Trading firm

Old Growth Harbor

Company hiring

Trader Maths does not track Old Growth Harbor's job board yet, so there are no open roles to show here. Browse open roles at other firms.

About Old Growth Harbor

Old Growth Harbor LLC is a proprietary quantitative research and trading firm founded in Seattle, Washington, in 2026. Proprietary trading means the firm trades only its own capital. It does not solicit, accept, or manage client or external funds. The firm applies machine learning and large language models (LLMs) to capital markets, combining high-dimensional mathematics, deep learning, and the study of market microstructure, which concerns how orders are placed and matched.

Its research covers three areas. Reinforcement learning is used to build decision engines that operate in changing market conditions within set risk limits. Custom-trained LLMs derive quantitative signals from large unstructured datasets. High-frequency modeling of the limit order book, the record of outstanding buy and sell orders, supports systematic order execution. Positions are based on machine learning hypotheses tested against multi-decade datasets, and risk constraints are built into model loss functions.

The in-house Strategy Lab platform, written in Rust and using Arrow, Parquet, DuckDB and Polars, supports reproducible backtesting (simulating strategies on historical data) and walk-forward validation (out-of-sample testing across successive periods). It also supports Monte Carlo stress testing, parameter optimization and cross-engine verification. The firm uses TimescaleDB for tick and price-bar data and qgraph, an internal knowledge graph built on PostgreSQL with pgvector. Old Growth Harbor is occasionally open to hearing from scientists in reinforcement learning, mathematical linguistics or limit order book microstructure, and accepts resumes or research papers.

All trading firms