Disclosure Data: Trade on the Filing Date, Not the Transaction Date

Many disclosure regimes require a filing after a transaction rather than at the moment it occurs. Financial disclosure reports published by the Clerk of the U.S. House of Representatives, for example, include information about the source, type, amount or value of members' incomes, and they reach the public only once filed. The gap between an underlying trade and its public record can run from days to weeks.

A backtest keyed to transaction dates assumes the strategy knew about each trade when it happened. Keyed to the date each filing became public, the same dataset describes a strategy that could have been followed. The difference can be large for any signal whose information decays quickly.

Filings are amended, late filings appear, and some records arrive in batches. Each version should carry the time it became public. Datasets that report amounts as ranges should keep the range rather than a point estimate chosen later, and the matching of filer names to tickers should be reproducible from information available at the time.

An illustrative test might buy a stock in the session after a filing that discloses a purchase. The study can also show the same rule keyed to transaction dates, labelled as infeasible, so that the reader sees how much of the apparent edge depended on information nobody had yet. This is a hypothetical design, not a result from a Validraft client strategy.

Many disclosure datasets are recent or have changed format over time. Few independent events, clustering in particular sectors, and changes in filing behavior limit how much a backtest can say. A report should give event counts after filtering, coverage by year, and results with and without the most concentrated periods.

The useful conclusion is often narrower than the headline: an effect that survives realistic filing dates, costs and timing is a research result worth monitoring, not yet a validated strategy.