in every verdict, and our own EAs are open to the same scoring.
• Scores are recalculated when new evidence appears, in either direction, and the change is logged on the verdict.
• Vendors can send verified records or corrections through the contact page.

Change log

• 17 October 2026: written scoring anchors added for all eight criteria. Both published verdicts were re-checked against them and no score changed.

Verdicts are independent commentary, not financial advice. Trading carries a high risk of loss. Read our risk disclaimer.