Tag
A study from Northwestern shows a 3-state Hidden Markov Model that detects market regimes to outperform traditional factor investing in S&P 500 trading, delivering 2% annual alpha and avoiding major crashes.
The author builds a stripped-down three-agent version of a 50-agent AI portfolio system described in a BlackRock paper, using Claude. The system produces a surprising regime analysis and adversarial critique that leads to a real portfolio adjustment.