A rule-based AI framework for portfolio optimization using investor profiling and asset allocation models

In this study, we introduce InvestiChat, a conversational framework that helps retail investors in India. The system creates an Investment Policy Statement (IPS) for each user and categorizes them as Conservative, Moderate, or Aggressive. We used real-time data from Yahoo Finance and Alpha Vantage, along with historical datasets, to build sample portfolios. Information of portfolio construction used Modern Portfolio Theory (MPT), and Monte Carlo simulations applied portfolio-per-user matching. Testing over an initial pilot study found evidence that the framework can represent a decrease in effort associated with portfolio selection. Barriers such as API rate limiting, and potential biases surfaced in the properties of classification processes.