The study, titled “Investor Sentiment States and Tail Risk Management in Cryptocurrency Markets,” focuses on an increasingly important problem in digital finance: cryptocurrency prices can move sharply within short periods, making conventional measures of average returns insufficient for understanding potential losses. Rather than asking only whether sentiment can predict cryptocurrency returns, Zhou and Zhusupova examined whether sentiment can identify different risk environments.
Fear and Greed as Market Risk Signals
The researchers used the Crypto Fear & Greed Index, a publicly available indicator ranging from 0 to 100. Lower scores represent stronger fear, while higher scores indicate stronger greed.
The approach recognizes that cryptocurrency markets can behave differently depending on the prevailing psychological environment. The researchers compared volatility, potential losses and portfolio drawdowns across the three sentiment states.
The study used daily closing prices for Bitcoin and Ethereum from January 1, 2020, through December 31, 2024, producing 1,826 daily observations. The two cryptocurrencies were combined into an equally weighted portfolio, with 50% allocated to Bitcoin and 50% to Ethereum.
To examine downside risk, the researchers used rolling volatility, Value-at-Risk (VaR), Conditional Value-at-Risk (CVaR) and drawdown measures. CVaR is particularly useful for identifying losses that extend beyond a specified risk threshold.
Fear Periods Show Much Higher Risk
The differences between sentiment states were substantial.
During Fear periods, average portfolio volatility reached 4.83%, compared with 2.91% during Neutral periods and 3.70% during Greed periods.
The same pattern appeared in downside-risk measures. Average rolling VaR was:
Because CVaR focuses on losses beyond the VaR threshold, the results indicate that extreme fear was associated with more severe losses in the tail of the portfolio's return distribution.
Drawdowns also differed sharply. Average current drawdown reached −51.76% during Fear periods, compared with −34.31% during Neutral periods and −18.57% during Greed periods.
By contrast, average returns changed only modestly across the three states. Average portfolio returns were 0.30% during Fear, 0.22% during Neutral, and 0.23% during Greed.
This distinction is important. The results suggest that sentiment was more closely connected with changes in market risk conditions than with differences in average returns.
Statistical tests reinforced the pattern. Differences across sentiment states were statistically significant for volatility, rolling VaR, rolling CVaR and drawdown, while differences in portfolio returns were not statistically significant.
Sentiment Adds Information Beyond Volatility
Zhou and Zhusupova also tested whether sentiment provided information about downside risk after accounting for market volatility.
Their first regression model found that the Fear & Greed Index was significantly associated with rolling CVaR and explained approximately 24.1% of its variation.
When portfolio volatility was added to the model, explanatory power increased to approximately 58.1%, while the sentiment variable remained statistically significant. This indicates that sentiment contained information about downside risk that was not fully captured by volatility alone.
A third model focused specifically on the Fear state. The model explained approximately 53.9% of the variation in rolling CVaR, while the Fear indicator showed a significant relationship with more severe downside-risk conditions.
Zhou and Zhusupova therefore present sentiment as a risk-monitoring signal rather than a standalone prediction tool. Their analysis suggests that a decline in investor sentiment can coincide with a market environment in which losses become more severe.
A Sentiment-Based Portfolio Strategy
The researchers also tested whether sentiment information could be used for practical risk management.
The benchmark portfolio maintained a constant allocation of 50% Bitcoin and 50% Ethereum. Under the sentiment-based strategy, exposure was reduced during Fear periods to 25% Bitcoin, 25% Ethereum and 50% cash. During Neutral and Greed periods, the portfolio returned to the 50%-50% cryptocurrency allocation.
The strategy reduced portfolio volatility from 3.67% to 3.15%, representing an approximately 14.24% reduction. Its Sharpe ratio also increased from 1.227 to 1.276, indicating improved risk-adjusted performance.
However, the strategy did not eliminate extreme losses. Cumulative return fell from 19.88% for the benchmark to 17.65%, while maximum drawdown was slightly worse at −76.71%, compared with −76.31% for the benchmark. Rolling CVaR at the latest observation was identical for both strategies at −5.29%.
The results therefore do not support treating sentiment as a guaranteed protection against cryptocurrency crashes. Instead, reducing exposure during fearful periods may lower ordinary portfolio fluctuations while sacrificing some participation in subsequent recoveries.
Implications for Cryptocurrency Risk Management
The findings have practical relevance for cryptocurrency investors, portfolio managers and financial analysts.
The Crypto Fear & Greed Index can serve as one additional indicator for identifying changing market conditions. Investors could potentially combine sentiment information with volatility, liquidity, market momentum and other risk indicators when monitoring cryptocurrency exposure.
The research also demonstrates why average returns alone can be misleading when assessing digital-asset risk. A market can show relatively similar average returns across different sentiment environments while experiencing very different levels of volatility, drawdown and extreme-loss exposure.
As Ningyu Zhou and Zhanat Mukhtarovna Zhusupova of Al-Farabi Kazakh National University report, investor sentiment provides additional information about dynamic downside risk, but the sentiment-based strategy has a limited effect on the most extreme losses. Their findings support using sentiment as part of broader risk monitoring rather than as an independent trading signal.
The authors acknowledge important limitations. The analysis covers only Bitcoin and Ethereum, uses a single sentiment indicator, assumes a simple allocation rule, and does not incorporate transaction costs or alternative rebalancing strategies. The findings may therefore not automatically apply to smaller or less liquid cryptocurrencies.
About the Authors
Ningyu Zhou is the corresponding author of the article and is affiliated with Al-Farabi Kazakh National University, Almaty, Kazakhstan.
Zhanat Mukhtarovna Zhusupova is a co-author affiliated with the same university.
The supplied manuscript does not state the authors’ academic degrees or provide detailed individual biographies, so specific degree titles and professional specialties cannot be confirmed from the source.
0 Komentar