Saad Ullah
⬤ Eight AI models got handed $100,000 each in late November to trade freely in real markets. The results? Claude Sonnet 4.5 crushed the competition. It's the only model that hit double-digit gains, climbing to 10.4% while the S&P 500 limped along at 2.7%. The experiment, shows Claude's portfolio now sitting at roughly $110,400 compared to the benchmark's $102,700.
⬤ The gap kept widening through January as Claude pulled ahead of every other AI in the race. Grok 4 came in second with 9.1%, followed by Opus 4.5 at 4.5%, DeepSeek V3 at 4.0%, Gemini 2.5 Pro at 3.8%, and GPT 5.2 trailing at just under 3.0%. While most models moved sideways or inched forward slowly, Claude's equity curve shot upward, especially after the new year hit.
⬤ You can watch the whole thing play out live through the Arena section on Rallies' platform. The experiment doesn't reveal specific trades, but the numbers tell the story—Claude's been consistently climbing while others stayed flat. From December through late January, it navigated typical market noise and shifting trends better than its AI peers and the broader market.
⬤ This matters because Wall Street's paying attention. As more traders explore AI-driven strategies, this experiment gives us real data on how different models actually perform when money's on the line. Claude didn't just beat other chatbots—it beat the market. That's making people rethink what AI can do beyond writing emails and answering questions.
Saad Ullah
