A free, open-source CLI that reads a stock's SEC filings. pip install pennytune
Note: This is a personal, independent project, not affiliated with or endorsed by any company or regulator. PennyTune surfaces evidence for your own due diligence. It does not constitute financial or investment advice.
A penny stock can file an 8-K disclosing a continued-listing deficiency, and the next morning most retail screeners will still show it as a green "buy." The warning was right there in the filing. The screener never read it.
That gap is the whole reason this tool exists.
Most penny stock and micro-cap screeners sort by price, volume, and a handful of ratios. They don't open the filings. But for a small, thinly covered company, the things that actually matter, dilution, going-concern doubt, a delisting notice, a restatement, are disclosed in SEC filings long before they show up in the price.
So I built a CLI that does the reading. It's called PennyTune, it's open source on GitHub, and it's live on PyPI today under the MIT license.
pip install pennytune
In the rest of this post I'll walk through what it does, why I built it, what's actually under the hood, and how you can use it for your own due-diligence work in under a minute.
The Gap Most Screeners Miss
Retail tools for vetting small, cheap stocks fall into two buckets.
On one end you have the screener. You sort a few thousand tickers by market cap, price, and maybe a P/E or a short-interest column, and you get a tidy table. It feels like research. But a screener is reading numbers someone else already aggregated. It is not reading the company's own disclosures. A stock can look statistically fine in a screener while its latest 8-K announces it has 30 days to regain compliance or be delisted.
On the other end you have reading every filing by hand. This is where the real signal is. But pulling up EDGAR, finding the right 10-K, 10-Q, and 8-K, and working through dilution, financial distress, insider transactions, and going-concern language, for every name you're curious about, takes serious time. Most people don't do it, because it doesn't scale.
So the signal that matters most is the signal almost nobody looks at, because looking at it is tedious.
That's the gap PennyTune fills. It does the filing-reading for you and surfaces the forensic red flags in a penny stock's disclosures, so you can spend your time on judgment instead of document retrieval.
What It Does in 30 Seconds
You install it. You set a one-time SEC identity (more on that below). You point it at a ticker. You get a forensic breakdown in your terminal.
Here's a real example:
pennytune inspect TICKER
That command pulls the company's filings from SEC EDGAR and computes a set of forensic risk signals from them: financial-distress and accounting-quality scores, dilution and capital-raising activity, insider buying versus selling, 8-K material events, and delisting and trading-suspension risk. It then shows you a decomposed score, so you can see which factors are dragging it down and which aren't.
If you want to rank a set of names instead of inspecting one, there's a scan:
pennytune scan TICKER1 TICKER2 TICKER3
That ranks a curated set of tickers you choose by their filing-derived risk signals, so you can triage a watchlist instead of going one by one.
No infrastructure. No dashboard. No account with a data vendor. No API key. Just a CLI that reads filings and returns an answer.
What's Actually Under the Hood
The headline is "it reads filings." The details are where the work lives.
One data source, on purpose: SEC EDGAR. PennyTune pulls exclusively from the SEC's public EDGAR system, the same filings the company is legally required to submit. No paid data feeds, no scraped third-party aggregators, no opaque "alternative data." Everything it reports is computed from public filings, which means the inputs are auditable and free.
Forensic risk signals. Rather than a single black-box score, PennyTune decomposes a company into separate signals so you can see what's driving the picture: financial-distress and accounting-quality scoring drawn from established, published models; dilution and shelf/capital-raise activity; insider transaction patterns; 8-K material events; and delisting and trading-suspension risk. For event-driven red flags, it names the specific 8-K item behind the flag (for example, a 3.01 continued-listing deficiency), so an event flag isn't just an abstract number.
Evidence, not verdicts. This is the part I want to be precise about, because it's a deliberate design choice and it shapes everything. PennyTune does not tell you a stock is "clean" or "a landmine." It does not give buy or sell signals. It does not assess tradeability or fetch live prices. It surfaces the forensic signals that are in the filings and leaves the judgment to you. The output is framed as evidence for your own due diligence, full stop.
No API key, no account. A lot of finance tooling assumes you'll sign up somewhere and manage a secret key. PennyTune doesn't. The only identity it needs is the contact string the SEC itself asks programmatic EDGAR users to send (a name and email in the request header, so the SEC can reach you if your requests cause load). That's the SEC's own fair-access requirement, not an account with me. It's stored locally on your machine, redacted in config output, and never sent to the author or anyone else.
Cross-platform. It runs on Linux, macOS, and Windows, on current Python versions, as a pure-Python package.
The Engineering Behind It
This is the part I think matters most. For a tool that touches financial data, "trust me" isn't good enough.
365 automated tests. Every commit runs 365 automated tests covering the scoring logic, the filing-parsing, CLI behavior, error handling, and edge cases. They run fully offline against fixtures, so the suite doesn't depend on a live network.
12 OS/Python combinations. The CI matrix runs on Linux, macOS, and Windows across Python 3.11, 3.12, 3.13, and 3.14. That's 12 combinations on every push, with no platform-specific skips, so if something works on Linux but breaks on Windows, I find out before users do.
Established models, not invented math. The financial-distress and accounting-quality scoring is built on published academic models, not formulas I made up. The code is open source, so anyone can read exactly how each signal is computed and check it against the source models.
Honest about its limits. PennyTune doesn't claim to be a verdict machine. It doesn't fetch live prices, it doesn't assess whether a stock is tradeable, and it's explicit in its own output and documentation that it's a research tool, not advice. Being upfront about what it does not do is part of the design.
Why I Built This
I kept wanting a fast way to answer one question about a small, cheap stock: is it cheap for a reason that's sitting in its filings?
The honest answer to that question is almost always in the SEC filings, and almost never in a screener. So I'd end up on EDGAR, opening 10-Ks and 8-Ks, reading through dilution and going-concern language by hand. It worked, but it was slow, and it didn't scale past a name or two.
The tool started as a way to automate the filing-reading I was doing manually. Then I added the decomposed scoring so I could see what was driving a flag instead of just getting a number. Then I added the watchlist scan so I could triage more than one name. Then I added the test suite, because for anything that parses financial filings, a quiet bug is worse than a loud one.
At some point it was useful enough that I figured other people poking at micro-caps would want it too, so I open sourced it under MIT.
Getting Started
The full workflow takes about a minute.
Install:
pip install pennytune
Works on Linux, macOS, and Windows. Python 3.11 or newer.
Do the one-time setup (this sets the SEC-required contact identity and your preferences):
pennytune init
Inspect a single ticker:
pennytune inspect TICKER
Rank a watchlist:
pennytune scan TICKER1 TICKER2 TICKER3
See exactly what data sources it touches and the free-tier limits:
pennytune sources
Read the full disclaimer (also shown on first run):
pennytune disclaimer
What's Next
The launch version is feature-complete for the due-diligence workflow I had when I built it. One thing I'm considering for a future version: the filing references (the specific form, date, and accession) already exist internally where the data is fetched, so surfacing them next to each flag, so you could click straight through to the exact filing, is a natural next step. If you'd find that useful, that's good signal for me to prioritize it.
If you use it and find something missing, open an issue.
Try It
pip install pennytune
I built this because I wanted it. Putting it out there in case other people find it useful too.
PennyTune surfaces evidence for your own due diligence. It is not investment advice, and it doesn't tell you what to buy or sell. The judgment stays with you.