AalsiTrader

The problem

Trading manually means watching charts all day. I wanted a system that could scan the market, propose trades, manage positions, and report everything back to me in real time — while I get on with my life. The name is a joke at my own expense: aalsi means "lazy" in Hindi.

What I built

A full autonomous trading platform:

• Real-time dashboard — portfolio tracking, P&L, win rate, drawdown, and an equity curve, streamed over WebSockets.
• Paper trading engine — every strategy runs against virtual capital first, with a full trade journal and performance analytics (Sharpe, Sortino, Calmar, profit factor) before any real money moves.
• AI agent system — an orchestrator agent ("Sigma") reviews and approves or rejects trades proposed by the strategy layer, with every decision logged to an activity feed.
• Autonomous operation — scheduled scans every few minutes during market hours, position monitoring, and pre-market token health checks.

Interesting decisions

• A single Lambda monolith instead of microservices. One TypeScript Lambda handles the HTTP API, WebSocket connections, and trading logic. For a solo project, deleting infrastructure complexity beats theoretical scalability.
• Paper-first by design. The paper/live toggle is a first-class feature, not an afterthought — strategies must prove themselves on virtual capital before the switch flips.
• Event-driven scheduling with EventBridge rules rather than a long-running process: the system wakes up, does its job, and goes back to sleep. Fitting, for a lazy trader.

Outcome

Live in production at aalsitrader.com, trading autonomously on a schedule, with a mobile app build via Capacitor. The bigger outcome: I learned more about market microstructure, risk metrics, and AWS event architecture from this one project than from years of reading about either.