سیستم هشدار دهنده برای موسسات سرمایه گذاری جهانی در بازارهای سهام در حال ظهور بر اساس پیش بینی های یادگیری ماشین

At local emerging stock markets such as Korea, Hong Kong, Singapore and Taiwan, global institutional investors (GII) comprised of global mutual funds, offshore funds, and hedge funds play a key role and more often than not cause severe turmoil via massive selling. Thus, for the concerned local governments or private and institutional investors, it is quite necessary to monitor the behavior of GII against a sudden pullout. The main aim of this article is to propose an early warning system (EWS) which purposes issuing warning signal against the possible massive selling of GII at the local market. For this, we introduce machine learning algorithm which forecasts the behavior of GII by predicting future conditions. Technically, this EWS is an advanced form of the EWS developed by Oh et al. [Oh, K. J., Kim, T. Y., & Kim, C. (2006). An early warning system for detection of financial crisis using financial market volatility. Expert Systems, 23, 83–98] which issues a warning based on classifying present conditions. This study is empirically done for the Korean stock market.

Over the last 10 years, emerging stock markets such as Korea, Hong Kong, Singapore and Taiwan have been incorporated into the world financial market (Ghysels & Seon, 2005). Globalization and removal of the regulations at the local markets make global institutional investors (GII) to be major influences at the local markets. As a result, the movements of GII actually direct the local markets particularly when severe external or internal shocks hit the market. For example, GII, by May 2004, has occupied almost half the Korean stocks in terms of total market capitalization and in cases of external or internal shock such as Asian financial crisis in 1997, Russia moratorium in 1998, liquidity crises of Daewoo group in June–December 1999 and Hyundai group in June–December 1999, and 9/11 terrorists attack in 2001, the abnormal pullouts of GII has led the Korean stock market to the near collapse (Choe et al., 1999, Kim and Kwon, 2003 and Kim and Wei, 1999). In order to prepare against such a devastating situation, a proper early warning system (EWS) that can detect or predict abnormal pullout of GII, especially global hedge funds, is strongly desired at the local markets. In order to develop EWS for GII (EWSGII), we will follow a standard procedure developed by Oh, Kim, and Kim (2006) which employs machine learning algorithms for establishing EWS for financial markets. Technically, their procedure is based on classifying the current situation according to some standard measures such as market volatility. The core of the procedure is defining the gray zone as its main feature vector which may proceed to either stable condition or collapse and issuing a warning when the market enters the gray zone. In this article, we present an advanced EWS which issues warning by classifying and forecasting future conditions of a market. The core of the procedure is defining the oracle rule which determines the future market conditions in advance and then tracing (forecasting equivalently) the oracle rule with trained machine learning algorithm. The rest of this study consists of as follows: Section 2 reviews the previous studies about behavior of GII and the existing EWS algorithm. A discussion of technical aspects of the EWS monitoring GII (EWSGII) is also given. Section 3 presents the detailed construction procedure of EWSGII. Section 4 is devoted to the empirical case study for building EWSGII for the Korean stock market. Concluding remarks are given in Section 5.

Globalization and deregulation of local stock markets make GII major influences at the local markets. A problem with GII is that they may try to sell the stocks as much and fast as possible and leave the local stock market in the case that external shocks like Asia crisis arises since their main goal is to gain an absolute return for their customers. Therefore, an efficient and predictive EWS is strongly needed for monitoring or forecasting the movements of GII. In this study, the EWSGII is proposed which forecasts the movements of GII by classifying the future market condition. For this the oracle and trained classifiers are introduced. Through empirical study, it is shown that performance of EWSGII depends on the oracle rule and the type of training dataset. More precisely, it is noticed that for training dataset based on the ESP against the market trend, conservative oracle classifier appears desirable, while for training dataset based on the ESP with the market trend, sensitive oracle classifier appears desirable. It is also interesting to report that there exists an optimal target lag for EWSGII based on forecasting, optimal in the sense that its correct hit rate with testing data is uniformly high.