Published on April 24, 2026
The rise of prediction markets has captivated both regulators and the public. Platforms like Kalshi and Polymarket are gaining traction, inspiring many to claim they are striking it rich using AI models. This surge in interest is accompanied by a wave of social media buzz around potential profits.
However, new research from Arcada Labs challenges this notion. A study published on arXiv put six advanced AI models to the test, allocating each $10,000 to trade on prediction markets over 57 days. The results were sobering and unveiled a stark reality for those hoping for easy gains.
During the trial, every model sustained losses, ranging from 16% to 30.8% on Kalshi, while Polymarket saw less dramatic declines. Grace Li, co-founder of Arcada Labs, notes that the varied performance might stem from how freely models can trade. On Polymarket, models accessed a broader array of markets compared to the fixed options on Kalshi.
Despite disheartening losses, Li signals a silver lining. She believes that AI models may gradually improve, potentially surpassing human performance in trading decisions. Yet, the research underscores a crucial perspective: the true value lies not in profits alone, but in understanding the implications of advanced intelligence for society as a whole.
Related News
- How to Combat Rapid Battery Drain on Your Pixel Device
- DeepSeek Launches Advanced AI Model, Setting New Industry Standards
- Google Expands Desktop Presence with New Windows and MacOS Apps
- Alibaba's Happy Horse AI Emerges as Leader in Video Generation Technology
- AI Models Struggle with Soccer Betting Predictions
- Samsung Launches Micro RGB TVs, Revolutionizing Home Viewing Experience