Results from my trading bot
This page combines the editorial explanation of the project with the live data that used to be published from the bot runtime: current portfolio, capital evolution and latest executed orders.
Live results
The interactive section is protected with Google reCAPTCHA before enabling the bot queries.
What the bot tried to do
The goal was to reduce impulsive trading decisions and test whether a programmed set of rules could improve portfolio management. That led to several building blocks: portfolio state, historical capital evolution and the latest executed operations.
The experiment should be read as a software and data project, not as financial advice. The interesting part is how the bot turns market information into a repeatable process: read the current state, compare it with the strategy, execute an action and store the result for later analysis.
Environment considerations
Any trading agent first needs a market model: current price, history, liquidity, cash, held shares and transaction cost. That simplified model is the base on which automated strategies can be tested.
A real strategy also needs guardrails: maximum exposure per asset, maximum number of trades, stop conditions, logging, data quality checks and a way to disable the bot when the data provider or broker behaves unexpectedly.
Limitations and useful lessons
The hardest part is not drawing a chart or storing the portfolio. The hard part is avoiding false confidence. A backtest can look good because of overfitting, missing fees, survivorship bias or a period that happens to favor the selected rule.
For that reason, the value of this page is educational: it documents the pieces required by an automated trading workflow and links them with Python, market data and API articles in the rest of the site.