hindcast

Guides

As-reported vs restated data — which one should you use?

As-reported data is the figure a company published at the time. Restated data is that figure as it stands after later amendments — corrections, reclassifications, changes in accounting standards, or a restructuring that redefined segments. Both are legitimate; they answer different questions.

Restated figures are the better choice for understanding a business over time. They apply consistent definitions across periods, which is what makes a five-year revenue trend comparable rather than an artifact of changing disclosure. Analysts building a fundamental view of a company almost always want them.

As-reported figures are the only correct choice for evaluating anything that happened. The market priced the original number; a strategy, a manager or a model that acted on that date acted on that version. Using a restatement published two years later to judge a decision made before it existed is not conservatism, it is time travel — and it systematically flatters the decision.

The practical failure is storage rather than analysis. A database that keeps a single value per period has already chosen for you, usually the restated one, and it has thrown away the information needed to recover the alternative. Once the original print is gone it cannot be reconstructed from the current view, which is why the choice has to be made at query time and not at ingestion.

What's covered

Questions

How common are restatements?

Common enough that any long-horizon study will contain some, and concentrated in exactly the situations that are analytically interesting. The right posture is to assume any given historical figure may have been revised and to check rather than to hope.

Can I get both from one dataset?

Only if it was built to keep both. Once ingestion collapses to a single current value, the original is gone — which is why the distinction has to exist in the storage model rather than being a query-time filter over a flattened table.

Sources

Join the waitlist for early access.

Join the waitlist. Vote the roadmap. First in, first served.

← Back to the data catalog