The rise of Bitcoin has sparked intense academic and public interest, not only due to its groundbreaking technology but also because of its extreme price volatility and complex relationship with financial regulation. This study investigates the dynamic interactions among user behavior, media attention, search interest, price, and trading volume in the Bitcoin market—particularly before and after a major regulatory intervention in China. By comparing Bitcoin’s volatility with that of traditional stock markets, and analyzing shifts in causal relationships post-regulation, this article sheds light on how policy impacts market structure and investor behavior.
Bitcoin vs. Stock Markets: Assessing Relative Volatility
To understand Bitcoin’s inherent risk profile, we compare its price fluctuations with those of two major stock indices: the Dow Jones Industrial Average (DJIA) for the U.S. market and the SSE 300 Index (HS300) for China. Using daily data from January 1, 2013, to August 12, 2018, we apply detrending techniques to isolate pure volatility from long-term trends.
Three distinct trend-fitting methods are used:
- Convolution-based Voigt function for Bitcoin Price Index (BPI)
- 9th-order polynomial fitting for DJIA
- S-shaped BiDoseResp function for HS300
After removing underlying trends, we calculate the detrended ratio—a measure that reflects short-term deviations from expected values. The results reveal stark differences in market stability:
- Bitcoin’s detrended ratio standard deviation: 0.6214
- DJIA: 0.0355
- HS300: 0.1110
👉 Discover how Bitcoin's volatility compares to traditional assets in real-time.
This means Bitcoin’s price swings are approximately 17 times more volatile than the U.S. stock market and 6 times more than China’s equity market during the observation period. Additionally, Bitcoin exhibits high positive skewness and excess kurtosis—indicative of frequent extreme movements and "fat tails"—reinforcing its classification as a high-risk, speculative asset.
These findings align with broader research suggesting that while traditional markets are influenced by macroeconomic fundamentals, Bitcoin prices are more sensitive to sentiment, speculation, and network effects.
Measuring Investor Attention: Key Indicators
Understanding what drives Bitcoin’s price requires examining proxies for investor interest. Prior studies have used:
- Google Trends search volume (Garcia et al., 2014)
- Wikipedia page views
- Twitter mentions
- Forum activity (e.g., Bitcointalk.org)
In this study, we focus on three China-specific indicators:
- Search Volume (SEARCH): Daily Baidu search queries for “Bitcoin”
- Media Coverage (NEWS): Number of news articles mentioning Bitcoin in Chinese media
- User Growth (USER): New registrations on major domestic cryptocurrency platforms
Together with Bitcoin Price Index (BPI) and Daily Trading Volume (VOL), these variables form the core dataset analyzed through a Vector Autoregression (VAR) model, allowing us to explore dynamic feedback loops between sentiment and market performance.
Regulatory Shock: The 2017 Chinese Exchange Shutdown
A pivotal moment in Bitcoin’s history occurred on September 30, 2017, when Chinese regulators banned all domestic cryptocurrency trading activities. This policy event serves as a natural experiment to assess how regulatory interventions alter market dynamics.
We divide the dataset into two periods:
- Pre-regulation: January 1, 2013 – September 30, 2017
- Post-regulation: October 1, 2017 – August 12, 2018
This split enables comparative analysis of whether the relationships between investor attention metrics and market outcomes changed significantly after the ban.
Stationarity and Model Specification
Before estimating VAR models, we conduct Augmented Dickey-Fuller (ADF) tests to ensure time series stationarity. Results show that BPI and USER are non-stationary in levels. To address this, we transform them into:
- RETURN: Log-difference of BPI (Bitcoin returns)
- USER%: Log-difference of user count (user growth rate)
All five variables—RETURN, VOL, SEARCH, NEWS, USER%—are stationary in their transformed forms. We then estimate four VAR(2) models with optimal lag order selected via information criteria (AIC, BIC).
| Model | Period | Dependent Variables | Independent Variables |
|---|---|---|---|
| M1.1 | Pre-event | RETURN | USER%, SEARCH, NEWS |
| M2.1 | Post-event | RETURN | USER%, SEARCH, NEWS |
| M1.2 | Pre-event | VOL | USER%, SEARCH, NEWS |
| M2.2 | Post-event | VOL | USER%, SEARCH, NEWS |
Granger Causality: Uncovering Directional Relationships
Granger causality tests help identify predictive relationships between variables. A variable X “Granger-causes” Y if past values of X improve predictions of Y beyond what is possible using only past values of Y.
