Method before message.

BarnGlow works from the public record. We list the source family for every data point rather than hiding it behind a single “market data” credit.

• Annual and quarterly reports — including 10-K, 10-Q, 20-F and 6-K filings where applicable.
• Current reports and issuer announcements — material events, guidance releases and dated company statements.
• Historical price and volume series — daily observations used for context and anomaly flags.
• Public industry classification — the peer group is named, so the comparison can be checked.

We do not ingest paywalled research, broker estimates, private data, or information obtained from a confidential source.

The model does three narrow jobs: it finds changes, it labels unusual observations, and it drafts a short summary with a source trail. It is a reading aid, not an authority.

• Compares a new filing with the prior one to locate changed sections and restated figures.
• Measures price and volume observations against a stated baseline window.
• Assembles peer context from named, public classifications and reported figures.
• Drafts a dated note that links each material line back to a passage or series.

Every output shows its inputs. The model does not promise an accuracy rate, and it does not choose a conclusion for the reader.

These limits are product boundaries, not disclaimers added after the fact. An output that crosses one is a defect.

• It does not tell you what to buy or sell, or suggest a position size or timing.
• It does not issue a rating, a valuation level to act on, or a direction call.
• It does not produce personalized advice or account-specific recommendations.
• It does not present a sample as current, live, or historically profitable.
• It does not fill a gap with a confident guess. A missing source is stated as missing.

We sample outputs and compare each material claim with the passage or series it cites. A claim that cannot be matched to its source is removed or marked unresolved. Validation is not a promise that every note is correct.

• The source trail is read back against the output before a sample is published.
• Changed figures are checked against the prior filing, not only the new one.
• Every correction is dated and attached to the archived note.
• Errors can be reported to [email protected] or by using the contact form.

The public record is not a clean dataset. These are the cases we know are hard, and they are visible in the product rather than buried here.

BarnGlow publishes no return figures, no success metrics, and no comparison with an index. There is no performance history on this site to read, because the product does not issue trade instructions and has no strategy to score.

The sample reports show the form of an output. They are not a record of what the product knew, recommended, or achieved. Any number that looks like a return should be treated as a format demonstration and not reused.