RedHood Systems

AI-Powered Market Intelligence Dashboard

Portfolio Project by Tazeem Chowdhury Transforming information overload into actionable trading insights using AI

---

Project Overview

Problem: Retail traders spend 2-4 hours daily monitoring 20+ information sources (X/Twitter, Substack) to identify market opportunities. This creates information overload and missed signals.

Solution: RedHood Systems aggregates multi-source feeds and uses Claude AI to extract the top 3 market narratives with entropy risk scoring (quantified uncertainty) and trade hypotheses. Every run generates a styled RedHood Reads HTML report and persists all data to SQLite.

Impact: Reduces research time by 80% (from 180 min → 30 min) while improving signal quality through systematic AI analysis.

---

Key Features

• AI Narrative Extraction: Claude AI (Opus 4.8) processes 50+ feeds to identify top market themes
• Entropy Risk Scoring: Quantifies market uncertainty (1-10 scale) using physics-inspired framework
• Thermodynamic Regime Detection: Auto signal generation from market temperature/entropy state (redhood_regime_detector.py)
• Trade Hypothesis Generation: Specific, actionable trade ideas with entry/exit logic
• Hypothesis Grading: 0–20 quality grade (Specificity / Catalyst / Risk Management / Cohesion) + A–F letter, with long-side ticker extraction (redhood_grader.py, score_narratives.py)
• Long-Only P&L Tracking: $2,500-per-ticker ledger over extracted tickers via yfinance (redhood_pnl.py)
• Star-Schema Persistence: Runs, feeds, narratives, tickers, grades, prices, earnings + Power BI views in redhood.db
• Power BI Export: SQLite → Power BI pipeline with Power Query (M), DAX measures, and report layout (powerbi/)
• RedHood Reads HTML Report: Styled editorial card report generated every run
• Account Management: CLI tool to manage tracked X/Twitter accounts
• Trading System Analysis: Thermodynamic position-sizing model via run.ps1
• Multi-Source Aggregation: X/Twitter (via Nitter RSS, no API key required), Substack RSS, and public Telegram channels (via the t.me/s/ web preview, no bot token required — includes @redhoodtrades)

---

Architecture

┌─────────────────────────────────┐
│  Data Sources                   │  Nitter RSS (X/Twitter), Substack RSS,
└─────────────┬───────────────────┘  Telegram (t.me/s/ preview), Yahoo Finance
              │
              ▼
┌─────────────────────────────────┐
│  Python Aggregator              │  redhood_aggregator.py
│  - Feed collection              │  + redhood_regime_detector.py (regime)
│  - AI narrative extraction      │  + ticker_extraction.py (tickers/sides)
└──────┬──────────────┬───────────┘
       │              │
       ▼              ▼
┌────────────┐  ┌─────────────────┐
│ SQLite DB  │  │ RedHood Reads   │
│ redhood.db │  │ HTML Report     │
└─────┬──────┘  └─────────────────┘
      │
      ├──▶ score_narratives.py ──▶ grades + long tickers (redhood_grader.py)
      ├──▶ redhood_pnl.py ───────▶ long-only P&L ledger (yfinance)
      └──▶ powerbi/export_to_powerbi.py ──▶ Power BI (Power Query + DAX)

---

Quick Start

Prerequisites

• Python 3.9+
• Anthropic API key (Get one here)
• PowerShell 5+ (Windows, for run.ps1)

Installation

# Clone the repository
git clone https://github.com/tazeemc/Redhood-Systems.git
cd Redhood-Systems

# Install dependencies
pip install -r requirements.txt
# (yfinance is optional — only needed for redhood_pnl.py / scoring with --with-pnl)

# Set up environment variables
echo ANTHROPIC_API_KEY=sk-ant-your-key-here > .env

Run via PowerShell (Recommended)

