A quant library you can look at but not touch

What it does

MlFinLab is a commercial Python toolkit for financial machine learning, covering the full pipeline from data structures and feature engineering to backtest statistics and bet sizing. The company, Hudson & Thames, sells it under Business and Enterprise licenses with Slack support included.

The interesting bit

The GitHub repo is deliberately hollow: no source code, no commits to browse, just an issue tracker and marketing copy. It is an unusually honest admission that the “open” in open-source here means open a support ticket, not open the codebase. The actual library ships through a separate portal with lecture videos and example notebooks.

Key highlights

• Fourteen modules spanning labeling, sampling, clustering, synthetic data generation, network analysis, and codependence measures
• Extensive documentation and video lectures bundled with purchase
• Example notebooks showing complete strategy pipelines
• Slack community access for licensed clients
• “All rights reserved” license—no trial, no freemium tier visible

Caveats

• The README claims “every line of code… is extensively tested” but provides no evidence; there is no code to inspect
• “What was only possible with huge R&D teams” is marketing boilerplate, not a benchmark
• No pricing transparency; purchase flow requires portal sign-in

Verdict

Worth evaluating if you are a portfolio manager with budget and a preference for vendor-supported quant tools over rolling your own. Skip it if you need to audit algorithms, modify implementations, or avoid recurring license costs.