Stock trading has a more significant influence on the global economy. Stock trading with portfolio optimization became challenging due to the complexity of analyzing the high variance in time series stock data. Efficient portfolio management increases profit and avoids risky situations when investing. The present work aims to model a Variable Neighborhood Search Multi-Agent System for Portfolio Optimization (VNSMASPPO) to optimize the profit on defined trading constraints on buying, selling, and holding trading decisions. This work proposes a novel Variable Neighborhood Search-based Multi-Agent System (VNASMAS) algorithm for profit computation with a constraint-based multi-agent system. The stock price history experimental data sets are collected from 8th August 2016 to 31st March 2023 with 14,567 records. The proposed model achieved an RMSE of 10.11, MAE of 2.75, and MAPE of 0.017, outperforming the literature models. VNSMASPPO maximizes the portfolio profit and is a reliable, adaptable approach.