การศึกษาและจัดการข้อมูลทางทวิตเตอร์เพื่อเป็นตัวแทนความเชื่อมั่นของนักลงทุนในดัชนีตลาดหลักทรัพย์ กรณีศึกษาระยะสั้น : สงครามการค้า สหรัฐอเมริกา-จีน
This study presents the creation of information obtained from Twitter to represent investor’s sentiment proxy in DJI during an uncertain event, such as US-China Trade war. The data collection is divided into 2 periods, which are 10th –14th September 2018 (Normal period) and 17th –21st September 2018...
Saved in:
Main Author: | |
---|---|
Other Authors: | |
Format: | Theses and Dissertations |
Language: | other |
Published: |
เชียงใหม่ : บัณฑิตวิทยาลัย มหาวิทยาลัยเชียงใหม่
2020
|
Subjects: | |
Online Access: | http://cmuir.cmu.ac.th/jspui/handle/6653943832/71060 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Chiang Mai University |
Language: | other |
Summary: | This study presents the creation of information obtained from Twitter to represent investor’s sentiment proxy in DJI during an uncertain event, such as US-China Trade war. The data collection is divided into 2 periods, which are 10th –14th September 2018 (Normal period) and 17th –21st September 2018 (Abnormal period). The abnormal period is a peak period of twitter users discussed about the US-China Trade war. The samples of tweets were collected by using 5 key hashtags on Twitter API. 9 new variables from Twitter data at the same period were created, including hourly twitter volume and hourly twitter moods (8 Basic emotion).
The results show that, during normal period time, we are unable to use Twitter data to represent investors’ sentiment proxy. As both Twitter data have no statistically significant correlation with DJI performance. During abnormal period time, The results showed a statistically significant positive correlation between Twitter data and DJI performance. Moreover, The results also demonstrated that the MA (1) with GARCH (1,1) - FEAR and MA (1) with GARCH (1,1) - JOY models provided more accurate volatility in DJI returns than the MA (1) with GARCH (1,1), which was created without using Twitter data |
---|