Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/80980
Title: On the hawkes' processes and its application on A-H shares
Authors: Liu, Qian
Advisors: Yiu, Ka-fai Cedric (AMA)
Keywords: Finance -- Mathematical models
Liquidity (Economics) -- Mathematical models
Stock exchanges -- Mathematical models
Issue Date: 2019
Publisher: The Hong Kong Polytechnic University
Abstract: There are many indicators for measuring the financial market liquidity. The thesis explores recent academic literature related to one-dimensional and two-dimensional Hawkes' processes applied in financial market. Hawkes' processes are derived from a model of multivariate point processes and has recently been sought-after for the applications in high frequency financial models in the past decade. In this thesis, the Hawkes process with exponential response function is applied to model the arrival process of the trades, where the expected intensity and flow branching ratio are calculated from the estimated values of the model parameters. The expected intensity represents the transaction intensity, which is an indicator of the adequacy of liquidity; and another liquidity indicator is the branching ratio, which is the measure of the degree of aggregation of transactions. After giving a brief overview of the main definitions on properties and characteristics of how Hawkes' processes are applied, we describe procedures for simulating and deriving maximum likelihood estimation (MLE) functions for parameter estimation on one dimensional process and extend to two dimensional case. Then we survey various empirical studies using one-dimensional and two-dimensional Hawkes' processes in A-H shares both in Mainland and in the Hong Kong stock markets. Comparable studies of both the Mainland and Hong Kong markets are beneficial because both markets differ, so we would like to focus on different investors investment behavior. We choose two pairs of A-H shares both in Mainland and the Hong Kong stock markets in different industries classification. It seems intuitive to analyze the different regulated markets simultaneously since the process of market evolution can be observed as well as the differences and similarities of the two markets. Our model accounts for the arrival of bid-ask orders intensities on both two markets that influence activities, trigger one-sided or two-sided clustering of trades. We use branching ratio and expected intensity as measurement of activity on A-H shares that provides a direct way to access their level of endogeneity and also relates to market liquidity. From our numerical results, We can observe that two-dimensional Hawkes' processes have better explanatory power on market liquidity than one-dimensional Hawkes' processes. Also the resulting from confusion matrix, the overall accuracy on trading strategy would be more than 70%. Furthermore the performance on H-shares in Hong Kong stock market did better than A-shares in Mainland China stock market. There are still many applications of Hawkes' process, as we discuss in relating with two-dimensional Hawkes' processes its bid-ask side, calculate its branching ratio and expected intensity. Another contribution of this thesis surrounds the bid-ask expectation intensity we proposed a trading strategy on confusion matrix. Through confusion matrix comparison, we conclude that the forecast performance of expected intensity of H-shares in Hong Kong market is obviously better than that of A-shares in the domestic market.
Description: xviii, 99 pages : color illustrations
PolyU Library Call No.: [THS] LG51 .H577M AMA 2019 Liu
URI: http://hdl.handle.net/10397/80980
Rights: All rights reserved.
Appears in Collections:Thesis

Files in This Item:
File Description SizeFormat 
991022232429203411_link.htmFor PolyU Users168 BHTMLView/Open
991022232429203411_pira.pdfFor All Users (Non-printable)2.85 MBAdobe PDFView/Open
Show full item record
PIRA download icon_1.1View/Download Contents

Page view(s)

12
Citations as of Aug 21, 2019

Download(s)

7
Citations as of Aug 21, 2019

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.