Hybrid intelligent system for pricing the indices of dual-listing stock markets

Abstract

In previous studies (Howe et al., 2002; Alaganar and Bhar, 2002; Samant and Korth, 1998) it has been found that there is strong correlation between the American Depositary Receipts (ADRs) and the corresponding stocks in their native/primary markets in the long run. The results found that such correlation is mainly one way from the native markets to the ADRs. The method they employed is the econometrics and what they had studied are the correlation properties of these individual stocks. A simplified automated system is outlined to study the index of dual-listed stocks, namely ong Kong's Hang Seng Index and the Hang Seng London Reference Index. They have different trading hours and the possibility of incorporating London Reference Index into the pricing of the Hong Kong's Hang Seng Index is tested. One of the difficulties of AR method and neural network is how to choose suitable input. Evolutionary computation is outlined here of how to overcome this difficulty by employing it to simulate the markets' interactive dynamics. The study of the dual-listed index as a whole and the employment of the intelligent system in its pricing make it distinct from these previous econometrics studies in ADRs. The modelings here show positive results for incorporating the index movements in the secondary market into the pricing of the primary market. The percentage error is found to reduce by about 10%.