| چکیده انگلیسی مقاله |
Objective
Knowing the behavior of capital markets and their orientation is a foreground for analyzing the behavior of return during the events governing the society. Current political, economic, and social issues can simply influence the parameters of the economic cycle. The stock market as a significant part of the economy is not an exception. Precisely predicting and recognizing fluctuations increases investors’ confidence and leads to accurate and timely asset management decisions. Also, knowing the most efficient tools for predicting return is vital for analyzing the behavior of the market. The present research aims to cluster current companies in the stock market based on their susceptibility to sanctions using the best method, artificial intelligence, for forecasting stock returns.
Methods
This research study utilizes weekly return data from 200 active companies, along with information on variables such as industry type, size, liquidity, and profitability of the selected companies from 2016 to 2021. Additionally, the study incorporates data on key political, economic, and social events during this period. In the first step, four of the best models of deep learning and machine learning including LSTM (Long-Short Term Memory), DQN (Deep Q Network), RF (Random Forest), and SVR (Support Vector Machines) were compared. Next, the prediction about stock return was made by applying the most superior model. In the second step, scenarios were developed based on the susceptibility of return changes to each input variable, including industry type, size, liquidity, and profitability. These scenarios were then analyzed to assess their sensitivity. In the final step, applying partitional clustering, the results were categorized into three clusters: economic, political-economic, and economic-social, and the findings were analyzed afterward.
Results
Comparing deep learning (LSTM, DQN) and machine learning models (SVR, RF) revealed that LSTM is the superior model for predicting stock return. Also, the results from clustering provided a broad range of analyses based on the needs of investors. The results could be used as a basis for analyzing changes in stock return rate on facing issues. Generally, political events have the most significant effects on the stock return of companies followed by economic events. Finally, social events are the least effective factors. In terms of criteria, company size, type, liquidity, and profitability were effective factors in fluctuations, respectively.
Conclusion
Iran’s stock market is affected by political, economic, and social news as well as government actions and statements; the susceptibility depends on the type of news. The events directly impact stock returns, and it is established that during political, economic, and social events, the stock market return of companies fluctuates based on their type, size, liquidity, and profitability. Political events exert the most significant influence on stock returns and warrant particular attention from capital market participants. |