Open-source LLM trading bots vs. the government bond, on paper money (blueprintau.com)

In a bold experiment launching on November 1, 2023, a team is deploying 100 open-source AI trading bots, each with A$63,000 in simulated capital, to navigate the volatile world of trading by attempting to "catch falling knives." This strategy entails buying during market downturns and selling during recoveries, with each bot guided by an AI model that evaluates potential trades. Notably, these bots will operate without stop losses and can leverage investments up to 40 times, meaning a small market shift could wipe them out entirely. The competition culminates in comparing their performance against the stagnant yield of Australian 10-year government bonds, providing an intriguing benchmark for their success. This project is significant for the AI/ML community as it pushes the boundaries of how AI can be utilized in real-time decision-making in finance, showcasing both experimental strategies and the resilience of AI models under high-risk conditions. By tracking the trading bots' performance—where a mere 80 out of 100 are expected to fail before reaching the bond benchmark—the initiative aims to illuminate the effectiveness of AI in generating returns compared to a risk-free asset. The weekly reports on these bots will offer valuable insights into their strategies' viability, ultimately contributing to a broader understanding of AI's capabilities in challenging financial environments.