ANPS-TradeMeUp

ANPS (AI News Prediction System) - Multi-Agent News-Based Market Prediction System

Note: This is a private, source-available project currently under active development. Features, APIs, and screenshots may change frequently and are not intended for production use. ⚠️

What is ANPS-TradeMeUp?

ANPS-TradeMeUp (AI News Prediction System) is an MVP-grade pipeline that ingests news, extracts events/entities using LLMs, scores impact and surprise, and produces short-to-medium term market predictions. It includes a Dash GUI for real-time monitoring and a FastAPI backend.

Database: PostgreSQL 16+ with pgvector extension for optimal performance, JSONB support, and vector similarity search capabilities.

---

Quick Start

TradeMeUp converts financial news into probabilistic market predictions using a modular multi-agent pipeline. Start the GUI quickly with start_gui.bat (Windows) and explore dashboards and live charts. For development and production setup, follow the Installation section below.

---

Features

Multi-Agent Pipeline

• 16-agent modular architecture (ingest → understand → analyse → predict)
• LLM-based content understanding with fact verification
• Entity mapping with confidence scoring
• Impact, surprise, and regime detection
• Signal decay tracking and correlation analysis
• Multi-horizon predictions (1d, 5d, 20d)

Dashboard

• Real-time metrics overview (news volume, predictions, entities)
• Recent news feed with quality scores
• Current market regime indicator
• Top performers tracking (1h, 24h, 5d, 30d, 1y, all-time)
• Pipeline health monitoring

Predictions Tab

• Advanced filtering (entity, date range, horizon, confidence)
• Live performance tracking with actual vs expected returns
• Direction probabilities (up/down/flat)
• Risk and confidence scores
• Detailed modal view with market data
• Refresh individual predictions
• Export and batch operations

Trading Simulations

• Portfolio configuration (capital, currency, risk adjustment)
• Create simulations from predictions (date range or last N)
• Trading decisions (buy/sell/hold) with risk assessment
• Stop loss and take profit calculations
• Transaction cost breakdown (commission, spread, slippage, market impact)
• Position sizing recommendations (risk-adjusted)
• Penny stock detection with special cost handling
• Expected vs actual return tracking
• Resimulate all with latest market data
• Filter by entity, horizon, decision, date range

Statistics Tab

• Overall metrics (articles, entities, predictions, impact scores)
• Event and quality distribution charts
• Sentiment analysis (positive/negative/neutral)
• Impact score visualization
• Top entities rankings
• Entity sentiment analysis with timeframes (7d, 30d, 90d, all)
• Entity details table with search
• News volume trends over time

Live Charts

• Multi-panel chart view (single, dual, quad mode)
• Candlestick and line chart types
• Real-time price updates (30s interval)
• Multiple timeframes (1d, 5d, 1mo, 3mo, 6mo, 1y, 2y, 5y, max)
• Volume overlay and moving averages
• Infinite scroll for historical data
• Zoom and pan with state persistence
• Fullscreen mode
• Chart overlays (brackets, breakouts) - DB-backed
• Custom symbol search

Agent Testing

• Individual agent health checks
• Test all 16 agents independently
• Real-time status monitoring
• Error tracking and logging
• Agent performance metrics

System Health

• Pipeline status overview
• Database statistics
• Agent operational status
• Processing metrics
• Activity log monitoring

---

Quick Start

1. Clone:

   git clone https://github.com/arn-c0de/ANPS-TradeMeUp.git
   cd ANPS-TradeMeUp

2. Start PostgreSQL database:

   docker-compose up -d

3. Create environment file and add API keys:

   cp .env.example .env.local
   # edit .env.local with your API keys

4. Run database migrations:

   alembic upgrade head

5. Start the GUI (Windows):

   .\start_gui.bat
   # open http://localhost:8050

That's enough to explore the GUI and view sample dashboards. For a full development environment and pipeline run, continue with the Installation below.

