The search for digital gold: Google trends data as a predictor for cryptocurrency price volatility
| dc.contributor.advisor | Alburo, Florian A. | |
| dc.contributor.author | Maliwanag, Rafael John S. | |
| dc.contributor.author | Uligan, Ileanna April D. | |
| dc.date.accessioned | 2024-07-26T06:22:25Z | |
| dc.date.available | 2024-07-26T06:22:25Z | |
| dc.date.issued | 2023-07 | |
| dc.description.abstract | Cryptocurrency price determination is one of the most interesting phenomena in modern finance. Polarizing as it is volatile, return and, consequently, risk in cryptocurrencies dwarf those of other traditional financial markets, and hence, constructing more holistic models of prediction undoubtedly has practical value. Proving to be useful in predicting market outcomes, the benefits of incorporating web search activity in forecasting models of cryptocurrency may be augmented due to the digital nativity of the market. Using time series data obtained from Google Trends, this paper examines the predictive capability of Google search volumes on cryptocurrency price volatility beyond prominent volatility models such as the GARCH(1,1). The study employs a two-step approach consisting of a GARCH(1,1) process and an ordinary least squares (OLS) regression. The conditional variance from the former, along with short-term and long-term indicators of Google Trends data are used as independent variables in an OLS regression of log volatility. The study has three main findings: 1.) Google Trends data predict cryptocurrency price volatility beyond the GARCH(1,1) model; 2.) Cryptocurrency price volatility is more responsive to the short-term Google Trends indicators rather than the long-term; and 3.) Among examined categories, keywords relating to the name and ticker of the coin (e.g. “Bitcoin” and “BTC”, for Bitcoin) and on widespread cryptocurrency-related terms have the most significant additional forecasting capabilities on the volatility of cryptocurrency returns beyond the GARCH(1,1). On the other hand, economic, geopolitical, and pandemic-related words seem to have little to no additional predictive capabilities. Ethereum is found to be most responsive to Google Trends data, while Cardano was found to be least responsive. | |
| dc.identifier.uri | https://selib.upd.edu.ph/etdir/handle/123456789/67 | |
| dc.language.iso | en | |
| dc.subject | cryptocurrency | |
| dc.subject | Google trends | |
| dc.subject | price volatility | |
| dc.subject | GARCH | |
| dc.title | The search for digital gold: Google trends data as a predictor for cryptocurrency price volatility | |
| dc.type | Thesis |