TY  - JOUR
TI  - Software application for investment portfolio structure management
T2  - 
IS  - 
KW  - Investment portfolio
KW  - genetic algorithms
KW  - neural networks
KW  - modeling
AB  - Alexey Isavnin - Professor, Department of Mathematical Methods in Economics, Division of Economics, Kazan Federal University, branch in Naberezhnye Chelny.Address: 1 (5/10), Stroitelei bul., Naberezhnye Chelny, Tatarstan, 423819, Russian Federation.E-mail: isavnin@mail.ruDamir Galiev - Student, Department of Mathematical Methods in Economics, Division of Economics, Kazan Federal University, branch in Naberezhnye Chelny.Address: 1 (5/10), Stroitelei bul., Naberezhnye Chelny, Tatarstan, 423819, Russian Federation.E-mail: damir.galiev@mail.ruForming an effective investment profile is one of the most common tasks in financial sphere. The quantitative method allows raising the quality of investment profile. We understand quality as is the required ratio between risk and yield of investment profile. The majority of today’s analytical programs does not have the opportunity to design advanced quantitative models of optimal investment profile choice.This paper describes mathematical models and algorithms, as well as logics and architecture of computer software for forming an effective structure of investment profile. The feature of the approach is using the most practicable models and designing a friendly interface for solving practical problems in the financial sphere. The developed system is rather flexible: it allows running a wide class of models for choosing an effective investment profile, applying prediction models for time series of returns on assets, and considering outside experts’ reviews. The key feature of the system is an opportunity to cooperate with stock market sales terminal.The starting point of the system’s logic is the concept of optimizing "risk-return on asset." The architecture of the computer software uses two types of prediction models of the time series of asset returns: autoregressive moving-average model (ARMA) and a two-layered neural network model for forecasting time series. Module of genetic algorithms is also used to solve optimization problems with non-differentiable objective functions and limitations.Experiments were performed based the data of the Russian stock market (the Moscow stock exchange). The results of the experiments show the viability of the proposed models and the possibility of their application in practice for a wide range of economic agents, from individual investors to large funds (including pension). Further development of the research is to make and add new models and to improve the interface of computer software
AU  - А. Isavnin
AU  - D. Galiev
UR  - https://bijournal.hse.ru/en/2012--3(21)/65565361.html
PY  - 2012
SP  - 52-62
VL  -