Publications
Detailed Information
Predicting microeconomic factors using machine learning and applications in optimal asset allocation : 기계학습을 활용한 미시경제 변수 예측 및 최적자산배분에서의 응용
Cited 0 time in
Web of Science
Cited 0 time in Scopus
- Authors
- Advisor
- 이재욱
- Major
- 공과대학 산업공학과
- Issue Date
- 2018-08
- Publisher
- 서울대학교 대학원
- Description
- 학위논문 (박사)-- 서울대학교 대학원 : 공과대학 산업공학과, 2018. 8. 이재욱.
- Abstract
- The market participants predict and interpret market views in accordance with their interests and set various strategies to cope with them. Participants who determine important policies in financial markets will implement various policies to stabilize the markets based on macroeconomic & microeconomic variables. In this process, market regulation can be strengthened or mitigated depending on the purpose and investors modify their investment strategy to maximize their profits in response to the policies. Predicting future market views is the most important key to investors, and the value of their assets can be determined by their predictability. If the predicted market view is in line with the actual market trends, it is possible to maximize profits, but the market moves against the prediction, it must endure a huge loss. Therefore, predicting market views is an important factor for all market participants and many researches are actively conducted to improve the predictability.
Financial-related variables and corporate defaults are among the most important components of microeconomic factors. Financial-related variable forecasts can be stock prices and its volatilities and firm-related variables in financial statements. Financial-related variables can be easily observed in the market, and studies are being actively conducted to explain not only the predictions but also the relationships among the variables. Predicting financial variables is a key element in investors, and it is necessary to predict variables related to invested assets for the successful investment. Corporate bankruptcy forecasts are highly correlated with financial variables and can be the most important factor for investors who invest in the firm. Although macroeconomic variables are important, the variables in the financial statements indicating to the firm's financial soundness may play a major role in predicting bankruptcy. The bankruptcy of a company can have a significant impact on financial variables such as the stock price of the company and the loss of related companies should also be considered. Therefore, it is necessary to analyze firms accurately and develop an early warning system is required.
It is also necessary for investors to construct an optimal portfolio using predicted microeconomic factors. If the portfolio is not constructed effectively, even if the factors are accurately predicted, it may not earn the expected profits. Therefore, it is important to construct a portfolio that reflects the factors exactly, and it should be able to demonstrate excellence by comparing it with the market portfolio.
The aim of this dissertation is to predict microeconomic variables which are market index, stock volatility and corporate defaults and to construct a portfolio that reflects them. To do so, (i) verifying the existing microeconomic factor prediction models, (ii) improving the predictability of the factors, and (iii) constructing an optimal portfolio that reflects the predicted factors are needed. In addition, corporate bankruptcy forecasts, one of the forecasts of microeconomic variables, will be discussed. A bankruptcy prediction model will be proposed and an early warning system that can explain the systemic risk of the industry in advance will be developed.
Prediction research of microeconomic factors can be used effectively in practice. It will be also valuable as a future study and can be the basis of investment profits. In particular, corporate bankruptcy prediction is expected to predict the systemic risks of industries in advance, and prevent and counteract disasters such as the 2008 financial crisis. A portfolio that reflects the factors can be simplified through an automated system in practice. As the degree of certainty of the predicted factor can be reflected in the portfolio composition, it is possible to construct an optimal portfolio that is not overly biased toward the investors' view and in balance with the market situation. Therefore, it is expected that investors will be able to choose the optimal investment strategy that can generate excellent profit in the financial market by selecting a portfolio considering their preferences.
- Language
- English
- Files in This Item:
Item View & Download Count
Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.