MomentumFlow AI

Real-time AI-Powered Trading Signals for S&P 500 Momentum Strategies

🏆 Confluent AI Partner Catalyst Hackathon 2025

---

📖 Overview

MomentumFlow AI is a real-time momentum trading scanner that leverages Confluent Kafka streaming to deliver institutional-grade market analysis. Built for the S&P 500, it combines proven momentum trading strategies with live data streaming and AI-powered pattern recognition.

Key Features

• 📊 Stock Scanner — Scans 534 S&P 500 stocks against 5 momentum criteria
• 📈 Trend Indicator — EMA-based trend detection (Bullish/Bearish/Neutral)
• 🚩 Bull Flag Detection — Automated pattern recognition with adjustable thresholds
• 📰 Live News Feed — Google News streaming with sentiment analysis
• 🧮 Risk Calculator — Position sizing with scaling protocols
• 💡 Trade Ideas — Auto-generated setups based on score + patterns

---

🏗️ Architecture

┌──────────────────┐     ┌─────────────────────┐     ┌──────────────────┐
│  Yahoo Finance   │     │   Confluent Kafka   │     │    R Shiny       │
│  Google News     │────▶│   (Cloud - Zurich)  │────▶│    Dashboard     │
└──────────────────┘     │                     │     └──────────────────┘
                         │  • stock_prices     │
   Python Producers      │  • stock_news       │      REST API Consumer
                         └─────────────────────┘

Kafka Topics

Topic	Content	Update Frequency
stock_prices	OHLCV + Momentum Scores	~2 min cycle
stock_news	Headlines + Sentiment	~5 min cycle

---

📁 Project Structure

MomentumFlowAI/
├── app_dark_theme.R          # Main Shiny application
├── kafka_producer.py         # Stock price streaming producer
├── kafka_news_producer.py    # News streaming producer
├── README.md                 # This file
├── LICENSE                   # MIT License
└── docs/
    └── MomentumFlowAI_Hackathon_Submission.pdf

---

🚀 Quick Start

Prerequisites

• R 4.x with RStudio
• Python 3.x
• Confluent Cloud account (free tier works)

1. Clone the Repository

git clone https://github.com/Tianning-lab/MomentumFlowAI.git
cd MomentumFlowAI

2. Install R Dependencies

install.packages(c(
  "shiny", "shinydashboard", "quantmod", "tidyverse", 
  "plotly", "DT", "TTR", "lubridate", "scales",
  "PerformanceAnalytics", "httr", "jsonlite", "base64enc", "xml2"
))

3. Install Python Dependencies

pip install confluent-kafka yfinance pandas numpy feedparser requests

4. Configure Confluent Kafka

1. Create a Confluent Cloud account at confluent.cloud
2. Create a cluster (Basic tier is fine)
3. Create topics: stock_prices and stock_news
4. Generate API keys

Update credentials in both Python files:

KAFKA_CONFIG = {
    'bootstrap.servers': 'YOUR_BOOTSTRAP_SERVER',
    'sasl.username': 'YOUR_API_KEY',
    'sasl.password': 'YOUR_API_SECRET',
    ...
}

5. Start the Producers

Terminal 1 — Stock Prices:

python kafka_producer.py

Terminal 2 — News:

python kafka_news_producer.py

6. Run the Shiny App

In RStudio, open app_dark_theme.R and click Run App.

---

📊 Momentum Criteria (Adapted for S&P 500)

#	Criterion	Threshold	Rationale
1	Price Range	$5 - $500	Covers most S&P 500 stocks
2	Daily Change	≥ 0.5%	Meaningful price movement
3	Relative Volume	≥ 1.0x average	Confirms participation
4	Momentum	Price > Open	Intraday buying pressure
5	Volatility	Range ≥ 0.5%	Sufficient profit opportunity

---

🚩 Bull Flag Detection

Thresholds adapted for large-cap stocks:

Parameter	S&P 500	Penny Stocks
Pole Return	≥ 1.5%	≥ 3%
Flag Range	< 7%	< 5%
Volume	Decreasing	Decreasing

---

🛠️ Technology Stack

Layer	Technology
Streaming	Confluent Cloud (Apache Kafka)
Producers	Python, confluent-kafka, feedparser
Data Sources	Yahoo Finance, Google News RSS
Frontend	R Shiny, shinydashboard
Visualization	Plotly, DT
Analysis	quantmod, TTR, PerformanceAnalytics

---

📸 Screenshots

Landing Page

Stock Scanner with Trend Indicator

Bull Flag Detection

Live News Feed

---

🗺️ Roadmap

• Q1 2026: Options flow analysis via Kafka Streams
• Q2 2026: Claude AI integration for natural language queries
• Q3 2026: Multi-asset support (crypto, forex)
• Q4 2026: Mobile app with push notifications

---

👨‍💻 Author

Tianning Ning
Founder & CEO, New Way Capital Advisory Limited

• 🐙 GitHub: @Tianning-lab
• 📧 Email: tnk@newwaycapital.com
• 📍 Geneva, Switzerland

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

🙏 Acknowledgments

• Confluent for the amazing Kafka platform
• Yahoo Finance for market data
• Google News for news feeds

---

© 2025 New Way Capital Advisory Limited
Made with ❤️ in Geneva, Switzerland

---

Live Demo

MomentumFlow AI is live at nwc-advisory.com/app — part of the New Way Capital Advisory analytics platform.

Related Tools

• Portfolio Analyser — Institutional portfolio analysis
• Risk API — 7-signal market risk scoring
• Property Comps — Comparable sales across 11 markets
• Portfolio X-Ray — Free fund look-through analysis