[Paper Review] On the Time-Varying Efficiency of Cryptocurrency Markets

Akihiko Noda|arXiv (Cornell University)|Apr 20, 2019

Blockchain Technology Applications and Security4 citations

TL;DR

This study applies Ito et al.'s time-varying generalized least squares model to test the adaptive market hypothesis (AMH) in major cryptocurrencies. It finds that market efficiency fluctuates over time, with Bitcoin showing higher efficiency than Ethereum and Ripple, and overall market efficiency evolving dynamically, supporting AMH in crypto markets.

ABSTRACT

This study examines whether the market efficiencies of major cryptocurrencies (e.g., Bitcoin, Ethereum, and Ripple) change over time based on the adaptive market hypothesis (AMH) of Lo (2004). In particular, we measure the degree of market efficiency using Ito et al.'s (2014, 2016, 2017) generalized least squares-based time-varying model. The empirical results show that (1) the degree of market efficiency varies with time in cryptocurrency markets, (2) the market efficiency level of Bitcoin is higher than that of the other markets over most periods, and (3) the market efficiency of cryptocurrencies has evolved. We conclude that the results support the AMH for the established cryptocurrency market.

Motivation & Objective

• To investigate whether cryptocurrency market efficiency changes dynamically over time.
• To test the validity of the adaptive market hypothesis (AMH) in digital asset markets.
• To compare the time-varying efficiency levels across major cryptocurrencies like Bitcoin, Ethereum, and Ripple.
• To assess how market efficiency evolves in response to changing market conditions and investor behavior.

Proposed method

• Employs Ito et al.'s (2014, 2016, 2017) generalized least squares-based time-varying model to estimate dynamic market efficiency.
• Applies the model to high-frequency price data of major cryptocurrencies to capture short-term efficiency shifts.
• Uses the adaptive market hypothesis (AMH) framework to interpret efficiency as a function of market conditions and investor learning.
• Estimates time-varying efficiency by modeling the predictability of returns through a rolling window approach.
• Compares efficiency levels across Bitcoin, Ethereum, and Ripple using consistent statistical methodology.
• Validates results through robustness checks on different time windows and return horizons.

Experimental results

Research questions

• RQ1 Does the degree of market efficiency in cryptocurrency markets vary over time?
• RQ2 How does the efficiency of Bitcoin compare to that of Ethereum and Ripple across different periods?
• RQ3 To what extent has the overall efficiency of cryptocurrency markets evolved since their inception?
• RQ4 Do the observed patterns of efficiency support the adaptive market hypothesis (AMH) in digital asset markets?

Key findings

• The degree of market efficiency in cryptocurrency markets is not constant but varies significantly over time.
• Bitcoin exhibits a higher level of market efficiency compared to Ethereum and Ripple over most observed periods.
• The efficiency of all major cryptocurrencies has evolved, indicating dynamic adaptation to market conditions.
• The time-varying efficiency patterns align with the predictions of the adaptive market hypothesis (AMH).
• Efficiency levels are sensitive to market shocks and changes in investor behavior, particularly during periods of high volatility.
• The results suggest that market efficiency is not a fixed property but a dynamic outcome shaped by market evolution and participant learning.