How to Find Financial Restatements in SEC Filings

Financial restatements can be useful research signals. They show that a previously reported figure changed after becoming public—sometimes because of an error, sometimes because of an amendment, and sometimes because the company changed how a fact was tagged or presented.

For researchers, the central question is not merely: What does the latest filing say?

It is: What was first reported, when did it become public, and what changed afterward?

Answering that question requires point-in-time data. A current-only financial database may replace an original value with a revision, obscuring how the historical record changed.

This guide explains how to find financial restatements in SEC filings, compare original and revised values, and avoid false positives.

What counts as a financial restatement?

In a broad research workflow, a restatement flag can mean a meaningful change to a previously reported financial fact for the same company, fiscal period, and accounting concept.

That is broader than searching only for filings explicitly labeled as restatements. A changed value may appear in:

• An amended filing such as a 10-K/A or 10-Q/A
• A later filing that presents comparative prior-period figures
• An XBRL fact using the same concept and period but a revised value
• A filing that corrects or updates an earlier disclosure

Not every change means fraud, distress, or even an accounting error. Units, decimals, contexts, fiscal periods, taxonomy choices, and presentation changes can all create apparent differences. Treat a restatement flag as a research lead, not a verdict.

The reliable method: preserve original and revised values

To detect revisions, retain filing-time observations rather than storing only the latest value.

For each fact, capture relevant fields such as:

• Company identifier and ticker
• SEC filing accession or source filing
• Filing or acceptance date
• Form type
• Fiscal period and fiscal year
• XBRL concept or tag
• Units
• Context dimensions
• Reported value
• Whether the filing is an amendment

Then group comparable observations by company, period, concept, units, and compatible context.

Within each group, sort observations by the date they became public. The earliest valid observation becomes the original value. The most recent comparable observation becomes the latest value.

A simple relative-change test is:

relative_change = abs(latest_value - original_value) / abs(original_value)

Tradevo Data labels a row as restated when the same-tag value changes by more than 0.5%, including amendments. This is its documented classification rule, not a determination that a change is legally or economically material.

The edge case is an original value of zero. Relative change is undefined there, so a detection system needs a separately documented rule rather than an arbitrary percentage.

Why the same-tag requirement matters

Comparing values across different XBRL concepts can produce misleading alerts.

For example, a company may move from a standard taxonomy concept to a custom extension or disclose a related but not identical measure. Even when labels look similar, the accounting meaning may differ.

A conservative detector can begin by comparing values reported under the same tag. This may not capture every economically equivalent revision, but it reduces semantic mismatches and makes each flag easier to audit.

A more advanced pipeline can map extensions to standard concepts, but that requires validation and should preserve the original tags for inspection.

A practical SEC EDGAR workflow

1. Identify the original filing

Start with the relevant 10-K or 10-Q in SEC EDGAR. Record the filing date, accession, fiscal period, units, context, and reported value.

2. Search for amendments

Check for 10-K/A or 10-Q/A filings. Amendments are an obvious place to look, but do not stop there. A later annual or quarterly filing may present a changed comparative prior-period value.

3. Match the accounting fact

Compare the same XBRL concept for the same fiscal period, units, and compatible context. Watch for consolidated versus segment facts and continuing operations versus total-company measures.

4. Compare chronologically

Sort matched observations by when they became public. Preserve the first-reported value even if a later filing changes it.

5. Apply a documented rule

Use a documented classification rule, such as Tradevo Data's greater-than-0.5% same-tag rule. Keep the underlying observations so researchers can investigate or override the classification.

6. Review the filing narrative

Read amendment explanations, footnotes, auditor language, accounting-policy disclosures, and internal-control sections. A numeric change identifies where to look; the filing may explain why it happened.

Why current-only databases can hide the signal

Suppose a company first reports an accounting value and later revises it. If a database keeps only the revised figure, a historical backtest may behave as though that figure was known on the original date.

That creates lookahead bias.

A point-in-time record can instead separate:

• first_filed: the date the value became public
• original_value: the first-reported, point-in-time-safe value
• latest_value: the current revision
• restated: whether the same-tag change exceeds 0.5%
• qa_status: a quality-control field

For an as_of query, Tradevo Data includes rows where first_filed <= as_of, with the selected date included. For a deeper treatment, see https://tradevodata.com/blog/lookahead-bias-fundamental-backtests?utm_content=blog-how-to-find-restatements-in-sec-filings and https://tradevodata.com/blog/point-in-time-fundamentals-data?utm_content=blog-how-to-find-restatements-in-sec-filings.

Using Tradevo Data to surface revisions

Tradevo Data provides point-in-time US equity fundamentals sourced from public-domain SEC EDGAR data. Its annual dataset contains 632,466 point-in-time rows across 5,168 US companies, covers 16 annual concepts, and labels 37,804 restatements. It also contains 1,089,635 quarterly rows across 5,382 companies and seven quarterly concepts.

The API exposes one fundamentals endpoint:

/v1/fundamentals?ticker&as_of[&concept][&period=annual|quarterly]

Annual is the default period. Server-side as_of filtering uses first_filed <= as_of, including the selected date.

Each row includes first_filed, original_value, latest_value, a restated flag, and qa_status. Researchers can use these fields to screen for changed facts before reviewing the underlying SEC filings.

Tradevo Data covers up to 12 fiscal years. It does not cover non-US or delisted companies and does not provide TTM figures. Quarterly Q4 is reported where tagged or derived and labeled where supported; Q4 EPS and share values are not derived. Parquet format is not included.

Options compared fairly

Competitor capabilities and pricing can change. Review each provider's current documentation, methodology, and pricing page rather than comparing labels alone.

When another option wins

Use SEC EDGAR directly when you need complete filing narratives, exhibits, custom tags, or provenance beyond normalized numeric rows.

A broader vendor may win when you require international issuers, delisted-company coverage, additional asset classes, or a larger integrated data catalog. Sharadar or Tiingo may be a better fit when their broader products, coverage, or existing integration matches your workflow. QuantConnect may win when research, data, backtesting, and execution need to live in one environment.

Tradevo Data is positioned as an honest budget tier of research-grade point-in-time fundamentals—not as the only credible or affordable option.

When to build it yourself

Build your own pipeline when restatement detection is core intellectual property, when you need custom materiality rules, or when vendor normalization would conceal details essential to your research.

Be prepared to handle:

• SEC submissions and accession histories
• Filing dates and acceptance timestamps
• XBRL contexts, units, dimensions, and custom extensions
• Fiscal-calendar irregularities
• Duplicate and conflicting facts
• Amendments and later comparative disclosures
• Taxonomy migrations
• Reproducible quality checks

The advantage is control. The trade-off is continuing maintenance and validation, not just the initial parser.

Verify the method before paying

The public proof pack contains 225 rows for five companies, covering their latest three fiscal years, with full methodology and no signup:

https://github.com/christianpichichero-max/pit-fundamentals

Inspect the chronology, source logic, original values, and revisions before deciding whether the normalization fits your research.

If it does, the card-backed 7-day free trial at https://tradevodata.com/?utm_content=blog-how-to-find-restatements-in-sec-filings covers 10 companies, a rolling 3-year history, and 100 requests per day without bulk. It renews at $29 per month unless canceled. Pro provides complete available history, 5,000 requests per day, and bulk annual or quarterly downloads through /v1/download and /v1/snapshot?as_of. Documentation is at https://tradevodata.com/docs?utm_content=blog-how-to-find-restatements-in-sec-filings.

Restatement flags are best used as triage: identify changed facts, inspect the filing sequence, read the disclosures, and decide whether the revision is relevant to your research.

Not investment advice.