AI trading for the masses

We analyse financial data, then trade on it.

An AI trading system built from the ground up, tested against ten years of market history, and run under rules that code enforces.

Read the market

Every morning the system scans thousands of stocks for the ones that are moving on real news: earnings, guidance, deals. Price, volume and headlines, all before the open.

Judge the news

Not every headline is worth a trade. The system reads each one and keeps only real, material, company-specific news. Hype, dilution and vague chatter get thrown out.

Trade with a hard stop

Code decides the size of every trade and where the stop goes. The AI can suggest, but it can never raise risk. A bad day, a bad week or a deep drawdown shuts trading down automatically.

How a trade happens.

The same four steps, every trading day. Nothing is discretionary.

1
Scan before the open
Find stocks gapping up on news, in a healthy long-term trend, liquid enough to trade cleanly.
2
Grade the catalyst
Read the pre-market headlines and keep only the days where the news is real and the reason is clear.
3
Enter on the break
Wait for the first fifteen minutes. If the stock breaks above that range, buy, with the stop just underneath.
4
Manage by rules
Take part of the profit when the trade works, trail the rest, and exit on a set day if it goes nowhere. No feelings involved.

Where we are, honestly.

Built: 2025 to 2026
Tested on: Ten years of market history, including delisted stocks
Right now: Paper trading with real market data and fake money, since September 2026
Real money: Only when the paper record earns it

No performance numbers are published until there is a real-money track record. Nothing here is investment advice.

The trading journal.

What was tested, what failed, what held. Kept in public so the story can't be rewritten later.

Date	Entry	Status
2026-09-02	A free news grader replaces the paid one Rewrote the catalyst judgement as plain rules and checked it against thousands of graded headlines. The bot now runs at zero API cost.	Running
2026-09-02	Paper trading starts The full loop runs live on a demo account: scan, grade, enter on the opening-range break, manage at the close, report by message.	Running
2026-09-02	Include the stocks that died Added 1,500 delisted companies to the history so the tests can't quietly ignore the losers. The edge held.	Closed
2026-09-01	Enter on the day, not the day after Buying the morning after a news gap lost money. Buying the break of the first fifteen minutes on the day itself did not. The entry is the edge.	Closed
2026-08-29	Does the news actually matter? Graded 2,453 news gaps A, B or C and traded each group. A beat B beat C, in order. The filter is real signal.	Closed
2026-08-25	Foundation Data store, broker connection, alerts, and a risk gate that no model can override.	Closed

Catherina.

I'm building an AI system that trades the market the way it should be traded: on real news, with strict rules, without emotion.

The idea is simple: the best short-term trades come from real news, a stock that was quiet before it, and a strict stop. Doing that consistently, every morning, without fear or greed, is what a machine is for. The system is being proven on paper first. When it has earned real money, it will trade real money.

Role: Founder
Based: Oslo, Norway
Focus: AI trading systems
Also built: Snakke, Dishy, Notax

Be first when it goes live.

Leave your email and you'll hear from Catherina when the system opens up. No newsletter, no noise.