Component Diagram

This document details the internal structure of each main container using the C4 Component level.

Backend API Components

• 🌐 Data Ingestion Module: Fetches and normalizes data from external sources (APIs, CSV).
  • 📉 Market Data
    • Yahoo Finance API (via yfinance)
    • Alpha Vantage – Free API for stock, forex, and crypto
    • IEX Cloud – Real-time and historical data
    • Polygon.io – High-quality market data (paid tiers)
    • Finnhub – Real-time and fundamental data
  • 📊 News & Sentiment (Optional)
    • NewsAPI – Financial news aggregation
    • Sentiment APIs – For NLP-based sentiment scoring
• ETL/Processing Module: Handles batch and streaming ETL using Python (Polars, PyArrow, etc.).
• Query Engine: DuckDB (lightweight, SQL on Parquet)
• Backtesting & Strategy Engine: Runs trading strategies (rule-based and event-driven), backtests, and generates signals.
  • backtrader – Strategy backtesting
  • bt – Portfolio backtesting
  • zipline – Quantitative finance backtesting (used by Quantopian)
• Trade Execution Module: Interfaces with broker APIs to place/cancel orders.
• Authentication & Authorization: Verifies Auth0-issued JWTs (RS256) via JWKS, applies scopes/roles for protected endpoints; user profile/settings stored in app DB as needed.
• Reporting & Analytics: Generates reports, dashboards, and logs (numpy – Numerical operations, ta – Technical analysis indicators, scikit-learn – Machine learning (optional), statsmodels – Statistical modeling).
  • matplotlib / seaborn – Static plots
  • plotly – Interactive charts

Data Storage Components

• NoSQL Database: Stores earnings event data, trade signals, logs, and unstructured information, Real-time access for dashboards or agents Flexible schema that may evolve.
  • DBMongoDB (open-source, easy Docker setup)
• Parquet Storage: Stores historical and analytical data for fast querying. Storing historical price data, backtest results, performance metrics Efficient analytics and batch processing Integration with tools like DuckDB
• Database Abstraction Layer: Provides a unified interface for data access (NoSQL/Parquet).

🧠 Would a Vectorized Database Help?

Vectorized databases (like Pinecone, Weaviate, Qdrant, Milvus, or Chroma) are designed for storing and searching high-dimensional vectors—typically used for:

Semantic search (e.g., finding similar news, stocks, or patterns) Retrieval-augmented generation (RAG) for LLMs Clustering and anomaly detection in large datasets

Infrastructure Components

• Docker Containers: Encapsulate backend, UI, and supporting services for portability.
• CI/CD Pipeline: Automates build, test, and deployment (GitHub Actions).
• Hosting Environment: Linux server (cloud or bare metal) for running containers and services.
• LLM API (OpenAI, Azure OpenAI, Cohere, etc.)
• Orchestration: Prefect (open-source workflow orchestration)

Documentation Components

• VitePress Site: Vue 3 + Vite-based static site generator for comprehensive technical documentation.
  • Architecture documentation (C4 models, diagrams)
  • API reference
  • Strategy documentation
  • Developer guides
  • Component documentation
• Markdown Source: Version-controlled documentation alongside code
• Automated Builds: Documentation built and deployed as part of CI/CD pipeline

AI Components

Building Your Own AI Agent can:

• Automate data ingestion, screening, and reporting
• Generate trade ideas or alerts based on your strategy
• Summarize news, filings, or social sentiment
• Interact via chat or dashboard (using LLMs like GPT-4, Llama, etc.)

Web UI Components (Svelte)

• Dashboard: Main interface for monitoring and control (Svelte).
• Strategy Editor: UI for creating and editing strategies (Svelte).
• Results Viewer: Displays backtest and live results (Svelte).
• Auth Client: Integrates with Auth0 Universal Login using Authorization Code with PKCE; stores tokens in memory and attaches bearer token to API calls; handles token refresh and logout.

Cross-Cutting Concerns

• Configuration Management: Centralized config for all modules.
• Logging & Monitoring: Tracks system health, errors, and performance.
• Security: API key management, access control, and secure communication.
  • Authentication: Auth0 (Free); SPA uses OIDC PKCE, API validates JWT; environment-specific settings via config (domain, audience, client id).
  • Authorization: start with coarse-grained scopes; expand to role-based permissions as strategies and UI features grow.