I'm a retail investor — interested in the market, but never to the point of digging deep into each company's financials. A little over a year ago, I picked up a book on fundamental analysis, looking to take my investing a step further. Somewhere in the first chapters, there was a quote attributed to Warren Buffett:

"You have to understand accounting and you have to understand the nuances of accounting. It's the language of business... unless you are willing to put in the effort to learn accounting — how to read and interpret financial statements — you really shouldn't select stocks yourself."

That stuck with me. I started looking into how to actually read financial statements, and quickly realized the data for US-listed companies was all publicly available on the SEC's EDGAR database. But applying even basic fundamental analysis across several companies meant opening dozens of tabs, doing calculations by hand, and keeping track of everything in a spreadsheet. That wasn't going to work long-term.

I also noticed that the existing sites presenting company financials all looked identical — the same table structure regardless of the company, the same pre-selected line items. That felt wrong to me. Financial statements are supposed to be a window into a specific company's financial life. Flattening everything into a generic template loses exactly the granularity that makes them interesting.

As a software developer by trade, I decided to build the tool I couldn't find. I spent over a year engineering a custom engine to pull data directly from the XBRL documents companies file with the SEC — 15 years of filings, every line item preserved exactly as reported, in tables that reflect the actual structure of each business. No opaque aggregation.

Investing real money on the basis of these numbers means you need to be able to trust them. So every figure in every table links back to its source filing, and every ratio exposes its exact formula. I've put every effort into making the data as accurate as possible — but given the scale and the diversity of reporting styles across thousands of filings, errors cannot be entirely ruled out. At least here, they are detectable.

Before trusting a number, it helps to know where it comes from.

All the data comes directly from the filings companies submit to the SEC — parsed as-is, without going through a third-party vendor or any reprocessing layer. What you see is what was filed. The whole point of this project is making that readable across fifteen years.

The same financial concept can appear under different accounting labels across a company's history. Rather than surfacing these as separate fragments, the engine identifies them as the same concept and consolidates them into a single line — so the full history reads as one continuous series, not a patchwork of disconnected periods. Every company has its own labelling history — this work is done from scratch for each one.

Two columns are the only exception to this — they don't come from any filing because companies simply don't publish them. There is no standalone Q4 report: the fourth quarter only ever appears rolled into the full-year figure. Flow statements accumulate over time, so subtract the nine-month cumulative from the annual and what remains is Q4. LTM — Last Twelve Months — follows the same logic: the sum of the four most recently known quarters.

The promise was transparency. Here's what that looks like in practice.

Every value in every table links back to its source filing. Click a cell and the original SEC document opens.

Ratios work the same way. Click any ratio cell and a breakdown panel opens showing the exact formula used, with each component value pulled from the filing. If a number looks off, you can check the calculation yourself and decide whether to trust it.

Clicking a row traces it as a line on the chart at the top of the page. You can overlay multiple series to compare trends across line items — revenue against gross profit, operating income against net income, whatever makes sense for the analysis you're doing.

The full dataset is also exportable to Excel. If you want to build your own model or run calculations the tool doesn't offer, the raw numbers are yours to work with.

And if something still looks wrong — a figure that doesn't match the filing, a calculation that seems off — there's a bug report button in the interface. The data is correctable. Nothing here is set in stone.