RA-MAIP: A Continuous Risk-Scored SEBI-Compliant Robo-Advisory System for Multi-Asset Investment Planning in NSE
Abstract
In India, the retail investment market has changed fast in the last decade due to mobile-first applications, prevalent digital payment systems, and the consistent growth of investment advisory services registered by SEBI. Retail penetration of capital markets is reflected in the fact that the number of demat accounts has now topped 140 million by the end of 2023, and monthly inflows to mutual funds in SIPs have been above Rs.15,000 crore since early 2023. Although this has increased, the advisory layer that caters to these investors is structurally constrained. The current robo-advisory services put an investor into broad risk groups, usually conservative, balanced, and aggressive, and use pre-defined allocation templates to assign to each group. This practice does not represent the true spectrum of investor risk capacity, and does not meet the SEBI suitability requirement that all portfolio recommendations have a documented, quantifiable evaluation of the financial position and risk tolerance of the investor. This paper introduces RA-MAIP (Regulation-Aware Multi-Asset Investment Planner), a four layers system that fills these gaps in a single architecture. It uses six structured investor attributes to derive a normalised and continuous risk score, maps this score to asset weights in logistic and Gaussian allocation curve functions, validates each allocation with SEBI-inspired exposure limits in a deterministic compliance engine and reports quantitative performance metrics against Indian market indices such as expected annual return, annual volatility, Sharpe ratio, and Value at risk. An experiment using simulation on three representative investor profiles (conservative, balanced and aggressive) using simulation gave Sharpe ratios of 0.61, 0.89 and 1.14 respectively, which were better than a published robo-advisory baseline average of 0.54. Violation of all compliance tests in all profiles tested were identified and corrected in two rebalancing steps and the average cost of the Sharpe ratio of the constrained frontier-optimal allocation was 0.04 compared to the unconstrained frontier-optimal allocation. The architecture is planned to directly connect with live NSE data feeds and scale in the environments of Indian FinTech deployments.