Análisis del comportamiento de clientes del mercado de valores mediante machine learning e inteligencia artificial

Abstract

CONDITION FOR PUBLICATION OF PROJECT. In the Ecuadorian stock market, brokerage firms face the challenge of tailoring their business strategies to specific customer segments based on information supported by their historical trading records. However, investment proposals are mainly based on the experience of the trader or commercial advisor, which introduces a high degree of subjectivity. One of the main limitations is the absence of advanced tools capable of transforming large volumes of data into useful knowledge, which restricts the design of effective strategies and reduces the competitiveness of these institutions. This project aims to analyze customer behavior using machine learning techniques, integrating traditional variables to segment and identify customer profiles. To this end, historical data from the period 2014–2024 was used and supervised (Random Forest) and unsupervised (K-Means) models were applied, complemented by external validation through financial advisors. The results show that factors such as the amount to be invested, the effective value, and the total value of the transaction are more decisive than conventional technical variables, allowing for more accurate segmentation and the design of differentiated commercial strategies. In conclusion, the project proposes an innovative tool that optimizes customer management in securities firms, promotes strategic decision-making, and strengthens competitive position in the stock market. Keywords: Stock Market, Machine Learning, Customer Segmentation