Editorial policy

Review workflow

Each channel review follows the same editorial workflow. Public Telegram messages are collected for a defined time window. The raw data is processed through structured extraction — classifying posts, identifying signals, result updates, analysis, and promotional content. An article draft is then assembled from the structured evidence. Every draft undergoes human editorial review before publication and may be edited, revised, or rejected at the editor's discretion.

Observed fact, editorial inference, and unverifiable claims

Reviews distinguish between three categories of information:

• Observed fact — directly visible in the reviewed messages. Examples: "42 signal-like posts were found in the review window", "the channel posts daily market analysis for EUR/USD and GBP/USD".
• Editorial inference — a reasonable conclusion drawn from the evidence, stated as such. Examples: "the posting pattern suggests the channel prioritizes analysis over structured trade setups", "promotional frequency appears higher than content frequency in the reviewed sample".
• Unverifiable claims — statements made by the channel that cannot be independently confirmed from the reviewed data. Examples: profit claims without visible outcomes, guaranteed-return promises, references to private VIP performance records.

Absence of evidence is not treated as proof of wrongdoing. When data is missing or incomplete, the review describes the gap factually rather than filling it with speculation.

When content is revised or rejected

A review may be revised when new evidence becomes available, when the editorial team identifies factual errors, or when a reviewer requests changes to wording, framing, or emphasis. Revision preserves the article structure and section layout while updating the content.

A review may be rejected entirely if the evidence base is too thin to support a meaningful assessment, if the channel data is unreliable, or if the review does not meet editorial quality standards.

Role of AI in the review process

AI assists with two stages of the review process: structured data extraction (classifying messages, identifying instruments, recognizing signal patterns) and article draft generation (assembling the structured evidence into readable prose). Deterministic fields — message counts, review period, channel metadata — are computed directly from the source data, not generated by the AI model.

The AI does not make the final editorial decision. Every article is reviewed by a human editor who can accept, modify, or reject the draft. The editorial verdict and rating are evidence-driven and validated before publication.