A few months ago I needed to look up a company's debt-to-equity ratio. Not a complicated request. I ended up on Bloomberg's pricing page and closed the tab.
$2,000 a month. For data I needed once.
I'm an engineer. So I did what engineers do when something is broken or expensive: I decided to build a replacement. Not the full Bloomberg Terminal — I'm not delusional — but the specific slice of it I actually needed. Fundamental analysis. Clean data. No paywall. No sales call required.
That became pomegra.io.
What I actually built
The core is a stock analysis tool that pulls fundamentals — P/E, EPS, revenue growth, debt ratios, FCF — and surfaces them without requiring a finance degree to interpret. The interface is opinionated: it shows you what matters, not everything at once.
Under the hood it's not magic. Public data sources, some normalization logic, a bit of AI to turn the raw numbers into readable analysis. The hard part wasn't the data — it was deciding what to leave out. Financial data has a way of expanding to fill whatever space you give it.
The market analysis feature (pomegra.io/feature/market-analysis-simple) is where I spent the most time. The goal was to make it feel like you have a smart friend who read the 10-K so you didn't have to.
The part nobody warns you about
When you build a finance tool, you suddenly become responsible for accuracy in a domain where being wrong has real consequences. I'm not providing investment advice — the legal disclaimer is there, I mean it — but the data still has to be correct.
This meant writing a lot of validation logic. Cross-checking numbers against multiple sources. Building in uncertainty signals so the UI doesn't pretend to be more confident than the underlying data actually is.
Engineers are used to shipping fast and iterating. Finance data doesn't let you do that cleanly. A cached bad P/E ratio is worse than no P/E ratio.
Why free
I made most of pomegra.io free because the alternative felt wrong. Financial data has been locked behind expensive terminals for decades, mostly because that's how it's always been — not because the data itself is hard to get. It's a distribution problem, not a scarcity problem.
The engineers and individual investors who need this data aren't going to expense a Bloomberg subscription. They're going to Google it and get a half-answer from a random finance blog with a 2019 dataset.
I'd rather give them something that actually works.
Still building
The portfolio analysis and rebalancing tools are in active development. AI agents for deeper market research are on the roadmap. It's a side project — I have a day job, I maintain ai-tldr.dev, I live my life — but I ship something every few weeks.
If you want to poke around, it's live. No sign-up required for most features.
Top comments (0)