All Projects

Polymarket Trade Collector

Cloud-based real-time data pipeline capturing prediction market microstructure from Polymarket via WebSocket. The system collects full order book depth and trade streams, currently amassing 650K+ rows and 13K+ trades with automated cloud-to-local replication and daily archival. Built to study market microstructure dynamics in prediction markets, with early analysis revealing potential alpha signals in tail-size trades and asymmetric Yes/No position dynamics.

Key features include real-time WebSocket connection to Polymarket API, SQLite storage with automated syncing, structured data capture of bids/asks/trades, and cloud infrastructure for 24/7 collection with local backup. Future work includes statistical analysis of order flow patterns, visualization dashboards, and potential strategy development.

LSTM Algorithmic Trading Model

Designed an LSTM-based deep learning model for t+1 Bitcoin price prediction using Fear and Greed Index data. Incorporates hyperparameter optimization via Optuna, comprehensive visualization, and robust performance analysis.

Neural Network Trading System

End-to-end neural network trading system covering data preprocessing, model training, backtesting, and performance evaluation. Demonstrates applied quantitative research capabilities in algorithmic trading.

NFL Ticket Market Analysis

Comprehensive 120-page quantitative research report analyzing multi-season NFL ticket market behavior with probabilistic price capitulation modeling, historical trend analysis, and 1K+ generated data visualizations.

Implied Volatility Models

Developed historical implied volatility model charts for 15 commodity, currency, and index futures markets at Mt. Lucas Management. Provided quantitative tools supporting derivatives analysis and market monitoring.

Portfolio Optimization Tool

Interactive tool that constructs optimal investment portfolios based on user-defined risk tolerance and desired return profiles. Implements Modern Portfolio Theory to find the efficient frontier and recommend asset allocations.

Lookahead Bias Research

Research article on "Predicting the Future Without Cheating: Overcoming Lookahead Bias with the Window Method" - exploring best practices in backtesting trading strategies to avoid common pitfalls.

Crypto Lending Analytics

Financial models developed at TradeStation Crypto to assess loan performance, counterparty exposure, and collateral adequacy. Supported scaling institutional crypto lending portfolio from $60M to $180M.

Market Data Pipeline

Automated data extraction and analysis pipeline for secondary ticket market research. Integrates multiple data sources including SEO analytics, demographic data, and social signals for demand forecasting.

Relative Value Pricing Engine

Pricing strategy tool designed to enhance bid-offer decision quality in market-making operations. Assists with data model architecture, exploratory analysis, and automated strategy implementation.