Projects

A Python system I built to measure whether discretionary macro managers collectively permit equity risk. It uses normalized 13F exposure (relative to each manager’s own historical max), rate of change, and cross-manager dispersion to detect belief strain before macro breaks. This is a regime-state indicator over months/quarters—not a timing model or stock-selection tool—and is designed to stay transparent and configurable.

Data infrastructure I built to track real-time short-sale borrow fees for 17,000+ securities across 18 global markets. Collects Interactive Brokers' borrow rates and margin requirements every 15 minutes to support my flow trading analysis and short squeeze detection. Features efficient delta encoding (98% compression), Parquet caching for fast queries, and automated daily rebuilds. Designed to integrate with market-state-detector for comprehensive flow regime classification.

A Python tool I built to detect high-uncertainty market conditions using volatility spikes, price gaps, and abnormally wide trading ranges. It automates the checks I'd do manually to identify unstable market regimes and helps me avoid making trading decisions during periods of elevated uncertainty. Includes context detection to distinguish between broad market stress, sector-specific issues, and stock-specific events.