Undergraduate Thesis Completed

Abstract

With the proliferation of digital communication, social media platforms have increasingly become pivotal in the dissemination of financial information, thereby shaping investor behavior and influencing market dynamics. This study leverages empirical data from Xueqiu, a prominent social trading platform, by integrating advanced Natural Language Processing (NLP) techniques, sentiment dictionaries, BGE-M3 word embedding models, and Generative Pre-trained Transformer models (GPT-4). Our primary objective is to scrutinize the alignment between investors’ discourse on social media and their actual trading activities.

The analysis reveals a pronounced congruence between the narratives shared by investors on Xueqiu and their market behaviors. By incorporating variables such as gender and the clarity of information, this research elucidates the factors that drive consistency between verbal expressions and trading actions. Notably, it is observed that male participants and individuals who articulate their thoughts more clearly are predisposed to higher levels of behavioral consistency.

Moreover, the study quantifies the market effects of this consistency by calculating Cumulative Abnormal Returns (CAR) and Investment returns (Inv_ret), underpinning the substantive role of social media information in financial markets. These insights offer crucial decision-making support to both investors and regulatory bodies, highlighting the transformative impact of online financial dialogues.

Awards

• Second Prize, Third ”Xueshi Cup” Academic Paper Competition of Nanjing University
• Outstanding Undergraduate Thesis

python
# Can you decrypt my WeChat number using Caesar cipher?

# Step 1: This is the encrypted version of my WeChat number using Caesar cipher with a shift of 3.
encrypted_wechat = "dpvwurqjcbi"

# Step 2: Use your programming skills to decrypt this Caesar cipher and find my WeChat number.

# Hint: The original WeChat number is a lowercase string, and the shift used is 3.
# You'll need to shift each letter in the encrypted_wechat string backwards by 3 to reveal the original.

# Good luck!