УДК 339
DOI: 10.36871/ek.up.p.r.2024.06.03.032
Авторы
J. V. Torkunova,
Kazan State Power Engineering University, Kazan, Russia
V. Y. Ilichev,
Kaluga Branch of Bauman Moscow State Technical University, Kaluga, Russia
V. E. Drach,
Sochi State University, Sochi, Russia
F. L. Chubarov,
Russian State Agrarian University – Moscow Agricultural Academy named after K. A. Timiryazev, Kaluga, Russia
A. N. Paсukevich,
Sochi State University, Sochi, Russia
Аннотация
The work is devoted to evaluating pricing models for calculating the profitability of individual financial instruments, for example, such as stocks, using a multilayer generative-adversarial artificial neural network (GAN) and developing its own model based on the analysis. A huge amount of specially selected data is supplied to the inputs of the neural network, changing over time (dynamic). To improve the objectivity of the model, this work does not implement the arbitration capabilities of the markets. This is how one can analyze and explain variations and errors in pricing, as well as identify the key factors affecting asset prices. As the outcome, it was found out that the resulting asset pricing model exceeds all standard approaches for its assessment using the Sharpe ratio. It was shown that machine learning is an acceptable tool in term of investment forecasting, on the example of Russian Federation market.
Ключевые слова
asset, Russian Federation, machine learning, deep learning, pricing model, neural network, market arbitration, Sharpe ratio, stochastic discount factor