Findings on Price Drivers (RETURN)
Before Regulation (M1.1):
- Bidirectional causality exists between NEWS and RETURN
- SEARCH → RETURN: Search interest predicts price changes
- RETURN → USER%: Price increases attract new users
After Regulation (M2.1):
- Only SEARCH → RETURN remains significant
- Media coverage no longer influences prices
- The feedback loop between price and user growth weakens
👉 See how real-time sentiment affects crypto prices today.
Findings on Trading Volume (VOL)
Before Regulation (M1.2):
- Strong bidirectional causality among VOL, USER%, SEARCH, and NEWS
- All attention metrics actively drive trading volume
After Regulation (M2.2):
- USER% → VOL and NEWS → VOL links disappear
- Only SEARCH → VOL remains weakly significant
These results indicate that regulatory suppression disrupted the ecosystem where media narratives and user acquisition fueled trading activity. With exchanges shut down, information dissemination lost its market impact.
Supplementary Causal Links Among Sentiment Indicators
Further pairwise Granger tests reveal:
- Pre-event: Nearly all sentiment variables (SEARCH, USER%, NEWS) influence each other bidirectionally
- Post-event: Only VOL → RETURN breaks down; otherwise, internal sentiment dynamics remain intact
This suggests that while investor attention still correlates internally, its ability to translate into market action diminished post-ban.
FAQ Section
Q: Why is Bitcoin more volatile than stock markets?
A: Unlike equities tied to earnings and economic data, Bitcoin lacks intrinsic cash flows. Its value hinges largely on adoption, speculation, and network effects—making it highly responsive to sentiment shifts and macro-level regulatory news.
Q: Did the Chinese ban eliminate Bitcoin trading entirely in China?
A: While centralized exchanges were shut down, peer-to-peer trading and offshore platforms allowed continued participation. However, trading volume dropped sharply due to reduced accessibility and increased risk.
Q: Can search volume predict Bitcoin prices?
A: Yes—especially in emerging markets like China. Rising search interest often precedes price rallies, reflecting growing retail participation. However, this signal weakens under strict regulation.
Q: What does “Granger causality” mean in this context?
A: It measures whether one time series can statistically predict another. It doesn’t imply true causation but indicates informational precedence—e.g., rising searches may signal upcoming buying pressure.
Q: How reliable are VAR models for crypto analysis?
A: VAR models are effective for capturing short-term interdependencies among multiple variables. However, they assume linearity and stability—limitations when applied to rapidly evolving crypto markets.
Q: What are the implications for crypto regulation?
A: This study shows regulation can decouple investor attention from market outcomes. Effective oversight may reduce speculative frenzies but could also stifle innovation and market efficiency.
Core Keywords
Bitcoin volatility, policy intervention, investor sentiment, VAR model, Granger causality, financial regulation, market dynamics, media attention
👉 Explore advanced analytics tools to monitor crypto market sentiment and trends.
Conclusion
This empirical analysis demonstrates that Bitcoin operates as a fundamentally different asset class compared to traditional equities—characterized by extreme volatility and strong sensitivity to investor attention. Regulatory actions, such as China’s 2017 exchange ban, significantly alter the structure of information flow and market responsiveness. While search interest retains some predictive power, media influence and user growth lose their impact post-intervention.
These findings underscore the importance of integrating behavioral indicators into financial models and highlight the transformative role of regulation in shaping digital asset ecosystems. Future research should explore cross-country comparisons and the rise of decentralized finance (DeFi) as alternative channels for crypto engagement under restrictive regimes.