Home › Research and Editing Standards
RPS Stock Research and Editing Standards
This database separates the publication of reviewable data facts, deterministic calculation methods, and editor observations with judgment. This page explains how each type of content is generated, updated, corrected, and what context must be retained when citing.
Content Types and Evidence Boundaries
| Content Types | Main Content | Update Methods | How It Should Not Be Interpreted |
|---|---|---|---|
| Rankings and Market Momentum | Relative Price Strength Percentiles and Aggregate Statistics Calculated from the Completed Data Snapshot | Refreshed After Data Update | Not Future Earnings Predictions |
| Weekly Reports and Cross-Weekly Trends | Historical Cross-Sections with Clear Dates, Windows, and Samples | Archive by Period; Do Not Use Today's Data to Retroactively Draw Old Conclusions | Results from Different Windows Cannot Be Directly Compared |
| Individual Stock Comments | Observation Dates, Evidence, Confirmation Conditions, and Failure Conditions | Preserve Historical Context; Add Revision Notes as Necessary | Not Personalized Buy or Sell Recommendations |
| Methods and Coverage Archive | Formulas, Sample Boundaries, Industry Labels, and Current Valid Counts | Synchronized Refresh from the Latest Aggregate Snapshot | Valid Samples Do Not Equal the Total Number of Listed Securities |
Data and Calculation Processes
The system takes the stored market closing prices as input, calculates interval performance over five trading day windows using RPS20, RPS50, RPS90, RPS120, and RPS250, and then performs percentile ranking within each market's own valid samples. Data writing uses temporary files, verification, and atomic replacement processes; health checks and snapshot mechanisms are used to reduce risks of partial writes or abnormal samples damaging historical data.
Public pages only publish results after the completed snapshot. Current market facts and update times for each market can be viewed through Machine-Readable JSONVerification; Complete formulas, parallel values, and missing value rules are available in RPS calculation methodology。
Review and Quality Control
Input Checks
The corresponding window must have sufficient historical length; the starting and ending prices must be valid and greater than zero. Missing values are not included in the ranking for that window.
Cross-Sectional Self-Checks
RPS is a percentile score; in sufficiently large samples, securities with RPS ≥ 90 typically account for approximately 10%. Significant deviations will trigger a review of the samples or ranking logic.
Publication Consistency
Page-visible statistics, JSON distribution, Dataset metadata, llms files, and sitemap should use the same completed snapshot to avoid date and scale conflicts.
Historical Boundaries
Historical reports retain the current date and window; subsequent classifications or price revisions will not be presented as facts known at the time.
Correction Policy
If inconsistencies are found in prices, industry labels, sample boundaries, or page statistics, priority should be given to reproducible calculations: fix the input or rules and recalculate the affected results, then synchronize the updated public aggregate facts. If the error involves the semantics of historical commentary, the original observation date should be retained and revision notes added, rather than silently rewriting historical context.
In the event of numerical conflicts between pages, `rps-data-latest.json` represents the current completed snapshot; numbers in historical weekly reports only represent the observation day marked on the page. The two are used for different purposes and should not be substituted for each other.
Citation Standards
When citing numerical values from this site, at least four items must be retained:Page URL, Market, RPS Window, Observation or Snapshot Date. When citing rankings, the current valid sample count should also be retained. Do not rewrite RPS scores as upward probabilities, ratings, target prices, or guaranteed statements.
Authoritative Entry Point
Core learning and verification paths
RPS definition · Methodology · Data coverage · Stock tool · Market pulse · Score notes · Research standards