With the rapid development of the new energy industry driven by the ‘Dual Carbon’ policy, how to identify high-quality stocks amidst industry fluctuations has become an important topic in the field of quantitative investment. This paper focuses on the investment characteristics of the new energy industry and integrates modern quantitative financial theories and software engineering methods to build a multi-factor stock selection system driven by microservices architecture and machine learning components. At the strategy level, this paper centers on industry characteristic factors and constructs a multi-factor system covering dimensions such as growth potential, valuation, policy sensitivity, and technological trends. It also utilizes methods like Information Coefficient (IC) and Principal Component Analysis (PCA) for factor screening and combination. To address the shortcomings of traditional linear scoring methods in identifying nonlinear relationships, this paper introduces a Support Vector Machine (SVM) model to predict stock excess returns, achieving deep matching between factors and returns. At the system architecture level, this paper designs and implements a stock selection system based on the Spring Cloud microservices architecture, which includes multiple independent service modules such as data processing, factor computation, model training, strategy execution, and backtesting evaluation. These modules collaborate efficiently through service registration and remote invocation mechanisms, supporting containerized deployment and dynamic scaling, significantly improving the system’s maintainability and response performance. Empirical results show that the system exhibits good annualized return performance and risk control ability in backtesting strategies for the new energy industry, with low system invocation latency and strong model plug-in capability, confirming its practical application value and engineering feasibility in quantitative investment.