Multi-agent financial news analysis and alpha factor mining platform.
Monitor live financial news, run agent debates, and mine alpha signals with multi-LLM support.
FinnewsHunter is an enterprise-grade multi-agent platform designed to analyze real-time financial news streams and generate quantitative alpha signals. It suits quantitative researchers and developers looking to deploy local agent teams for automated market sentiment assessment.
The system crawls major financial news sources and uses specialized agents like NewsAnalyst alongside debate workflows to interpret market impact. It pairs this textual analysis with akshare market data, Milvus vector storage, and a React frontend for deep stock and K-line exploration.
It suits quantitative analysts, developers, and finance professionals with Docker experience who want to build custom news-driven trading pipelines. It does not suit non-technical traders or those without API keys for supported large language models.
Built natively on AgenticX using core abstractions like Agent, Tool, and Workflow.
Switch between Bailian, OpenAI, DeepSeek, Kimi, and Zhipu models directly in the UI.
Pre-loaded with 5000+ A-share data points and integrated with akshare daily and minute K-lines.
Performs background vectorization tasks using Milvus and PostgreSQL without blocking the analysis flow.
Free under its open source licence, though you must pay for your own LLM API usage and infrastructure hosting costs.
880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
The platform supports Bailian (Alibaba Cloud), OpenAI, DeepSeek, Kimi (Moonshot), and Zhipu.
Celery workers continuously crawl financial news sources like Sina Finance, National Business Daily, and Securities Times.
See Zekai first in Google