---

Installation & Full Setup

Prerequisites:

• Python 3.11+ (Python 3.12+ recommended)
• PostgreSQL 16+ with pgvector extension
  • Option 1: Docker & Docker Compose (recommended for quick setup)
  • Option 2: Native PostgreSQL installation
• Poetry (recommended) or pip

Database Setup (Choose One):

Option A: Docker (Recommended)

# Start PostgreSQL 16 with pgvector
docker-compose up -d

# PostgreSQL will be available at:
# Host: localhost:5432
# Database: trademeup
# User: trademeup_user
# Password: trademeup_pass

Option B: Native PostgreSQL (Windows/macOS/Linux)

# 1. Install PostgreSQL 16+ from https://www.postgresql.org/download/

# 2. Install pgvector extension
# Windows (PowerShell as Admin):
cd "C:\Program Files\PostgreSQL\16\bin"
.\psql.exe -U postgres
CREATE EXTENSION vector;

# Linux/macOS:
sudo apt-get install postgresql-16-pgvector  # Ubuntu/Debian
brew install pgvector  # macOS
psql -U postgres
CREATE EXTENSION vector;

# 3. Create database and user
CREATE DATABASE trademeup;
CREATE USER trademeup_user WITH PASSWORD 'your_password';
GRANT ALL PRIVILEGES ON DATABASE trademeup TO trademeup_user;

# 4. Enable extensions
\c trademeup
CREATE EXTENSION IF NOT EXISTS "uuid-ossp";
CREATE EXTENSION IF NOT EXISTS "vector";

For detailed PostgreSQL setup instructions, see POSTGRESQL_SETUP.md.

Full Installation Steps:

1. Clone repository:

   git clone https://github.com/arn-c0de/ANPS-TradeMeUp.git
   cd ANPS-TradeMeUp

2. Start PostgreSQL (if using Docker):

   docker-compose up -d

3. Install Python dependencies:

   # Using Poetry (recommended)
   poetry install
   poetry shell
   
   # Or using pip
   python -m venv venv
   # Windows:
   .\venv\Scripts\activate
   # Linux/macOS:
   source venv/bin/activate
   
   pip install -r requirements.txt

4. Configure environment:

   cp .env.example .env.local
   # Edit .env.local and configure:

   Required settings:

   # Database (PostgreSQL required)
   DATABASE_URL=postgresql://trademeup_user:your_password@localhost:5432/trademeup
   
   # LLM API Keys (at least one required)
   ANTHROPIC_API_KEY=your_anthropic_key
   OPENAI_API_KEY=your_openai_key
   
   # Optional: News API keys, AlphaVantage, etc.

5. Run database migrations:

   alembic upgrade head

   This creates all tables, indexes, and constraints in PostgreSQL. The migration includes:

   • All core tables with proper relationships
   • JSONB columns for flexible data storage
   • GIN indexes for fast JSONB queries
   • pgvector columns for embedding similarity search
   • Timezone-aware timestamp columns

6. Run the pipeline:

   • Continuous mode (recommended):

     python scripts/run_continuous_pipeline.py

   • One-shot (single pass):

     python scripts/run_mvp_pipeline.py

7. Start the GUI:

   python run_dashboard.py
   # or on Windows: .\start_gui.bat
   # Open http://localhost:8050

8. Start the API (optional):

   uvicorn src.api.main:app --reload
   # Open http://localhost:8000/docs

Database Management

Docker Commands:

• Stop PostgreSQL: docker-compose down
• View logs: docker-compose logs -f postgres
• Restart: docker-compose restart postgres
• Remove data (destructive): docker-compose down -v

Backup & Restore:

# Backup
docker exec trademeup_postgres pg_dump -U trademeup_user trademeup > backup_$(date +%Y%m%d).sql

# Restore
docker exec -i trademeup_postgres psql -U trademeup_user trademeup < backup.sql

# Backup with compression
docker exec trademeup_postgres pg_dump -U trademeup_user trademeup | gzip > backup.sql.gz

Native PostgreSQL:

