Developer sstklen released trump-code on GitHub on March 15, 2026, an open-source AI system that analyzes former President Trump's social media posts to predict stock market movements. The project claims a 61.3% hit rate after testing 31.5 million model variations, garnering 329 GitHub stars within approximately 24 hours of release.
The system monitors Trump's posts in real-time across both Truth Social and X (formerly Twitter), analyzing content, timing, sentiment, and keywords to identify correlations with subsequent market movements. The claimed 61.3% accuracy rate exceeds the 50% baseline of random prediction, though whether this performance generalizes beyond backtesting remains unvalidated by independent community testing.
Multi-Platform Monitoring and Arbitrage Capabilities
Trump-code tracks posts across Trump's primary communication channels simultaneously, capturing potential information asymmetries when content or timing differs between platforms. The system includes:
- Real-time post monitoring on Truth Social and X
- Stock market correlation analysis
- Polymarket API integration for prediction market data
- Cross-platform arbitrage detection between different prediction markets
- Multilingual support for Chinese, English, and Japanese markets
The Polymarket integration enables the system to identify price discrepancies between prediction markets, potentially allowing users to exploit arbitrage opportunities on Trump-related political and market events.
Technical Approach and Overfitting Concerns
The repository's "brute-force" tag alongside the "31.5M models" claim suggests the system tested millions of model variations to find patterns correlating post characteristics with market movements. This exhaustive search approach raises standard concerns about overfitting—where a model performs well on historical data but fails to generalize to future predictions.
The project's tags include ai, machine-learning, brute-force, open-data, prediction, signal-analysis, stock-market, trump, truth-social, and polymarket. The multilingual support reflects international interest in Trump's market impact, particularly relevant for Chinese markets responding to trade policy signals and Japanese markets reacting to currency and Asia policy comments.
Market Context and Regulatory Considerations
The release timing coincides with the 2026 election cycle, during which Trump actively campaigns for midterms and potentially the 2028 presidential race. Trump's posts have historically moved markets on topics including trade policy, company-specific comments, and geopolitical positions, creating opportunities for algorithmic trading systems.
Automated systems trading on public figures' social media posts exist in uncertain regulatory territory and raise several concerns:
- Market manipulation through amplified reactions to posts
- Front-running capabilities via faster-than-human analysis
- Potential reduction in effectiveness as more traders adopt similar signals
- Ethical questions about democratizing versus concentrating trading advantages
The developer's decision to open-source rather than commercialize a potentially profitable system enables community validation of claimed performance metrics while democratizing access to sophisticated market prediction tools.
Key Takeaways
- Trump-code analyzes Trump's social media posts across Truth Social and X to predict stock market movements, claiming 61.3% accuracy after testing 31.5 million models
- The open-source system gained 329 GitHub stars within 24 hours of its March 15, 2026 release
- The project includes Polymarket integration for prediction market arbitrage and supports Chinese, English, and Japanese languages
- The claimed 61.3% hit rate exceeds random prediction baselines but may suffer from overfitting given the extensive model testing approach
- Automated trading on political figures' social media posts raises regulatory and ethical questions about market manipulation and front-running