Literature review: Risks and Returns of Cryptocurrency by Liu and Tsyvinski (2018)

Summarized findings

The researchers analyzed the relationship between cryptocurrencies and stocks, commodities, currencies, macro factors and sentiment. Next to momentum sentiment is the most promising factor with predictive power.

  • Bitcoin, Ethereum, and Ripple’s XRP are not correlated with five major world currencies
  • there is no consistent evidence of systematic relationships between cryptocurrencies and the evaluated commodities (gold, platinum, and crude oil)
  • Proxies used for macroeconomy: non-durable consumption growth, durable consumption growth, industrial production growth, and personal income growth (apart from some probably insignificant relationship between ETH and durable consumption growth)
  • The authors find a significant time-series momentum present for Bitcoin, Ripple and cryptocurrency momentum at the daily and weekly frequencies (e.g.: “a one-standard-deviation increase in the current day’s Bitcoin return predicts a 0.33 percent increase in the daily return over the next day”)
  • Positive investor sentiment, proxied via Google searches and “Bitcoin” mentions on Twitter and negative sentiment, proxied via Google searches for “Bitcoin hack”, exhibit positive and negative relationships with future returns of up to 2.75%
  • Proxies based on the cost of mining (as a way to capture supply factors) show no relationship with returns for Bitcoin and Ripple while for Ethereum, there is some evidence that returns are exposed to the stock returns of AMD

Our comments:

    • It is remarkable how many different factors the researchers analyzed with regards to cryptocurrency prices
    • We find the method how positive and negative sentiment was operationalized (positive = search terms of coin, negative = “bitcoin hack”) does not live up to the full emotional spectrum users express on crypto social media