# Last 45 minutes (default), includes trading system analysis
.\run.ps1

# Last 5 hours, custom symbols
.\run.ps1 -Hours 5 -Symbols "AAPL","MSFT"

# Last 24 hours (full day)
.\run.ps1 -Hours 24

# RedHood aggregator only (skip trading analysis)
.\run.ps1 -SkipTrading

# Trading analysis only
.\run.ps1 -SkipRedHood

Run Python Directly

# Default (last ~10 minutes)
python redhood_aggregator.py

# Last 24 hours
python redhood_aggregator.py --hours 24

Score, Grade & Track P&L

# Grade every narrative + extract long-side tickers (idempotent)
python score_narratives.py

# Score only recent narratives, also refresh prices/P&L
python score_narratives.py --since 2026-03-01 --with-pnl

# Long-only P&L leaderboard ($2,500/ticker; needs yfinance)
python redhood_pnl.py --report

Run Tests & Static Analysis

pip install pytest ruff

# Unit tests (44 tests — extraction, grading, Telegram scraper, DB schema)
pytest

# Static analysis (also runs in CI on every push/PR)
ruff check .

Manage Tracked Accounts

# List all tracked Twitter accounts
python accounts_db.py --list

# Add an account
python accounts_db.py --add SomeHandle --category macro --notes "Description"

# Toggle active/inactive
python accounts_db.py --toggle FirstSquawk

Example Output

📰 Fetching RSS feeds...
   ✅ Found 18 RSS items

🐦 Fetching Twitter feeds...
   📋 Active accounts from DB: @unusual_whales, @FirstSquawk, @AutismCapital...
   ✅ Found 12 tweets

📢 Fetching Telegram channels...
   📋 Channels: @redhoodtrades
   ✅ Found 9 Telegram messages

📊 Total feeds collected: 39

🧠 AI Analysis Phase...
✅ Extracted 3 narratives

============================================================
📋 DAILY BRIEF - TOP NARRATIVES
============================================================

[1] Fed Signals Dovish Pivot
    Entropy Risk: 🟢 LOW (3/10)
    💡 Hypothesis: Long QQQ calls, 2-week timeframe
    📝 Rationale: Multiple Fed speakers indicate willingness to pause
        rate hikes if inflation continues cooling.
    📅 Catalysts: CPI data, FOMC minutes

💾 Results saved to: data/redhood_insights_20260222_083045.json
📰 Report saved to:  data/redhood_reads_20260222_083045.html
🗄️  DB: run #5 saved — 30 feeds, 3 narratives

---

Project Structure

Redhood-Systems/
├── redhood_aggregator.py       # Main aggregator + RedHood Reads HTML generator
├── redhood_regime_detector.py  # Thermodynamic regime detection + auto signals
├── ticker_extraction.py        # Extract tickers + Long/Short/Hedge sides from hypotheses
├── redhood_grader.py           # Hypothesis grader (0–20 / A–F) + long-ticker extraction
├── score_narratives.py         # CLI: grade narratives → narrative_tickers + narrative_grades
├── redhood_pnl.py              # CLI: long-only P&L ledger via yfinance
├── backfill.py                 # Backfill narrative_tickers from historical narratives
├── accounts_db.py             # CLI: manage tracked X/Twitter accounts in SQLite
├── models.py                  # SQLite star schema (tables + Power BI views) + init helpers
├── demo.py                    # Demo mode with sample data (no API key required)
├── run.ps1                    # PowerShell runner: trading analysis + aggregator
├── redhood.db                 # SQLite database (star schema)
├── requirements.txt           # Python dependencies
├── pyproject.toml             # pytest + ruff configuration
├── tests/                     # Unit test suite (pytest, network-free)
├── .github/workflows/ci.yml   # CI: ruff static analysis + pytest matrix
├── .env                       # ANTHROPIC_API_KEY (not committed)
├── powerbi/                   # Power BI export pipeline
│   ├── export_to_powerbi.py   #   SQLite → Power BI dataset
│   ├── power_query.m          #   Power Query (M) source
│   ├── dax_measures.dax       #   DAX measures
│   └── report_layout.md       #   Report layout spec
├── docs/                      # Published RedHood Reads + Power BI integration guide
├── PRD_RedHood_Systems.md    # Product Requirements Document
├── Market_Research_Analysis.md# Competitive analysis & market sizing
├── CASE_STUDY.md              # Portfolio case study
├── README.md                  # This file
└── data/                      # Output directory (gitignored)
    ├── redhood_insights_*.json # Raw feed + narrative data
    ├── redhood_reads_*.html    # Styled RedHood Reads report
    └── TradingAnalysis_*.json  # Trading system output

---

Portfolio Artifacts

This repository contains key documents demonstrating PM skills:

1. Product Requirements Document (PRD)

• File: PRD_RedHood_Systems.md
• Contents: Problem statement, user personas, feature specs, success metrics, roadmap
• Demonstrates: Strategic thinking, user research, technical specification

2. Market Research & Competitive Analysis

• File: Market_Research_Analysis.md
• Contents: TAM/SAM/SOM analysis, competitive landscape, pricing strategy, GTM plan
• Demonstrates: Business acumen, market sizing, competitive positioning

3. Working Prototype

• File: redhood_aggregator.py
• Contents: Production Python code — feed scraping, Claude AI integration, HTML report generation, SQLite persistence
• Demonstrates: Technical execution, coding ability, systems thinking

---

Technical Stack

Backend:

