Loading...
Loading...
Browse all reviews, rankings, guides, strategies, and trust documents.
The exact criteria, weights, data sources, and audit cadence we use to score every Expert Advisor on this site. Reproducible by design.
Does the product have a verified live (not demo) account? We weight Myfxbook-verified accounts highest, followed by MQL5 Signals, then self-reported screenshots.
Profit factor, total net profit, win rate, and Sharpe ratio — measured against MetaTrader-standard backtest data over the longest period publicly available.
Max drawdown, recovery factor, and longest losing streak. We score lower drawdown more favorably than higher win rate at the cost of catastrophic risk.
Is the developer publicly known? Do they list verifiable contact info, prior products, and a real address? Pseudonymous authors lose points.
How many activations are included, how long the license lasts, whether updates are free, and how renewals are priced.
Mean time to first reply, public Q&A activity, post-sale documentation depth, and responsiveness to bug reports.
Each dimension scores from 0–100. The composite score is the weighted average of all six. Products without verified live trading data are capped at 60 to prevent backtest-only EAs from outranking live-verified ones.
composite_score = sum(weight_i * score_i) for i in {live, profit, drawdown, transparency, license, support}
if !has_verified_live: composite_score = min(composite_score, 60)| Event | Cadence | Detail |
|---|---|---|
| Marketplace metadata refresh | Daily | Author, ratings, downloads, comments, screenshots scraped from MQL5. |
| Verified live performance refresh | Weekly | Myfxbook embeds and broker-API equity curves polled and stored. |
| Score recomputation | Weekly | Composite weighted score re-calculated for every product in the catalog. |
| Hub-page intro / FAQ regeneration | Monthly | Long-form intros and FAQs regenerated from the latest data. |
| Methodology audit | Quarterly | Founder reviews scoring weights and sources; changes are versioned in Git. |
Long-form review and guide content is drafted by GPT-class language models, but only after every numerical claim has been verified against our internal product database. Drafts pass automated quality gates (hallucination check, financial-advice phrasing check, internal-link validity, minimum word count) and a randomized weekly sample is reviewed by William Harris in person.
Each AI-generated section is tagged in the database with ai_model, ai_generated_at, and (after review) last_human_review. The full editorial workflow is documented at /about/team.