Understanding Cryptocurrencies

At Cloudwall we don’t publish only fantasy essays on the Blockchain Economy future, but we also develop risk and valuation systems for digital assets. And what better way to kick off our weekly paper review if not with a piece written by traditional asset allocation researchers? Even though it’s co-authored by one of the greats in asset allocation research, Campbell “Cam” Harvey, the analytics part of this paper is quite basic. The list of further research proposals is definitely worth a read. And there’s also a Jupyter Notebook with the code used in this paper. Neat, we like open research and will contribute as well.

Show me the data

The paper starts with a customary review of the blockchain ecosystem, kept simple so as to not scare us, TradFi guys. If you are starting from the basics, maybe just go directly to prof. Harvey’s book on DeFi, it just came out.

The dataset has been collected from disparate sources and runs from May 1, 2017, to June 30, 2019. The Notebook has an updated dataset up to October 5, 2020 (note that the paper has been first submitted to Arxiv in July 2020). Before May 2017 the data for some of the chosen tokens is either not available, or the assets are not liquid enough. Does the data have the same cut-off for different assets? We don’t know. Where exactly is the data coming from? We don’t know. But we’d like to know next time.

The authors then present a benchmarking analysis between the S&P 500 Index (SPX), SPDR Gold ETF (GLD), and the S&P 500 volatility index (VIX) and three cryptocurrencies: BTC, ETH, and XRP. Among these, the authors suggest that ETH is easier to value, given that it has a tangible component in the form of smart contracts. Would have loved to see this statement drilled down a bit further. The third token, Ripple, is presented as a challenger to the SWIFT system for transfers.

Correlation is not causation, but still…

The analysis part starts with Pearson correlations over the whole period, showing a strong positive relationship between ETH and BTC, while there is no evidence of the correlation between the selected cryptos and traditional assets. If we instead look at 250 days rolling correlations, we notice that the co-movement between BTC vs XRP and ETH increases monotonically through time, especially in the second part of the dataset. Interestingly, in the updated dataset we can see that from 2020 the BTC becomes more positively correlated with GLD and SPX, while at the same time the relationship vs VIX becomes more negative. We can’t assume just from this observation that BTC behaves as a risk-on asset, but it’s a good indication. Overall, the rolling correlations show that the relationships change relatively often, presenting us with both risks and investment opportunities.

The analysis of standard deviations confirms higher risks in crypto vs traditional assets, though there appears to be some convergence over the most recent period. Looking at the distributional properties, we see that returns on cryptos were far away from following a Normal distribution. Note that traditional assets in this paper are also far from zero skewness and kurtosis of 3, with observations that are more extreme than on average over the past 30 years. Interestingly, if we consider only the third and the fourth moments of the distribution of log returns, BTC looks more Normally distributed than SPX. As the skewness measures the relative size of the tails, this means BTC has a more symmetric risk distribution between the negative and positive returns. And being the kurtosis a measure of the combined tails relative to the center of the distribution, this means that the probability of the extreme events for BTC is closer to a Normal. Being the volatility so high, this doesn’t mean it’s less risky, as the size of the extreme events in BTC is clearly more significant. There’s an anecdotal “5x rule of thumb” in crypto: 8% drop is a tail risk in Equities? Make it 40% in cryptocurrencies. This is one of the reasons why smaller hedge funds make more sense in crypto if they can understand and manage the risks without imploding.

