About the project
A low-latency Bitcoin trading bot, built by students, designed for students.
This project is a student-built Bitcoin trading bot created for a competition. Instead of hiding everything behind a black box, we focused on clear structure: live data, an evolving strategy, a transparent UI, and a full trade history.
How the bot works
Our system uses a Genetic Algorithm (GA), an optimization method inspired by evolution. Instead of following fixed rules, the bot learns from experience, adapts to conditions, and evolves toward smarter decisions.
- 1Population of strategies
The bot generates multiple strategy variants, each with different moving-average periods, RSI thresholds, and risk parameters.
- 2Fitness evaluation
Each strategy is backtested against real market data. We score on net profit after fees, drawdown, and consistency — not raw profit alone.
- 3Crossover + mutation
The top strategies breed new ones by mixing parameters, and mutation introduces new variations so the bot can discover patterns it hasn't tried yet.
- 4Evolution over time
Over many generations, the bot learns what works and adapts automatically.
Engine specifications
- Market data: Binance BTC/USDT WebSocket + REST
- Strategy model: Genetic Algorithm, evolving every generation
- Population size: 24 individuals per generation
- Fitness function: Net equity, fee-adjusted, drawdown-penalized
- Execution mode: Simulated spot positions (no real funds)
- Platform stack: Next.js, Zustand, TradingView, WebSockets
Why we built it
About our team
Mission statement
Our values
- • Transparency over hype
- • Learning through real data
- • Engineering before shortcuts
- • Curiosity and experimentation