The duo found that digital currencies don’t behave like stocks or other currencies when it comes to price movements.
On August 6, Yale economist Yukun Liu and PhD candidate Aleh Tsyvinski published a working paper titled Risks and Returns of Cryptocurrency. In it, they argue that there are several factors that predict price trends of some of the most popular digital currencies. The study is said to be “the first-ever comprehensive economic analysis of cryptocurrency and the blockchain technology upon which it is based.”
The researchers start with the claim that many factors that are predictive of the prices of stocks, currencies, and precious metals do not apply to cryptocurrencies, and so calculating the risk-return tradeoff requires different methods. They then go on to discuss factors specific to the cryptocurrency markets, focusing on two that they find to be predictive of digital currency price trends: the “time-series momentum effect,” and the “investor attention effect.”
Time-series Momentum Effect
The authors first focus on a factor called the “time-series momentum effect.” This means that when asset or cryptocurrency prices are rising they tend to rise even higher. This method can be useful to predict the best time for investors to buy and sell their crypto. Tsyvunski described their findings to Yale News:
“We have designed a simple strategy that says an investor should buy Bitcoin if its value increases more than 20% in the previous week.”
Although the time-series momentum effect can be useful, investors should know that it is not always the best method to use. The argument is essentially, “if the price goes up, it will continue to go up,” which has some obvious flaws. For example, according to coinmarketcap.com, the price of bitcoin rose approximately 35 percent between December 31, 2017, and January 6 of 2018. According to the researchers’ argument, an investor would have been wise to buy bitcoin on January 6. However, in reality, on January 7, the price of bitcoin fell 6 percent. By January 13, the price had fallen by almost 13 percent.
Investor Attention Effect
The second factor the working paper focused on is what they call the “investor attention effect.” Basically, the authors asked, “If there is an abnormally high number of mentions of the cryptocurrencies we studied in either Google search or on Twitter, will their returns go up?” To conduct this research. Tsyvinski and Liu gathered data on how many times the words bitcoin, Ethereum, and Ripple were used in Google searches and tweets.
According to the report a one-standard-deviation increase in bitcoin Google searches over the course of one week, “leads to increases in weekly returns of 1.84 percent and 2.30 percent” at the one and two week marks, respectively. In other words, an increase in the number of Google searches containing the word bitcoin could be a useful way to predict price trends in the coming weeks.
Researchers also found that a one-standard-deviation increase in Google searches for Ripple over the course of one week indicated 10.86 percent increase in returns a week later. Their working paper states that under the same circumstances, Ether investors could reasonably expect a 4.36 percent gain on their return.
When looking at how investor attention on Twitter affected the price of bitcoin, the authors found that if the number of tweets containing the word bitcoin increased one week the price of bitcoin tended to rise by 2.5 percent in the one week ahead returns.
This isn’t exactly a groundbreaking idea. The more influencers such as Ethereum founder Vitlalik Buterin, or investors such as Barry Silbert, and Didi Taihuttu tweet about their respective digital currency, the more people are going to be aware and interested in these tokens. Moreover, when the general public sees tweets and conversations praising bitcoin, Ethereum, or Ripple by perceived experts, they will feel more comfortable and secure in investing their hard-earned money in these particular cryptocurrencies.
Interestingly, the authors also considered the impact negative investor attention had on the price trends of bitcoin. They found that an increase in Google searches containing the word “bitcoin hack” caused a decrease of 2.75 percent in bitcoin returns the following week.
The final question the authors asked was if the cost of mining crypto is an indicator of price trends in bitcoin, Ripple, and Ethereum. Surprisingly, even though the cost of mining rigs and the electricity it takes to mine crypto can be exorbitant, the researchers found that these costs had little effect on the price of bitcoin and Ripple. However, they did find some evidence that Ethereum prices are affected by the stock returns of Advanced Micro Devices Inc. (AMD), which is one of the main manufacturers of specialized mining hardware.
It is important to note that Tsyvinski and Liu are not providing investment advice, and do not advocate for one type of crypto or another. This paper was intended to “cast doubt on popular explanations that the behavior of cryptocurrencies is driven by its functions as a stake in the future of blockhain technology similar to stocks, as a unit of account similar to currencies, or as a store of value similar to precious metal commodities.”
The authors urge investors to do their own research, because as they told CNBC, “All things can happen. Maybe the statistical patterns that we find are going to completely change. Maybe tomorrow bitcoin is going to be prohibited by regulators, maybe it’s going to be completely hacked, there are many things one would take into account.”