Now in private beta
Point-in-time research data your team can trust.
Deltarecord delivers managed data reliability for investment firms. We run your feeds end to end in your own AWS account, catch broken vendor data before it reaches your models, and keep point-in-time research tables verified and ready every morning.
Onboarding a limited number of investment firms. See our privacy notice.
- Your AWS account
- Runs in your cloud; only operational metadata leaves
- Open Iceberg tables
- Stored in S3 with zero vendor lock-in
- Query anywhere
- Snowflake, Athena, DuckDB, or Postgres
- Point-in-time
- Bitemporal history with zero lookahead bias
The problem
Bad data rarely announces itself.
It surfaces months later—in an unexpected model drawdown, an anomalous fill, or a backtest that cannot be reproduced. Deltarecord verifies every delivery before publication so corrupt data never enters your research.
A vendor delivers late or misses a deadline
A cutoff alert with clear root cause, while your research tables build reliably from the last good version.
A vendor changes schema or format unannounced
The delivery is held before publication. Your team keeps querying the last good version while we resolve the parser.
A source restates historical data
Both versions preserved with exact source timestamps ('as known at T'), preventing lookahead bias in backtests.
Tickers, CUSIPs, or FIGIs change across corporate actions
A built-in security master maintains point-in-time entity mapping across tickers, codes, splits, and renames.
Deliveries contain duplicate or out-of-order records
Strict ODCS contract checks reject duplicates and enforce integrity before anything is committed.
How it works
A managed service, not another tool.
We operate your data reliability end to end. Your team focuses entirely on alpha and research.
Deploy in your VPC
We deploy the framework into your AWS account via Terraform. We configure source adapters for your regulatory feeds, vendor deliveries, and APIs, storing all raw deliveries in immutable S3 storage.
Verify & publish
Every morning before your markets open, we land, parse, and verify feeds against strict ODCS contracts. If a delivery fails, it is placed on hold; your analysts keep working from the last good version while we resolve the incident.
Extend with regression safety
When you need a new feed or a vendor alters their delivery, we build the adapter and run regression runs against your historical landed data before anything touches production.
Security
Designed to pass due diligence.
Engineered to satisfy institutional allocators, compliance officers, and security teams from day one.
Client AWS account
Runs entirely inside your own AWS infrastructure and VPC. Your underlying financial data never passes through multi-tenant SaaS servers.
Zero data exfiltration
Only allowlisted operational run records leave your account for monitoring. Data values, holdings, and proprietary research stay in your VPC.
MFA & standing roles
Access is limited to three audited IAM roles. Operators connect using personal MFA credentials under strict least-privilege policies.
Regression-tested releases
Every framework release and proposal is verified against your recent landed deliveries before deployment, preventing breaking regressions.
FAQ
Common questions.
Which data sources do you support?
Regulatory filings (such as SEC EDGAR), vendor file drops over SFTP, and REST or streaming vendor APIs. If your research models require a source we don't yet support, we build and maintain the source adapter for you as part of onboarding.
Where does our data live and who owns it?
In Amazon S3 inside your own AWS account, stored as open Apache Iceberg tables. You retain full cryptographic control, access policies, and ownership. Your data is never copied into our account, and all tables remain permanently readable without our software.
How does our team query the data?
Analysts and quants can query Iceberg tables directly where they sit in S3 using Snowflake, Amazon Athena, DuckDB, or Starburst. For teams that prefer relational SQL workflows, we also sync published research tables directly into Amazon RDS Postgres.
How do you prevent lookahead bias in backtesting?
Every table is bitemporal, recording both source time (when the fact was published by the vendor) and landed time. When running a historical simulation at time T, your query retrieves facts exactly as known at T, guaranteeing that restatements and late filings never leak into past backtest windows.
What happens when a vendor delivers bad data or changes schema?
The delivery is placed on hold before publication. It never reaches your research tables. Your team continues querying the last good version without disruption, an automated alert is triggered, and our operator resolves the issue.
How is Deltarecord priced?
We charge a predictable flat monthly fee based on your feed volume and destination requirements, with zero per-query fees or data markups. Full pricing schedules are shared during private beta qualification.
Private beta
Request early access.
We are onboarding a select cohort of quantitative hedge funds, asset managers, and family offices. Get institutional-grade data reliability without hiring an internal data engineering team.
- Priority onboarding and custom source adapter development
- Direct communication with the founding engineering team
- Influence over roadmap priorities and supported data sources