Machine Learning Algorithms for Predicting the Returns of IPO Companies
1
조회
0
다운로드
2
인용
0
참고문헌
키워드
초록
Initial public offering (IPO) refers to the recruitment or sale of shares of a company to general investors to list the shares on the stock exchange. Determining the offer price is a type of corporate valuation. It has been consistently highlighted that the offer price does not properly reflect the corporate value in the Korean domestic IPO market. In most studies, the initial yield of IPO shares is around 20% to 30% on average but it is significantly affected by extreme values with high deviations. The wide variation in the yield of individual public offer shares makes it difficult to predict their stock price. Therefore, this study contributes to investors’ investment decisions by introducing machine learning algorithms to predict the closing price and closing price returns over the initial price of the listing. This study analyzes 38 communications and 1,165 IPO cases in the KOSPI and KOSDAQ from July 2004 to May 2021, provided by Korea Exchange. This empirical study shows that Random Forest, XGBoost, and Deep Neural Networks have a higher predictive accuracy than con- ventional statistical methods.
Abstract
Initial public offering (IPO) refers to the recruitment or sale of shares of a company to general investors to list the shares on the stock exchange. Determining the offer price is a type of corporate valuation. It has been consistently highlighted that the offer price does not properly reflect the corporate value in the Korean domestic IPO market. In most studies, the initial yield of IPO shares is around 20% to 30% on average but it is significantly affected by extreme values with high deviations. The wide variation in the yield of individual public offer shares makes it difficult to predict their stock price. Therefore, this study contributes to investors’ investment decisions by introducing machine learning algorithms to predict the closing price and closing price returns over the initial price of the listing. This study analyzes 38 communications and 1,165 IPO cases in the KOSPI and KOSDAQ from July 2004 to May 2021, provided by Korea Exchange. This empirical study shows that Random Forest, XGBoost, and Deep Neural Networks have a higher predictive accuracy than con- ventional statistical methods.
참고문헌 (10)
- 전진규, "IPO 저가발행이 이후공모에 미치는 영향", 한국증권학회지, 48(5), pp.589~616 (2019), 한국증권학회[DOI]
- 김주환, "공개정보와 IPO 주가성과", 재무관리연구, 36(2), pp.1~30 (2019), 한국재무관리학회[DOI]
- 이동원, "KOSPI200 Prediction through Low-Pass Filtered Long Short-Term Memory Algorithm", Quantitative Bio-Science, 39(1), pp.25~31 (2020), 자연과학연구소[DOI]
- 양준혁, "Forecasting the KOSPI 200 Stock Index Based on LSTM Autoencoder", Quantitative Bio-Science, 39(2), pp.101~109 (2020), 자연과학연구소[DOI]
- Rock K, "Why new issues are underpriced", J Financ Econ, 15, pp.187~212 (1986)
- Chen T, "Xgboost : A scalable tree boosting system", SIG KDD Int’l Conf on Knowledge Discovery and Data Mining, 22, pp.785~794 (2016)
- 조기환, "매매시점 파악을 위한 유전자 알고리즘 기반 스코어링 모델", 한국데이터정보과학회지, 29(3), pp.735~745 (2018), 한국데이터정보과학회[DOI]
- Kim KJ, "Simultaneous optimization method of feature transformation and weighting for Artificial Neural Networks using Genetic Algorithm : Application to Korean stock market", J Intel Inf, 2, pp.323~335 (1999)
- 조득환, "인공지능 (AI)을 활용한 공모주 투자여부 및기준 수익률 달성 여부 예측 모델", 한국데이터정보과학회지, 31(3), pp.579~590 (2020), 한국데이터정보과학회[DOI]
- 신상훈, "비재무 정보를 활용한 IPO 주식의 상장일가격 등락 예측에 관한 연구", 한국데이터정보과학회지, 29(2), pp.425~439 (2018), 한국데이터정보과학회[DOI]