Methods for predicting securities prices by means of a trading platform for portfolio investment

Бесплатный доступ

As a result of the research, it was found that at the moment there are many trading platforms on the Russian software market, but they do not have the functionality to predict securities quotes on financial exchanges. This platform includes: Meta Trader (MT 4, MT5), Act Trader, Quick, TransaQ, Alfadirekt, Rikom Trade, Mirror Trader, Ninja Tracer, cTrader, Rox. On the other hand, there are platforms that allow you to predict the stock price for a short period (no more than 5 days), such platforms include: IKnow First, Stock Carts, Stock Neural. This article aims to analyze the existing methods and algorithms that are used to develop these trading platforms in order to improve their efficiency in predicting securities prices for portfolio investment. The authors of the article considered the following methods: genetic algorithms, Markov random decision making, GDQN and GDPG machine learning methods, naive Bayesian algorithm combined with Adaboost algorithm, semantic analysis of financial news, Decision Stump, simple linear regression, support vector machine using radial basis kernel functions.


Trading platform, stock price, course, methods, algorithms, machine learning, neural networks

Короткий адрес:

IDR: 148322465   |   DOI: 10.25586/RNU.V9187.21.03.P.119

Статья научная