30 day update: I gave several AIs money to invest in the stock market

30-Day Progress Report: Testing AI-Driven Investment Strategies in the Stock Market

In a recent experiment, I allocated real funds to several artificial intelligence (AI) trading agents, allowing them to independently analyze live financial data and execute investment decisions in the stock market. This initiative aimed to explore the potential of AI-powered trading systems, particularly those focusing on swing trades and long-term investments, rather than short-term day trading.

Initial Hypothesis

My premise was that these AI models could outperform traditional benchmarks like the S&P 500 over a sustained period. Given their access to real-time financial data and their strategic allocation approach, I believed they might generate meaningful alpha after a month of operation.

One-Month Performance Overview

As we mark approximately 30 days into this experiment, I wanted to share an update on how these AI agents are performing. Over 100 community members followed the last update, and the early results are promising:

• Deepseek has achieved a 5% increase, outperforming the S&P 500, which has risen approximately 1% during the same period.
• Grok is up 4%, with GPT-based models showing a 3% to 1.15% gain.
• Both Claude models are also displaying positive returns within this timeframe.
• Conversely, Qwen and Gemini 2.5 have experienced slight declines.

Insights and Next Steps

While the preliminary data indicates that several AI models are successfully generating gains relative to the broader market, it’s important to emphasize that this is an initial snapshot. A single month is a limited timeframe to draw definitive conclusions, and sustained performance over a longer horizon is necessary to determine whether these AI agents can consistently produce alpha.

Moving forward, I plan to extend this experiment, monitor performance over additional months, and analyze factors such as risk, volatility, and consistency. If the trend persists, it could suggest that AI-driven investment strategies have a meaningful place in modern finance.

Additional Resources

For more details and ongoing updates, you can follow the experiment at Rallies.ai Arena.

Conclusion

These early results are encouraging and suggest that AI models, when given access to comprehensive real-time data, can potentially outperform traditional market benchmarks within a short span. However, patience and continued observation are essential to truly understand their capabilities and limitations in the complex world of stock market investing.