JPMorgan tests AI agents that can shift stock-bond allocations

JPMorgan is testing AI investment agents that can change the mix between stocks and bonds as market conditions shift. According to BlockBeats, the bank said the best-performing model in a 20-year historical backtest delivered an annualized return 0.7 percentage points above a traditional 60/40 stock-bond portfolio, while also showing lower volatility. JPMorgan added that all eight AI agents in its test produced better risk-adjusted returns. The bank also drew a clear line around the results. It said the findings came from simulated testing rather than live investment performance, meaning they should not be read as evidence of real-world returns. JPMorgan also warned that if AI is adopted on a large scale across markets, trading strategies could become more alike, which may increase crowding and amplify volatility during periods of market stress. The test points to potential gains in portfolio management, but the bank’s own caveat makes clear that the current evidence remains limited to backtested scenarios.