Why I'm Building Vintl

I started investing when I was 18. Nothing fancy — just a kid trying to figure out how markets worked. But as I went deeper into software and data engineering over the next five years, I kept coming back to one question: can you trade through math and data instead of gut feelings?

The answer is yes. But there's a catch.

Data is the edge

Every quant, every algo trader, every systematic fund — they all know the same thing: data is the most important part of finding a real edge. Not the model. Not the infrastructure. The data.

The problem? Financial data is a disaster.

Treasury yields come from Treasury.gov. Economic indicators come from FRED. Insider trades from SEC EDGAR. Options data from one vendor, commodities from another, government contracts from a completely different system. Each source has its own format, its own quirks, its own parsing headaches — if it even has an API at all.

I've spent countless hours writing scrapers, parsing HTML tables, normalizing date formats, handling edge cases. Every developer and trader who needs this data rebuilds the same brittle pipeline from scratch.

That's the problem I'm solving.

One API for all financial data

The vision behind Vintl is simple: every type of financial data, through one API, in one consistent format.

No more stitching together five different data sources with five different auth mechanisms and five different response formats. One key, one base URL, one JSON envelope. Treasury yields, economic indicators, insider trades — and soon, commodities, options, government contracts, and more.

Every endpoint follows the same patterns: cursor-based pagination, decimal strings (never floats — no precision drift), and a Stripe-inspired response envelope. Learn one endpoint, you know them all.

The killer feature: point-in-time queries

Here's something most people don't realize: economic data gets revised for years after it's first released.

GDP for Q3 2023 was first reported at $22,491B. Today, that same data point reads $22,840B. That's a $349 billion revision — and it happened silently, across multiple updates, over two years.

If your backtest uses today's number for a decision your model would have made in October 2023, your model is cheating. It's using information that didn't exist yet. This is called look-ahead bias, and it's one of the most common reasons backtests look great but live trading doesn't.

GET /v1/series/GDPC1/observations
→ value: "22,840.989"  (today's revised number)

GET /v1/series/GDPC1/observations?as_of=2023-10-26
→ value: "22,491.567"  (what markets actually saw)

One parameter. That's the difference between a backtest that lies and one that tells the truth.

No other API gives you this. FRED only returns the latest value. Academic databases like ALFRED exist but they're clunky and not built for developers who want to curl an endpoint and get JSON back.

The pipeline behind it

I built a data pipeline that continuously collects, parses, normalizes, and stores financial data from public sources. The database architecture is designed to preserve every version of every data point — so you can query any series as it looked on any historical date.

The pipeline is built to scale. Adding a new data source means writing a collector and a normalizer — the storage, versioning, API layer, and auth all come for free. That's why the roadmap can grow fast.

Then I thought: why keep this to myself?

The tedious work of scraping government websites, parsing SEC filings, normalizing inconsistent formats — I've already done it. Why not provide it as an API so other developers and traders don't have to rebuild the same pipeline?

That's Vintl.

What's live today

The first three modules are live:

• Treasury Yields — daily yield curves from 1990 to today. Par and TIPS curves, any maturity, any spread.
• Economic Indicators — 21 series (GDP, CPI, unemployment, and more) with full revision history and point-in-time queries.
• Insider Trading — SEC Form 4 filings with auto-detected buying clusters. Know when 3+ C-suite executives are buying the same stock.

What's coming

This is just the starting point. The pipeline is already collecting more data than what's exposed through the API today. On the roadmap:

• Commodities — oil, gold, natural gas, agricultural futures
• Options data — chains, greeks, historical surfaces
• Government contracts — federal spending, contract awards, vendor data
• Corporate filings — 10-K, 10-Q, 8-K parsed and structured
• Alternative data — as the pipeline grows, so does the API

The goal is simple: if it's financial data that's publicly available but painful to access, it belongs in Vintl.

Try it

There's a free tier. 1,000 requests per month, all endpoints except ?as_of. No credit card required.

If you're building anything that touches financial data — a trading bot, a research tool, a dashboard, a model — give it a try. One API key, one curl command, and you're pulling data in seconds.

Financial data infrastructure is broken. I'm building the fix.