Enough of anecdotes, let’s go back to hard science. The authors also conduct a basic co-integration test. Well, the Johansen test is not basic, but it’s more nuanced than the paper is showing. There are not enough methodological explanations and interpretation of results to consider this part a good piece of research, but the authors show that we can’t reject the hypothesis of no co-integration between BTC and GLD for that particular time frame and with whatever default settings there are in the R or Python co-integration packages they’ve used. Compared to correlation analysis, co-integration modeling is a better way of finding relationships between variables, as it checks whether certain time series share stochastic trends. Definitely worth more research on this topic. From our analysis, this has been changing as of late, though the relationship between BTC and GLD is not stable and we can’t assume that BTC is being considered an inflation hedge. And the mixed reaction of BTC and ETH to the recent US CPI figures does show that investors are also not sure about whether to bet on BTC because of inflation risks (higher CPI) or to stay out of it for a potential risk-off on Fed reducing their ocean of easy liquidity. Interestingly, ETH outperformed BTC. Is that because it lagged previously? Or because Fed’s liquidity tightening is expected to come without interest rate increases, keeping DeFi yields an interesting income opportunity vs ZIRP? We don’t know, but we are building tools to help investors understand these new market relationships.

Further research directions

For the last and sizable part of the paper, there is some literature review and potential research directions, and it’s a pretty neat collection, and arguably the best part of the paper. I’ll leave it here as a list of questions we are interested in, feel free to reach out if you want to collaborate on some of these. We have data, quant skills, and knowledge of both the traditional finance, as well as the digital assets markets, and we know how to do a proper co-integration analysis… We’d be happy to collaborate with both the academia and the private sector research teams.

Network design, valuation, and risk

• A combination of on-chain and off-chain trading yields high-dimensional inter-dependencies between different venues and assets. How do you apply network design to e.g. cross-chain transactions?
• How do we model different blockchain protocols in a game-theoretic way? How do we explain the behavior of miners and users?
• How do the micro-structure features such as exogenous structural constraints influence the dynamics and stability of blockchains?
• Is there an intrinsic value related to a network’s computing power and adoption? Can this value relate to the Value risk factor? Can we use it in asset pricing?
• There are new sources of information being used for digital assets. For instance, what social media activity can we use to assess the value and risk of digital assets? Can GitHub activity be useful in determining protocol risk?
• How does the hash-rate and price relationship impact demand and supply shocks?
• What frictions are introduced by DEXs and how do they impact risk?

Monetary systems and financial developments

• Can we model interactions between monetary policies and digital asset prices?
• What blockchain-based monetary systems are possible? And how do we model the impact of e.g. CBDCs on monetary policies?
• What effects do stablecoins have on “traditional” monetary policies?
• Is a financial equilibrium possible in a blockchain-based financial system, and how would that look?

Price discovery

• Can we develop a dynamic pricing model for cryptocurrencies and other digital assets?
• What are the optimal ways to drive user adoption? Is it through price appreciation expectations? How to use this for early-stage pricing of new protocols?
• What information can we get from order books? Is there any information asymmetry (e.g. between inner and outer layers of the order book)?
• How to spot cryptocurrency price manipulation?
• How do you deal with index constructions and digital asset-specific features such as forks?
• How do we detect bubbles in cryptocurrencies? Can this be used in real-time to monitor and manage risk?
• What are the cross-exchange arbitrage opportunities due to? Is it mainly about cross-border controls? Exchange fragmentation?
• What are the relationships between arbitrage opportunities and protocol latency, latency uncertainty, spot volatility, and risk aversion?

Portfolio diversification

• How does BTC compare to gold? Are their price dynamics related? Can they be both used for portfolio diversification?
• Can we apply traditional risk-based portfolio construction methodologies to lower the risk in digital asset portfolios?
• Can we find some stylized facts about the cross-section of digital asset returns?

In summary, there’s a lot of work to do here. We will be covering these and additional questions in our future research, so stay tuned and again, do reach out if you are open to research collaborations or working with us!

Disclaimer

Not a financial advice, solicitation, or sale of any investment product. The information provided to you is for illustrative purposes and is not binding on Cloudwall Capital. This does not constitute financial advice or form any recommendation, or solicitation to purchase any financial product. The information should not be relied upon as a replacement from your financial advisor. You should seek advice from your independent financial advisor at all times. We do not assume any fiduciary responsibility or liability for any consequences financial or otherwise arising from the reliance on such information.

You may view this for information purposes only. Copy, distribution, or reproduction of all or any portion of this article without explicit written consent from Cloudwal is not allowed.