CopyGov research guide
CopyGov methodology, sources, and coverage
How CopyGov reads public disclosures, models post-publication stock signals, handles missing data, and separates research from portfolio returns.
What the research contains
CopyGov studies disclosed stock-purchase signals, with links to the underlying public record. Its dated historical study covers 2014–2025, supplemented by House filing snapshots for 2025 onward. Senate coverage in this release remains historical.
The public profile pages show the existing five-profile Free sample. These profiles were chosen as an editorial sample independently of their returns. Their detail tables contain up to twenty recent available purchase signals; the aggregate statistics summarize the available history, not just those displayed rows.
How outcomes are modeled
The study evaluates stock purchases after their recorded publication dates, using the next trading-session entry in the underlying research. Stock and SPY outcomes share corresponding entry and end dates. Read the explicit dates on the interactive record for each observation.
The standard horizons are 63, 126, and 252 trading sessions. The current House signal calculation applies an illustrative 0.1% cost at entry and exit to both the stock and benchmark. Model assumptions do not reproduce an individual investor’s execution, taxes, or account expenses.
What is missing and what is excluded
A missing price is not a zero return. Incomplete horizons, unavailable entry prices, identity issues, and unpriced outcomes must remain distinguishable. Profile tables expose the number of priced observations, missing observations, and price coverage.
Direct-stock research excludes options, gifts, and exercises from the stock-buy sample. Comparing an option’s underlying stock does not reconstruct the option’s premium, leverage, expiration, or realized return. A disclosed purchase also does not prove that the position is still held.
How to interpret the statistics
Mean excess return is the average of stock return minus SPY return across priced signals. Median excess return is the middle observation. The win share is the fraction of priced signals that beat SPY over the selected horizon.
These are historical signal statistics, not portfolio returns or personalized recommendations. Repeated names and securities create correlated observations. Incomplete records and prices can create selection bias. Historical rankings are not proof of a persistent advantage.
Freshness and corrections
Each public profile displays the date of its bundled research snapshot. Interactive research may be newer. The crawlable profile snapshot is rebuilt with the site; it does not silently become current every time a visitor loads it.
The beta’s source checks currently depend on visits. Always-on collection, external notifications, and paid access are separate launch steps. If a record appears wrong, compare the linked original document and any amendments before using or sharing its outcome.
Sources and further reading
- House Clerk — original financial-disclosure records ↗
- Senate — original financial-disclosure search ↗