Selected side projects
Market modeling and public data.
Two projects with different goals: a machine-learning research pipeline for trading simulations, and a small transit app designed to be useful right away.
Stock Market Trading Bot
A Python pipeline for researching whether temporary price divergence between two share classes can generate useful trading signals.
Historical GOOG / GOOGL backtest · 2022
Cumulative simulated return
THE RESEARCH PIPELINE
From raw market data to a testable strategy.
The project ingests one-minute Tiingo data, engineers price and volume features, models the relationship between dual-class equities, and trains an RBF support-vector classifier across sliding time windows. A custom backtesting layer then simulates entries and exits and reports profit, precision, Sharpe, and Sortino measures.
The result is a historical simulation, not a claim of future returns. The project does not place real trades.
Market dataOne-minute Tiingo data for multiple dual-class equities
Feature engineeringRelative price, moving average, RSI, and money-flow signals
Statistical modelingOrnstein–Uhlenbeck spread model with sliding train/test windows
Machine learningRBF support-vector classification with scaled features
EvaluationCustom trade simulation, precision, Sharpe, and Sortino calculations
Open Transit Chicago
A free transit tracker for CTA trains and buses. I built the first version in a week to explore real-time data, API design, and PWA delivery.
WHAT IT DOES
Nearby arrivals without an account or subscription.
The app locates nearby stops, displays live rail and bus arrivals, and shows vehicles on a map. It is installable as a PWA and avoids user tracking.
Web clientReact, TypeScript, Leaflet, responsive PWA support
APIJava, Spring Boot, server-side CTA credentials
Upstream controlsCaching, rate limits, and usage budgets
DeliveryDocker, automated tests, secret scanning, and CI/CD