Open-source AI financial intelligence case study

Blum turns public market data into explainable equity and ETF intelligence.

This project is a technical case study, not a consumer trading app. It combines public RSS ingestion, yfinance market data, PostgreSQL persistence, modular quantitative scoring, FinBERT sentiment, sentence-transformer semantic search, a lightweight LLM reasoning layer and time-series anomaly analysis.

Open Intelligence Dashboard Read Methodology

FrontendNext.js

BackendFastAPI

DatabasePostgreSQL + Alembic

AI sentimentFinBERT

Semantic layerSentence Transformers

ReasoningQwen-compatible LLM

DeploymentDocker Space

Market Universe

Stocks, ETFs, sectors, countries, themes and exchanges are normalized into a PostgreSQL asset universe.

Signal Engine

Momentum, trend quality, volatility, technical indicators, semantic news intensity, ETF confirmation and anomalies produce a Blum Intelligence Score.

Explainability

Each surfaced asset includes why it emerged, what confirms it, what contradicts it, risk level, watch points and next evidence to monitor.