Examining the effectiveness of machine learning models in predicting the type of price hits with the price limit of Tehran Exchange Securities stocks with an emphasis on the model Histogram Gradient Boosting
نویسندگان
1 Department of System Management and Productivity, Tarbiat Modares University, Tehran, Iran
2 Department of Management, Mashhad Branch, Islamic Azad University, Mashhad, Iran/
doi
10.30495/ijfma.2024.77856.2157چکیده
This study examines the prediction of the type of stock prices and the price limit using End-Of-Day information in the Tehran Stock Exchange market. By using the information of 65 stock during the period of February 2021 to February 2023 and by using machine learning models, the type of hits has been investigated in terms of prices. By using some evaluation criteria, the performance of each model has been investigated. When the price hits the limit during the day and does not continue until the end of the day, and the price of the last transaction of the stock is not equal to the price of the price limit, non continuous price limit and when the price of the last transaction is equal to the price of about, the transaction is continuous price limit it will happen. The best performance was related to the Histogram Gradient Boost-ing model in February 2023 with an accuracy rate of 90 Percent. As a result, according to the level of accuracy in predicting the type of hits, it is possible to use the results of this study in the type of stock buying and selling strategy and the possibility of forming buying and selling queues
