What is included in earnings call transcript database?

company and fiscal period, call and publication timestamps, source URL and provider, speaker names and roles, prepared remarks and Q&A sections, document hash, availability and rights status. Coverage is explicit by company and quarter. A missing transcript, a discovered source, a restricted document, and a completed normalized transcript are different states.

Can I use this for historical backtesting?

Yes, where coverage and serving rights permit it. Save the query, source, known-at cutoff, schema version, and coverage result with the experiment.

Transcripts dataset

Earnings call transcript database

Organize calls as an auditable document corpus rather than an unversioned collection of pages.

Start with the API Read the API docs

Primary source

Issuer investor-relations pages and permitted transcript sources

API route

/v3/stocks/{symbol}/transcripts

Coverage

Explicit by stream

Serving

Rights-aware

What this dataset means

An earnings-call transcript is a textual record of prepared remarks and questions and answers. A usable corpus needs the company, fiscal period, call time, source, publication time, speakers, section boundaries, and revision status—not only a block of scraped text.

The buyer’s objective is straightforward: Acquire a research-ready historical transcript dataset. DataCedar keeps the research contract visible so the output can be inspected before it is trusted.

Fields delivered

company and fiscal period

call and publication timestamps

source URL and provider

speaker names and roles

prepared remarks and Q&A sections

document hash

availability and rights status

The endpoint is /v3/stocks/{symbol}/transcripts. Responses retain deterministic pagination and the metadata needed to connect normalized records to their source run.

How to evaluate a provider

Ask for company and quarter coverage, source ownership, missing-document states, publication times, revisions, speakers, sections, and export rights.

Transcript text may be copyrighted by the issuer or publisher. Public serving requires an owned, licensed, or otherwise permitted source; metadata can remain available when text cannot. DataCedar deliberately exposes that boundary in the product rather than leaving customers to infer it from a missing endpoint or a legal footnote.

Failure modes to test before purchase

Coverage counts unverifiable.
Rights unspecified.
Latest copy replaces historical version.
Acquisition access is treated as proof of redistribution rights.
An empty array cannot distinguish no event from incomplete collection.

Source, freshness, and reproducibility

Source. Issuer investor-relations pages and permitted transcript sources. DataCedar stores source identifiers and retrieval runs before normalization, then serves its own stable downstream schema.

Freshness. Collectors discover new calls after events and preserve source observations. Corrections create new document versions with their own hashes.

Rights. DataCedar does not bypass ownership restrictions. Decodo may retrieve public pages reliably, but it does not change copyright or redistribution rights.

Questions, answered.

The research schema includes company and fiscal period, call and publication timestamps, source URL and provider, speaker names and roles, prepared remarks and Q&A sections, document hash, availability and rights status. Fields remain connected to source, retrieval, coverage, and rights metadata.