A Bloomberg Terminal costs about $30,000 a year.
So you do what everyone does: You paste news headlines into ChatGPT.
Then a 10-K. Then a chart. Then another chat window.
By Friday you have 40 chats, 12 browser tabs, and no idea if any of what ChatGPT came up with works.
That's experimenting, which is fine, but the sooner you build a system, the sooner you get results.
Every morning brings new filings, prices and headlines for every stock you follow.
You can't read it all. Your AI can't either, unless you tell it what to read.
There's more to read than any person can handle, and most AI projects never become a system:
- The SEC’s EDGAR system processes about 4,700 filings per day
- Gartner predicts over 40% of agentic AI projects will be canceled by the end of 2027, because of escalating costs, unclear business value or inadequate risk controls
More AI won't fix this. A system that tells it what to read will.
The solution: The $0 research desk
Each step takes about a week but you end up with a power AI workflow.
Step 1: Wire up free data (week 1)
Install OpenBB and pull a year of daily closes for a 20-stock watchlist with obb.equity.price.historical(symbol, provider="cboe"), which needs no API key.
Then grab each ticker’s headlines with yfinance’s free yf.Ticker(symbol).news.
That’s the terminal’s data layer, for $0.
Step 2: Build the 2-sigma inbox (week 2)
I call this the "2-sigma inbox".
Python, not the AI, decides what matters.
A stock passes the screener only when today’s return sits more than two standard deviations from its last 60 days.
On a typical day, only a few names will pass the test.
The others never reach the agentic workflow. That’s the point.