NexPatch / finPatch Platform

Modern, data-driven trading and forecasting systems face the challenge of reliably processing large amounts of time-dependent data from different sources. This includes market and price data, blockchain data, measurement and event data as well as derived features for machine learning models. This data must be continuously captured, processed, stored and analyzed in order to enable reliable predictions and automated decisions. Classic trading or ETL systems usually only cover partial aspects and are often not designed for end-to-end workflows.

NexPatch / finPatch was developed as a modular, containerized end-to-end platform that maps the entire lifecycle of data-driven trading and forecasting systems. The platform combines visual data engineering, GPU-accelerated machine learning, signal and risk assessment as well as automated trading in a unified architecture. Data sources can be connected via the visual pipeline builder (nxpWarehouse), time series data can be transformed and prepared ML-ready. Modern PatchTST time series models are trained, monitored and used for continuous predictions. The finPatch web application provides users with a secure, subscription-based interface to monitor and control projects, signals, predictions and trading activities in real time.