Independent Project · Python · 2025–Present
Self-built Trading Research System
Paramo is a Python codebase designed to identify and validate patterns in extreme, news-driven price moves. This document covers the architecture behind it, the technical challenges solved along the way, and the skills required to build it.
What It Means to Build with AI
Paramo did not exist before AI coding tools made it possible for one person to build it. The platform spans data ingestion, storage architecture, bias correction, pattern recognition, and modular infrastructure capable of both live trading and backtesting. Each component would have required a specialist in a traditional development environment. AI was used as a force multiplier throughout, not as a shortcut. The AI handled code generation, infrastructure scaffolding, and integration problem-solving. The contextual judgement, architecture decisions, research logic, and quality control were mine. The result is a production-grade research system built solo that demonstrates what a competent AI pilot actually looks like in practice.
Key Technical Challenges Solved
Multi-Source Integration
Four financial data APIs with different schemas, coverage gaps, and undocumented inconsistencies. Built every ingestion pipeline independently and cross-validated data across sources before any analysis ran.
Dual-Mode Architecture
A backtest engine and live trading system share the same core logic but have fundamentally different runtime requirements. Designed a modular shared core where all strategy logic, pattern scoring, and exit management runs identically in both modes, with thin mode-specific adapters handling only the data layer.
Computational Scalability
Running full pattern analysis across 5,000 symbols with a 30-month lookback was projected to exceed 30 hours of runtime. Resolved by redesigning the screening pipeline to apply hard gap and price filters, dramatically reducing the symbol pool by 90% before hitting the expensive computation layer.
Lookahead Bias
Backtest results were implausibly strong. Diagnosed forward-looking price information leaking into entry logic, enforced strict temporal data boundaries, and revalidated the full backtest suite to confirm results reflected only information available at entry time.