Mahsa Safavi Iranji; Majid Zanjirdar; Mojgan Safa; Hossein Jahangirnia
Abstract
Given the widespread increase in classical and emerging models for asset allocation in investment portfolios available in the capital market, investors find it challenging to easily compare classical methods and machine learning techniques to identify the optimal investment combination. ...
Read More
Given the widespread increase in classical and emerging models for asset allocation in investment portfolios available in the capital market, investors find it challenging to easily compare classical methods and machine learning techniques to identify the optimal investment combination. The aim of this research is to compare asset allocation based on the Nested Clustering Algorithm (NCO) with classical portfolios. This study has been conducted in a practical and descriptive-analytical manner, with the statistical population consisting of all companies listed on the Tehran Stock Exchange and the Iran Farabourse from 2013 to 2022. After screening, adjusted daily data from 88 companies were selected as the final sample for statistical analysis. In this context, the Kruskal-Wallis test was used to examine the hypotheses, and Python, SPSS, and Excel software were utilized. Based on the overall performance evaluation criteria for portfolios (Sharpe ratio, Sortino ratio, maximum drawdown, value at risk, and expected shortfall), the results of the hypothesis tests in this research indicate that the methods based on the Nested Clustering Optimization Algorithm outperform their classical counterparts significantly. Therefore, it can be concluded that portfolios based on machine learning algorithms perform better than classical portfolios.
Zahra Pourahmadi; Dariush Farid; Hamid Reza Mirzaei
Abstract
Stock trading is a significant decision-making problem in asset management. This study introduces a financial trading system (FTS) that leverages artificial intelligence (AI) techniques to automate buy and sell orders specifically in Iran's stock market. Due to limited availability of labeled data in ...
Read More
Stock trading is a significant decision-making problem in asset management. This study introduces a financial trading system (FTS) that leverages artificial intelligence (AI) techniques to automate buy and sell orders specifically in Iran's stock market. Due to limited availability of labeled data in financial markets, the FTS utilizes reinforcement learning (RL), a subset of AI, for training. The model incorporates technical analysis and a constrained policy to enhance decision-making capabilities. The proposed algorithm is applied to the Tehran Securities Exchange, evaluating its efficiency across 45 periods using three different stock market indices. Performance comparisons are made against common strategies such as buy and hold, randomly selected actions, and maintaining the initial stock portfolio, with and without transaction costs. The results indicate that the FTS outperforms these methods, exhibiting excellent performance metrics including Sharp ratio, PP, PF, and MDD. Consequently, the findings suggest that the FTS serves as a valuable asset management tool in the Iranian financial market.
Maryam Moradi; Najme Neshat; Amir Mohammad Ahmadzade Semeskande
Abstract
Safe investment can be experienced by incorporating human experience and modern predicting science. Artificial Intelligence (AI) plays a vital role in reducing errors in this winning layout. This study aims at performance analysis of Deep Learning (DL) and Machine Learning (ML) methods in modellingand ...
Read More
Safe investment can be experienced by incorporating human experience and modern predicting science. Artificial Intelligence (AI) plays a vital role in reducing errors in this winning layout. This study aims at performance analysis of Deep Learning (DL) and Machine Learning (ML) methods in modellingand predicting the stock returns time series based on the return rate of previous periods and a set of exogenous variables. The data used includes the weekly data of the stock return index of 200 companies included in the Tehran Stock Exchange market from 2016 to 2021. Two Long Short-Term Memory (LSTM)and Deep Q-Network (DQN) models as DL processes and two Random Forest (RF) and Support Vector Machine (SVM) models as ML algorithms were selected. The results showed the superiority of DLalgorithms over ML, which can indicate the existence of strong dependence patterns in these time series, as well as relatively complex nonlinear relationships with uncertainty between the determinant variables. Meanwhile, LSTM with R-squared equals to 87 percent and the analysis of the results of five other evaluation models have shown the highest accuracy and the least error of prediction. On the other hand, the RF model results in the least prediction accuracy by including the highest amount of error.
