I built a real-time sentiment tracker for WallStreetBets and trading subreddits because I wanted to start trading, but got frustrated manually scanning hundreds of comments to figure out what retail traders were actually bullish on.
Built this over my Christmas break as a solo developer.
Tech stack: Next.js 16, Supabase PostgreSQL, adaptive rate limiting for Reddit's API, OpenAI for sentiment analysis (users provide their own keys), Vercel for hosting.
Challenging parts:
• Reddit's rate limits (100 requests/min) required building a deduplication system and adaptive backoff
• Detecting sarcasm in WSB posts ("this stock is going to the moon " could mean anything)
• Real-time updates without killing the database (30-second refresh with caching)
• 9-layer sentiment scoring system (volume, options flow, cross-subreddit divergence, etc.)
Free tier available. Premium tiers add historical data, AI predictions, and portfolio alerts.
Open to technical feedback!
philipwhiuk•1h ago
> Detecting sarcasm in WSB posts ("this stock is going to the moon " could mean anything)
shawnmfarnum•1h ago
Built this over my Christmas break as a solo developer.
Tech stack: Next.js 16, Supabase PostgreSQL, adaptive rate limiting for Reddit's API, OpenAI for sentiment analysis (users provide their own keys), Vercel for hosting.
Challenging parts: • Reddit's rate limits (100 requests/min) required building a deduplication system and adaptive backoff • Detecting sarcasm in WSB posts ("this stock is going to the moon " could mean anything) • Real-time updates without killing the database (30-second refresh with caching) • 9-layer sentiment scoring system (volume, options flow, cross-subreddit divergence, etc.)
Free tier available. Premium tiers add historical data, AI predictions, and portfolio alerts.
Open to technical feedback!
philipwhiuk•1h ago
So how does it do this?