• Python 3.9+
• Anthropic Claude API (claude-opus-4-8)
• feedparser (RSS + Nitter RSS parsing)
• urllib + regex (public Telegram channel scraping via t.me/s/)
• python-dotenv (environment config)
• yfinance (optional — long-only P&L tracking)

Data Storage:

• SQLite via redhood.db — star schema (twitter_accounts, runs, feeds, narratives, narrative_feeds, narrative_tickers, narrative_grades, tickers, prices, earnings, date_dim) plus Power BI compatibility views (dim_runs, fact_narratives, fact_narratives_ticker, fact_prices, fact_narrative_grades, …)

Reporting & BI:

• Self-contained HTML — RedHood Reads editorial card (Playfair Display + IBM Plex Mono design)
• Power BI export pipeline — Power Query (M) + DAX measures (powerbi/)

Trading Analysis (PowerShell):

• Yahoo Finance API (market data)
• Thermodynamic position-sizing: temperature, entropy, momentum, RSI

Deployment:

• Local execution (MVP)
• AWS Lambda + CloudWatch (planned)
• React dashboard (planned)

---

Roadmap

Phase 1: MVP (Complete)

• RSS feed aggregation (Substack)
• X/Twitter via Nitter RSS (no API key required)
• Claude AI narrative extraction
• Entropy risk scoring
• SQLite persistence (runs, feeds, narratives)
• Account management CLI (accounts_db.py)
• RedHood Reads HTML report (generated each run)
• Trading system analysis with thermodynamic sizing (run.ps1)
• .env support for API key management

Phase 2: Enhanced Analysis

• Thermodynamic regime detection + auto signal generation
• Ticker/side extraction into a star-schema bridge table
• Hypothesis grading (0–20 / A–F) + narrative scoring CLI
• Long-only P&L tracking ($2,500/ticker, via yfinance)
• Power BI export pipeline (Power Query + DAX)
• Historical backtesting of signal accuracy
• Sentiment trend tracking across runs
• Live ticker data in HTML report

Phase 3: Web Dashboard (Planned)

• React frontend
• FastAPI backend
• User authentication
• Trade journal UI
• Deployed demo (Vercel)

Phase 4: Scale (Post-MVP)

• Real-time alerts (Telegram bot)
• Mobile app (React Native)
• B2B features (team collaboration)
• API access for developers

---

Success Metrics

Product Metrics:

• Time saved: 2.5 hours → 30 min (83% reduction)
• Signal accuracy: 65% of flagged narratives = profitable trades
• User engagement: 5+ DAU with 70% weekly retention

Business Metrics:

• Target: 50 users (10 paid) in Month 1
• ARPU: $49/month
• Churn: <20% monthly
• LTV:CAC ratio: >3:1

---

Contributing

This is a portfolio project, but feedback is welcome!

How to provide feedback:

1. Open an issue with suggestions
2. Fork and submit a PR with improvements
3. Reach out directly: ctazeem@gmail.com

---

License

MIT License - feel free to use this code for your own projects.

---

About the Creator

Tazeem Chowdhury Scrum Master | Product Manager | Markets Analyst

• Background: Engineering degree with specialization in business analysis, data analytics, and enterprise service delivery. Currently pursuing CBAP and PMP certifications.
• Experience:
  • Project coordination and infrastructure delivery (Nav Canada, Mitel)
  • Business requirements gathering and solution design (RBC Capital Markets, IRCC)
  • Enterprise software implementation and QA (consulting engagements)
  • Cloud infrastructure and data visualization (Azure, Power BI)
  • Financial markets and cryptocurrency research and analysis
• LinkedIn: linkedin.com/in/tazeemchowdhury
• Twitter/X: @redhoodcapital
• Telegram: @redhoodtrades
• Email: ctazeem@gmail.com
• Substack: RedHood Reads

Why I Built This:

As a trader and market analyst, I was spending 3+ hours daily across Twitter, Substack, and financial feeds hunting for signals and synthesizing fragmented data. RedHood Systems automates the entire research pipeline — from feed aggregation to AI-extracted narratives to styled HTML briefings — demonstrating full-stack product thinking and technical execution.

---

Resources & References

AI & APIs:

• Anthropic Claude API
• Nitter — Twitter RSS proxy (no API key required)
• Telegram web preview — public channel scraping (no bot token required)
• Yahoo Finance API — market data for trading analysis (incl. $BLSH / Bullish)

---

Contact

Have questions about the project or want to discuss product opportunities?

Email: ctazeem@gmail.com LinkedIn: linkedin.com/in/tazeemchowdhury Newsletter: RedHood Reads on Substack Twitter/X: @redhoodcapital Telegram: @redhoodtrades

---

Last Updated: July 2, 2026 Version: 1.3 (Telegram source + $BLSH ticker)