Goal-Based Portfolio Diversification System

Abstract

This study introduces a Goal-Based Portfolio Diversification System, an AI-driven platform helping novice Indian investors build personalized portfolios. By merging rule-based logic with Large Language Models (LLMs), the system aligns investments with specific financial objectives through a three-tier architecture: a Goal Planner for risk assessment, a Portfolio Allocator driven by historical performance metrics, and an LLM module providing natural-language justifications. Utilizing AMFI and Yahoo Finance data, the platform generates recommendations across equity, debt, and gold while supporting Systematic Investment Plans (SIPs). Experimental results yield a portfolio efficiency score of 0.85 and an explanation fidelity of 0.312 (ROUGE-L), validating the system’s ability to mimic expert advisory. The modular design ensures scalability for advanced simulations, ultimately fostering financial literacy and disciplined investing without direct trade execution.