Surpriver is an open-source stock analysis tool on GitHub from Tradytics that uses machine learning and anomaly detection to find stocks showing unusual price and volume behavior before they make a big move. The project has collected roughly 1,900 stars and is written in Python on top of scikit-learn.
Most large stock moves are preceded by subtle changes in trading activity. Surpriver quantifies that idea: it loads historical price and volume data from Yahoo Finance for a watchlist of tickers, generates a rich feature set including returns and technical indicators, and runs anomaly detection models that rank which stocks look most abnormal compared with their own history. The repository is organized into clear modules for data loading, feature generation and the detection engine, so quants can swap in their own universes, features or models. Originally released in 2021, it remains a compact, readable reference implementation of ML-driven unusual-activity scanning for equities.
Note that this listing covers the open-source Surpriver tool itself rather than Tradytics’ separate hosted trading platform: everything described here is the free GitHub project that traders can download, inspect, modify and run against their own watchlists.
Surpriver is completely free and open source. It runs locally on your own machine with freely available market data, and there are no subscriptions or API fees required by the project itself.
Surpriver suits retail quants, data scientists and trading hobbyists who want to experiment with anomaly detection on equities, build watchlist scanners, or learn how ML features can surface unusual market activity worth a closer look.