Social Signal Harvester | James Murray

James Murray has developed the Social Signal Harvester, a tool that collects and analyzes social media signals from platforms like Twitter and Reddit to detect trend momentum. By leveraging sentiment analysis and social signals, the system can forecast potential market trends and shifts.

The harvester processes 500k+ posts/day, assigns sentiment scores, and detects breakout keywords 4-6 hours before price impact. It uses VADER + fine-tuned BERT for crypto slang.

Key Features

  • Sentiment Analysis: VADER + BERT with 91% accuracy on crypto tweets.
  • Trend Momentum Detection: Velocity and acceleration of keyword volume.
  • Cross-Platform Coverage: Twitter, Reddit, Telegram, Discord.
  • Breakout Alerts: Push when volume >5x 7-day average.
  • Influencer Tracking: Top 100 accounts by engagement.
  • API + Webhook: Real-time signal stream.

System Design & Architecture

The system collects data from social media APIs, applies sentiment analysis, and forecasts momentum using time-series models. Results feed into trading dashboards.

Technical Stack

  • Scraping: Twitter API v2 + PRAW
  • NLP: Hugging Face Transformers
  • Storage: Elasticsearch

Related Projects

Explore other related projects: