Modelling sentiment and trader aggressiveness in cryptocurrency markets: An empirical analysis of the Bitcoin limit order book.
Cryptocurrency assets are inherently susceptible to social sentiment. Due to emotional contagion, extreme emotions influence and drive traders’ order execution behaviour. This study aims to empirically model the relationship between sentiment and trader aggressiveness for two types of traders: Whale...
Saved in:
| Main Author: | |
|---|---|
| Format: | text |
| Language: | English |
| Published: |
Institutional Knowledge at Singapore Management University
2023
|
| Subjects: | |
| Online Access: | https://ink.library.smu.edu.sg/etd_coll/541 https://ink.library.smu.edu.sg/context/etd_coll/article/1539/viewcontent/GPBA_AY2019_DBA_AngeliqueNicoleThang.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Singapore Management University |
| Language: | English |
| Summary: | Cryptocurrency assets are inherently susceptible to social sentiment. Due to emotional contagion, extreme emotions influence and drive traders’ order execution behaviour. This study aims to empirically model the relationship between sentiment and trader aggressiveness for two types of traders: Whales and Retails. Additionally, I uncover the mediating roles of momentum, mean reversion, and market timing underpinning the dynamics of the off-chain Bitcoin market. Time series data collected from the Coinbase level 3 Bitcoin limit order book is first reconstructed to access its core structure. Next, sentiment data from Reddit is extracted and processed using an internal Natural Language Processing pipeline. These datasets are synthesized, and a suite of multivariate regression and autocorrelation models formulated to analyse the pertinent variables. Baseline results underscore the pivotal role of sentiment in significantly predicting Whale and Retail trader aggressiveness in the Bitcoin market. Mean reversion is evidenced by a negatively autocorrelating sentiment measure. Further regression analyses reveal the dynamics of the interplay of momentum, mean reversion, and market timing in predicting short-term and long-term sentiment. Whale order imbalance and Whale order aggressiveness are also found to be more effective in predicting Bitcoin returns. |
|---|