ai
AI Models Show Mixed Results in Live Trading Competitions, Some Profitable
Bloomberg reported on May 6 that tests of AI models from OpenAI, Anthropic, Google and xAI produced underwhelming results with many losing money. Nof1's Alpha Arena gave eight models $10,000 each to trade U.S. tech stocks over two weeks. Across four competitions only six of 32 outcomes ended in profit.
forbes.comAudio version
Tap play to generate a narrated version.
Developing·Limited corroboration so far. This page will refresh as more sources emerge.
Bloomberg published an article on 2026-05-06 detailing how AI bots auditioning for Wall Street trading are mostly losing money. Tests involving models from OpenAI, Anthropic, Google, and xAI delivered underwhelming results. In the tests many AI models lost money, traded excessively, and made erratic decisions despite receiving identical prompts.
The models appeared unable to stick to coherent strategies for more than a few trading sessions. One of the clearest examples came from Alpha Arena, a competition created by Nof1. U.S.
Tech stocks over a two-week period using different strategies including defensive approaches and leveraged bets. Across four Alpha Arena competitions the models collectively lost roughly a third of their capital. Only six of 32 outcomes ended in profit.
20 made 158 trades in one contest while Alibaba Group’s Qwen executed 1,418 trades under the exact same prompt. ZeroHedge reported that the experiments reflect growing interest in whether generative AI can eventually outperform traditional fund managers.
Wall Street firms including JPMorgan Chase and Balyasny Asset Management already rely on AI for research, fraud detection, and internal analysis.
The firms have largely stopped short of handing over actual investment decisions to AI. ” Research blog Flat Circle tracked 11 public AI trading competitions. In those 11 public AI trading competitions every event produced at least one profitable model but only two generated a profitable median return.
Some firms continue to bet on improvement with better tools and tighter guardrails. Intelligent Alpha runs an AI-driven fund that combines LLMs with earnings transcripts, analyst forecasts, corporate filings, macroeconomic indicators, and web searches. In late 2025 OpenAI’s ChatGPT correctly predicted the direction of earnings estimate revisions 68% of the time according to Intelligent Alpha.
