OBSCURA.TRADEManage supplier listing
To make alternative datasets available through a single interface. There are thousands of them - scattered across agencies, registries, vendors and archives, each with its own format, its own quirks and its own cost of entry. We are working through them one at a time, normalizing every one of them to the same schema, so that reaching the ten-thousandth dataset is no harder than reaching the first.
Datasets
1Historical subdomain count over time for every US company with an associated stock market ticker.
Updated DailyFrom Daily since 2010
About
Obscura - Alternative Data for Systematic Research and Model Training
Obscura operates a production alternative-data platform: 122 datasets, 1.13 billion rows, all normalized to a single schema. Coverage spans SEC EDGAR regulatory filings, equity order flow, congressional trading disclosures, sentiment, macroeconomic and government series, corporate intelligence, and web attention. One schema across the whole catalog means a consumer integrates once rather than per source.
Point-in-time correctness is the core asset. Every row on every dataset carries the calendar day the publisher actually made that record public, recorded day-of. Filtering on that single column is a no-look-ahead filter that means the same thing across all 122 datasets, so a research team does not rebuild point-in-time logic per vendor. Reporting periods, real-world event dates and revision vintages are held in separate fields, leaving disclosure lag measurable rather than collapsed. Where a publisher exposes no release history, we recover it from real-time archives and record the supporting evidence per dataset rather than inferring a schedule. This is the expensive part to build and the reason the data survives contact with a serious backtest.
Raw and parsed, both available. Every collector retains the original source payload alongside the structured table. For model training that means the publisher's own document — the bytes as sent — rather than a downstream interpretation of it, with every transformation between raw and parsed auditable.
Built to be evaluated quickly. Catalog metadata and per-dataset column schemas are public and unauthenticated, so utility can be assessed before any commercial conversation. Delivery is a typed REST API, Parquet bulk export, or a Model Context Protocol server for agent and LLM workflows.
Commercial posture. Subscription tiers for internal research use; redistribution and multi-seat rights are available under negotiated enterprise terms. We are open to scoping category coverage, history depth and licensing to a buyer's specific requirement.
Questions
What buyers ask about Obscura, answered from this page.
What does Obscura do?
Obscura - Alternative Data for Systematic Research and Model Training Obscura operates a production alternative-data platform: 122 datasets, 1.13 billion rows, all normalized to a single schema. Coverage spans SEC EDGAR regulatory filings, equity order flow, congressional trading disclosures, sentiment, macroeconomic and government series, corporate intelligence, and web attention. One schema across the whole catalog means a consumer integrates once rather than per source. Point-in-time correctness is the core asset. Every row on every dataset carries the calendar day the publisher actually made that record public, recorded day-of. Filtering on that single column is a no-look-ahead filter that means the same thing across all 122 datasets, so a research team does not rebuild point-in-time logic per vendor. Reporting periods, real-world event dates and revision vintages are held in separate fields, leaving disclosure lag measurable rather than collapsed. Where a publisher exposes no release history, we recover it from real-time archives and record the supporting evidence per dataset rather than inferring a schedule. This is the expensive part to build and the reason the data survives contact with a serious backtest. Raw and parsed, both available. Every collector retains the original source payload alongside the structured table. For model training that means the publisher's own document — the bytes as sent — rather than a downstream interpretation of it, with every transformation between raw and parsed auditable. Built to be evaluated quickly. Catalog metadata and per-dataset column schemas are public and unauthenticated, so utility can be assessed before any commercial conversation. Delivery is a typed REST API, Parquet bulk export, or a Model Context Protocol server for agent and LLM workflows. Commercial posture. Subscription tiers for internal research use; redistribution and multi-seat rights are available under negotiated enterprise terms. We are open to scoping category coverage, history depth and licensing to a buyer's specific requirement.
What data does Obscura offer?
Obscura offers Historical subdomain count over time for every US company with an associated stock market ticker..
How is Obscura's data delivered?
The data is updated daily, with Daily since 2010 of history.
How do I get a sample from Obscura?
A downloadable sample is available to Brickroad customers. Request a sample through Brickroad to start.
How is Obscura's data licensed and priced?
Obscura licenses under a Programmatic Data License. Terms and price bands are in the Buyer room.