Editorial guidelines.

1. Sourcing

Every numerical claim in a Nepse Signal article traces back to a primary source. We use, in order of preference:

• Direct disclosures from NEPSE-listed companies (audited quarterlies, annual reports, NEPSE filings).
• NRB (Nepal Rastra Bank) monetary policy documents and directives.
• SEBON regulatory filings, circulars, and license registers.
• Our own local floorsheet database (synced daily from NEPSE's official source — see our Methodology).
• Reputable secondary sources (Merolagani, ShareSansar, Bizmandu, mainstream press) — used for context, never as the sole source for a quantitative claim.

If a number cannot be tied back to a primary source, it is either qualified (e.g. "approximate", "as reported by X") or dropped from the article.

2. Drafting

Each long-form research article is drafted by a single named analyst — never by an anonymous content mill. Authors are listed at the top and bottom of every article and link to a dedicated /authors/:slug profile page detailing their credentials and expertise.

We write in plain language. Jargon is defined the first time it appears in an article (or links into our Glossary). We try to avoid the false-certainty register common in financial commentary; uncertain claims are flagged as such.

3. Review & fact-check

Before publication, every article goes through two additional sets of eyes:

1. Fact-check review — a second analyst independently verifies every number, date, name, and quotation against the cited source. Discrepancies are reconciled before the next step.
2. Editorial review — the founder reviews structure, framing, and balance. We deliberately avoid promotional language about our own platform inside research articles.

4. Editorial independence

Sponsors and advertisers have no editorial influence. Marketplace placements are clearly labelled "Sponsored" (see our Marketplace); research articles never reference advertisers, advertisers never see drafts before publication, and we don't take payment for positive coverage of any stock, broker, or company.

Our editorial team holds NEPSE positions personally. Where an article discusses a stock an author owns, that ownership is disclosed inline (e.g. "the author holds a long position in X at time of writing").

5. AI in our workflow

We use AI as a research and drafting assistant — not as a replacement for human judgement. Specifically:

• AI may help summarise quarterly filings, extract numbers from unstructured documents, or suggest article structure.
• All claims, framings, and recommendations in published articles are reviewed and verified by a named human analyst.
• We do not publish AI-generated content end-to-end. The by-line on a Nepse Signal article represents a human who stands behind every word.

6. Corrections policy

Mistakes happen. When they do, we fix them transparently:

• Factual error — corrected inline at the earliest opportunity. AUpdated YYYY-MM-DDstamp appears next to the published date.
• Material change — when a fix changes the article's conclusion, we add a "Correction" note at the top of the article describing what changed and why.
• Retraction — if an article is fundamentally wrong, we replace its body with a retraction notice but leave the URL live (with a redirect back to the index) so external links don't 404.

Spotted something wrong? Email info@nepsesignal.com. We credit external reviewers in the change-log of the article we fix.

7. Update cadence

Long-form research is reviewed at least once a year. If a regulatory change, market structure shift, or new data invalidates a published claim, the article is updated sooner.

Sector deep-dives (banking, hydropower, microfinance) are re-checked after each NRB monetary policy cycle, since policy changes ripple through bank fundamentals in 1-2 quarters.

8. What we won't publish