Trading Automation
Modular Python trading system that codifies discretionary execution, integrates live feeds and backtesting, and removes ~1 hr/day of manual work with built-in risk control.
Modular architecture in Python and Postgres with real-time data pipelines and integrations via API and WebSocket.
Duration
Present●
Introduction
Built a modular trading engine that removes ~1 hour/day of manual work, codifies discretionary execution, and accelerates experimentation. The system separates data, signals, execution, and analytics; AI speeds research while automated sizing and risk limits reduce errors.
Challenge
Discretionary trading requires constant attention, manual risk control, and consistent post-analysis. Existing tools didn't match my mental model or allow seamless iteration. The challenge was designing a system that didn't just execute tasks, but amplified my edge without losing flexibility or speed.
Approach
I designed a modular architecture aligned with how I actually think and trade. Prioritizing simplicity and extensibility, the system supports everything from visual analysis to structured backtesting and rapid strategy deployment. Every module is independent, composable, and easy to iterate. AI accelerates insight generation, but decision-making stays human.
Market Data & Signal Layer
- Real-time feeds from multiple exchanges via WebSocket and REST APIs
- PostgreSQL for storing price history and execution data
- Custom scanners filtering thousands of assets with dynamic criteria
- Automated alerts triggered by predefined setups
Execution Support & Analytics
- Signal engine codifying discretionary strategies into trigger logic
- Position tracker with live PnL, automated risk metrics, and exposure guardrails
- Backtesting module to validate hypotheses before going live
- Streamlit dashboards visualizing portfolio health and exposure
AI-Driven Research
- Sentiment analysis using OpenAI across news and macro headlines
- Pattern detection in historical time series data
- Automated daily market summaries and risk reports
- Ongoing experimentation with directional bias from diverse sources
Results & Impact
The platform shifted my time from execution to decision‑making: 1+ hour saved daily, fewer execution mistakes via auto‑sizing and risk limits, and faster iteration through structured backtests across 50+ strategy variants.
Efficiency Gains
- 1+ hour saved daily by eliminating repetitive execution workflows
- Instant visibility into risk, exposure, and performance history
- Automated sizing logic to prevent overexposure
- One-click access to strategy-specific performance data
Learning & Development
- Tested 50+ strategy variations through structured experiments
- Deepened skills in Python, data pipelines, and model deployment
- Improved understanding of market microstructure through repeatable analysis
- Laid groundwork for scaling into fully systematic strategies
The system continues to evolve as my core environment for backtesting, execution, and research. It's not finished, but it's already become a critical tool for exploring ideas at speed, with precision and control.