Financials and the Factor Library are coming soon...
Let's talk about how 3spread does it differently than other providers.
Most financial data vendors ask you to trust them. We'd rather show our work.
Every quarter, thousands of companies file financial statements with the SEC. Turning those filings into clean, comparable data is where most pipelines quietly fail. A mis-bucketed line item here, a silently-overwritten restatement there, a "0" that should have been "we don't know."
Here's how we keep our financial statements honest, end to end:
1. Deterministic from the source: no guesswork in the numbers path
Our extraction engine reads the company's raw data from the SEC submission: the facts they tagged and the structure of their filing. Every number in our schemas traces back to a specific fact in a specific filing. Standardization is rule-driven and reproducible; run it twice, get the same answer twice. No model "interpreting" a balance sheet into existence, which has become commonplace in the industry.
2. Every statement is scored, not just stored
Each statement we produce is validated on four core dimensions before it ships, alongside dozens of other internal quality assessment features.
- Footing: do the line items add up? Accounting principles are the foundation of our engine.
- Identities: does Assets = Liabilities + Equity, does the cash flow statement actually roll cash forward, and so on.
- Standardization: how cleanly did the filer's lines map to our canonical schemas? Regulatory and accounting guidelines give us direction; experience manually spreading thousands of statements gives us the ability.
- Completeness: did we retain what they filed? We track leaf-value retention above 99.9%.
The result is a composite quality score attached to every single statement, alongside a detailed subaccounting of our validator's outputs. Consumers don't have to wonder whether a statement is reliable; the score is part of the data.
3. Missing means missing. Never zero.
If a company doesn't report an item, our schemas say so explicitly. A null with a reason flag beats a fabricated zero every time, especially when that number feeds a model, a ratio, a screen, or a backtest.
4. Point-in-time, always
Restatements never overwrite originals. Every figure is versioned by the filing it came from and the date it became public. If you're backtesting a strategy, you see exactly what the market could have known on that date, not a quietly revised history.
Quality in financial data isn't a cleaning step at the end. It's an architectural decision at the start.
The financial datasets and our factor library were our original set of tools when we were founded in 2024, and our team is incredibly excited to be bringing them to you!
Getting ready:
- Read the docs at 3spread.com/docs for the endpoint reference and example queries.
- Sign up for a free account and provision an API key at 3spread.com/auth/signup, so you're ready the day they land.
- Join the community on Discord for release news first, and to tell us what you want from the factor library.
Free for the Community, always.