How Should Investors Evaluate AI Trading Agents? A New Study Has Answers
A May 2026 study from researchers affiliated with Pantera Capital, Stanford, IC3, and Ava Labs analyzed 1,900+ AI-tagged crypto projects to answer a practical question: how can investors tell a genuinely autonomous AI trading agent from a speculative wrapper? The research found even agents with fully public wallets couldn’t be verified as autonomous, and documents $191.7M in aggregate token holder losses across 925,323 wallets, with the top 1% of wallets capturing 81.4% of all gains ($1.81B). The paper’s proposed evaluation framework maps directly onto ChainAware’s Agent Trust Score, which screens both an AI agent’s wallet and its feeder wallet, giving investors a concrete way to check before they commit capital.