Integrated digital model of automated investment portfolio management: Economic efficiency and risks

E. Koroleva, Aleksandr Yu. Gorelik

Abstract

Subject. A comprehensive digital model of automated investment portfolio management. Objectives. Development of a comprehensive digital model of automated investment portfolio management and empirical verification of its effectiveness in various portfolio optimization settings. Methods. The research is based on the provisions of the classical portfolio theory of G. Markowitz and W. Sharpe. The software implementation is implemented in Python and is organized as a modular architecture of five functional blocks (parameter setting, data loading, iterative portfolio calculation, efficiency calculation and visualization, formatting of output tables). As part of the testing, two algorithms for finding optimal weights were compared: maximizing the Sharpe coefficient and maximizing the quadratic utility, as well as an analysis of sensitivity to the parameters of the retrospective analysis window and the frequency of revision. Results. It has been found that maximizing the Sharpe coefficient provides a higher and more stable trajectory of profitability compared to maximizing quadratic utility; maximizing quadratic utility is characterized by greater variability in the result and more pronounced fluctuations in profitability; reducing the parameters of the window of retrospective analysis and the frequency of revision worsens the final efficiency and increases the volatility of the result, with the negative effect being more pronounced for maximizing quadratic utility; accounting for transaction costs and rules for fixing/reinvesting gains significantly affects the efficiency assessment and increases the applied validity of the model. Conclusions. The scientific novelty of the work lies in the development of a reproducible digital model with a modular structure that combines portfolio optimization and practice-oriented efficiency calculation in a single computing circuit. The practical significance lies in the possibility of using the model for scenario analysis of strategy parameters and comparing the results with a market benchmark.

Source: semanticscholar

Read the AI summary, key takeaways and discussion on WOBR Quant Research.