Digital Nugget: Factors behind crypto returns - an introduction to sentiment scores
Sentiment scores can give short-term indications about market direction. In this article, we outline an approach to sentiment scores for cryptocurrencies – one of several factors that can be considered in evaluating the crypto markets.
At Sygnum, we analyse a number of factors for cryptocurrencies. This example with data from 30 November illustrates one approach using market and on-chain data for Bitcoin and Ethereum, with a time series of numerous factors. Statistical analysis is then used to assess the predictive strength of these factors for the following month’s returns.
Below we describe our process and the factors that we have found to be most useful when analysing short-term market direction.
First, we applied what is referred to in data science as “feature engineering”. This includes normalising the scales (to harmonise factors whose values range, for example, from -1 to +1 versus other factors whose values may be in the millions) and assessing whether the absolute level of a factor or its rate of change is a better indicator. This normalisation ensures the consistency of measuring factor dynamics in different regimes.
We then analysed the factors’ explanatory and forecasting power using the statistical method of cluster analysis. We split the values of each factor into four equally sized buckets (quartiles), with 25 percent of the data points in each. For each quartile of each factor, we measure the corresponding average return of Bitcoin and Ether in the following month. The factor quartile with the lowest return is then classed as a bearish indicator, and vice versa for the highest return.
Through rigorous analysis, we have selected the most significant factors, and we will regularly reassess. If necessary, we may reselect factors going forward.
For this factor, we refer to the Crypto Fear & Greed Index . Somewhat counterintuitively, high levels of the index (greed) tend to be associated with high returns in the following month (rather than a reversal). Lower levels of the index (fear) are associated with reasonable returns. The lowest returns have tended to occur when the index was hovering around average levels.
Monthly price momentum appears to have some persistence in predicting crypto returns, indicating that bearish and bullish regimes both tend to persist.
The alpha opportunity factor measures the dispersion of returns for crypto assets, i.e. the spread between the best performing and worst performing cryptocurrencies. High dispersion environments create a lot of opportunities for outperformance from token selection, and vice versa. We find that both very high and very low dispersion environments are associated with lower returns in the following month, while dispersion closer to the average appears to predict the highest returns.
Options skew measures the spread between the implied volatilities of put options (bearish/defensive strategy) and call options (bullish sentiment). We used a combination of 1-week and 1-month options, with a 0.25 market exposure (“delta”). We find that a high options skew in either direction was associated with the strongest returns in the following month.
Here we measure the average correlation between cryptocurrency returns (Bitcoin, Ether) and the returns of traditional asset classes including equities, bonds, commodities, and gold. We find that periods of higher correlation to traditional assets are usually followed by a strong performance for the crypto market in the following month.
The on-chain activity tracks the growth in the user base and user activity for Bitcoin and Ethereum. As such growth is value creative for crypto assets, we would expect a positive correlation with returns. This is generally borne out by most factors, and we combine six on-chain metrics to get a balanced indicator (active addresses, addresses with balance over USD 10, total fees, on chain flows, coin age, and token velocity).
Current factor speedometer
In the final step, we display a visual representation of the different factors. The four coloured quartiles represent the spectrum from bearish (red) to bullish (dark green). The most recent value of the indicator is shown by the black bar, with the precise value and the change relative to the previous month shown in the middle. The indicators below are as of 30 November 2021.
This snapshot indicates a mostly positive regime, with several indicators in the green, and the alpha opportunities indicator in a strongly bullish state. Meanwhile, the market momentum and market pulse signals are largely neutral.
This document is purely for educational purposes and has been issued by Sygnum Group. It is not intended for distribution, publication, or use in any jurisdiction where such distribution, publication, or use would be unlawful, nor is it aimed at any person or entity to whom it would be unlawful to address such a marketing communication. It does not constitute an offer or a recommendation to subscribe, purchase, sell or hold any security or financial instrument. It contains the opinions of Sygnum Group, as at the date of issue. These opinions and the information contained herein do not take into account an individual‘s specific circumstances, objectives, or needs. No representation is made that any investment or strategy is suitable or appropriate to individual circumstances or that any investment or strategy constitutes personalized investment advice to any investor. Therefore, you must verify the above and all other information provided in the document or otherwise review it with your external advisors. Some investment products and services, including custody, may be subject to legal restrictions or may not be available worldwide on an unrestricted basis. The information and analysis contained herein are based on sources considered as reliable. Sygnum Group uses its best efforts to ensure the timeliness, accuracy, and comprehensiveness of the information contained in this document. Nevertheless, all information indicated herein may change without notice.