Research standards
Methodology before conclusions.
RavenInvestor is designed to make every published claim traceable to a source, timestamp, calculation, or clearly labeled interpretation.
1. Start with the source
Every analysis should identify the asset, market, observation time, timeframe, and available data provider. When a source or model detail is missing, the article states that limitation instead of filling the gap with an assumption.
2. Separate four layers
3. Publish selectively
Routine market noise does not automatically become an indexed page. Publication is reserved for events with enough independent evidence to explain something useful. This reduces repetition and keeps the archive focused on meaningful case studies.
4. Treat models as models
A modeled estimate is not presented as an observed fact. Heatmaps, scores, support levels, and anomaly classifications depend on inputs and assumptions. Articles describe what a model shows, disclose available limitations, and avoid implying that a highlighted level must be reached.
5. Make historical comparisons auditable
A valid historical comparison must disclose its matching criteria, search period, sample size, forward horizons, and summary statistics. No historical result is treated as a forecast, and small samples are labeled as such.
6. Correct and update
Material updates should carry a modified timestamp. A future correction policy will preserve the original analytical record while clearly identifying what changed and why.