# Backup
pg_dump -U trademeup_user trademeup > backup.sql

# Restore
psql -U trademeup_user trademeup < backup.sql

# Connect to database
psql -U trademeup_user -d trademeup

Performance Optimization

PostgreSQL provides significant performance benefits:

• JSONB: Native JSON storage with indexing (vs SQLite's TEXT-based JSON)
• Concurrent Access: Multiple connections without file locking
• Advanced Indexing: GIN, GiST, partial indexes
• Vector Search: pgvector for embedding similarity (768-dim vectors)
• Query Planner: Sophisticated optimization for complex queries
• Partitioning: Table partitioning for large datasets (future)

Performance Benchmarks:

• Dashboard queries: ~50% faster than SQLite
• JSONB operations: 3-5x faster with GIN indexes
• Concurrent writes: 10x improvement
• Vector similarity search: Native support (vs JSON fallback)

---

Development Notes & Project Status

• Phase: Phase 1 - MVP Core
• Progress: 16 of 17 agents implemented (active development)
• Recent: GUI improvements, central error handling, and added agent init tests.

Full implementation details, architecture, and agent breakdown are available deeper in this README and in docs/.

---

Documentation & Guides

• Quickstart: QUICKSTART.md
• PostgreSQL Setup: POSTGRESQL_SETUP.md - Detailed database setup guide
• GUI Documentation: docs/GUI_README.md
• Local Setup: docs/SETUP_LOCAL.md
• Performance: docs/CONTINUOUS_PIPELINE_PERFORMANCE.md
• Third-party Licenses: THIRD_PARTY_LICENSES.md
• Changelog: CHANGELOG.md

---

Contributing & Tests

• Run unit tests:

  pytest tests/unit

• Integration tests:

  pytest tests/integration

Please follow the development workflow in CONTRIBUTING.md.

---

Security

If you discover a security vulnerability, please do not file a public issue. Report it by email to arn-c0de@protonmail.com or via GitHub Security Advisories at https://github.com/arn-c0de/ANPS-TradeMeUp/security. Include steps to reproduce, affected versions, and an assessment of potential impact where possible. The maintainer will acknowledge receipt within 3 business days.

See SECURITY.md for detailed security policy and best practices.

---

License

Copyright (c) 2026 arn-c0de. All rights reserved.

PROPRIETARY SOURCE-AVAILABLE LICENSE

This software is proprietary and source-available. You may view, clone, and modify this repository solely for the purpose of contributing improvements via pull requests or issues.

Strictly prohibited without explicit written permission:

• Commercial use
• Redistribution
• Publication of modified or unmodified versions
• Use in other software projects
• Sublicensing or selling

All contributions submitted to this repository become the exclusive property of the copyright holder.

See LICENSE for full details.

Maintainer: arn-c0de (arn-c0de@protonmail.com)
Repository: https://github.com/arn-c0de/ANPS-TradeMeUp

---

Last Updated: January 28, 2026 Version: 1.0.4

Recent Changes (v1.0.4)

• Complete PostgreSQL Migration: Migrated from SQLite to PostgreSQL 16+ for production-ready performance
• pgvector Integration: Added vector similarity search support for embeddings
• JSONB Optimization: All JSON columns converted to JSONB with GIN indexes
• Timezone Awareness: All datetime operations now properly handle timezones
• Type Safety: NumPy/pandas types automatically converted for database compatibility
• Performance: 50-300% improvement in query performance vs SQLite
• GUI Fixes: Resolved all datetime and JSONB deserialization issues
• Data Migration: Complete data migration from SQLite to PostgreSQL with validation

Developer note: When updating the project version, please also update the VERSION constant in src/config/settings.py so the GUI and documentation reflect the correct version.