Why this lab exists, and how decisions are computed
A trained engineer in software for over twenty years, I turned to artificial intelligence two years ago. I wanted to put it to the test on the stock market and see what it can actually deliver: analysis, decisions, and risk control. This site is the open laboratory: a simulated portfolio, a daily record, and a transparent view of every decision.
This site documents the daily outputs of a paper-trading laboratory. The system combines AI-assisted analysis, deterministic risk rules, and post-close simulated execution. You can follow:
An AI-assisted decision never automatically becomes a position: it follows a structured, logged process.
01
The system gathers market signals and research inputs to establish context.
02
A strategic conviction and thesis are formalized for each selected name.
03
The plan states whether to buy right away, wait for a better price, or add nothing for now.
04
Position size, available cash, and portfolio concentration are rechecked before any new exposure.
05
Orders are simulated once per day, after the markets close.
06
Decisions, statuses, and outcomes are recorded for longitudinal review.
Each public decision includes its date, conviction, plan, and observed status. The journal also retains paper-order events so favourable and unfavourable outcomes remain reviewable.
This project is a technical and educational experiment.
Artificial intelligence is a powerful decision-support tool, but it is by no means infallible. It can misread analysis or face the unpredictable swings of markets. Nothing on this site constitutes investment advice or an invitation to buy or sell financial assets.
This site shows a fictional portfolio: no real orders are sent. Each day, the system simulates what it would have done at the next market open.
Performance is recalculated daily from cash and position values, then compared with the S&P 500. Simulated fees and a price slippage allowance are deducted on each paper trade.
What the trader thinks and what it does are two different things.
Positive conviction does not necessarily mean a position is already held. The system may schedule a paper order for the next open, wait for a better price, or simply keep an existing position.
A paper order may schedule staged exposure for the next open. Until it is executed, the position is not held.
The system is interested in a stock but waits for a better entry point or a specific condition before considering a purchase.
The position or conviction is maintained, with no immediate new purchase.
Before any new position, the system checks that the portfolio can absorb the risk.
Predefined limits cover in particular:
After significant losses, the system may also temporarily reduce its activity.
