Abstract
Conventional financial models fail to explain the economic and monetary properties of cryptocurrencies due to the latter's dual nature: their usage as financial assets on the one side and their tight connection to the underlying blockchain structure on the other. In an effort to examine both components via a unified approach, we apply a recently developed Non-Homogeneous Hidden Markov (NHHM) model with an extended set of financial and blockchain specific covariates on the Bitcoin (BTC) and Ether (ETH) price data. Based on the observable series, the NHHM model offers a novel perspective on the underlying microstructure of the cryptocurrency market and provides insight on unobservable parameters such as the behavior of investors, traders and miners. The algorithm identifies two alternating periods (hidden states) of inherently different activity -- fundamental versus uninformed or noise traders -- in the Bitcoin ecosystem and unveils differences in both the short/long run dynamics and in the financial characteristics of the two states, such as significant explanatory variables, extreme events and varying series autocorrelation. In a somewhat unexpected result, the Bitcoin and Ether markets are found to be influenced by markedly distinct indicators despite their perceived correlation. The current approach backs earlier findings that cryptocurrencies are unlike any conventional financial asset and makes a first step towards understanding cryptocurrency markets via a more comprehensive lens.
Introduction
Traditional financial models, designed for fiat currencies or stocks, often fail to capture the "dual nature" of cryptocurrencies. Crypto isn't just a financial asset; it’s a technological protocol. Its price is influenced by both macroeconomic indicators (like the S&P 500) and on-chain metrics (like Hash Rate or transaction volume). This paper addresses the challenge of market microstructure: distinguishing between periods of rational, fundamental investing and periods of speculative frenzy. Because these "states" are not directly observable in the price alone, the authors use Bayesian inference to "peek" into the underlying dynamics that drive price volatility and returns.
Methodology
The core of the study is the Non-Homogeneous Polya Gamma Hidden Markov model (NHPG). Unlike standard Hidden Markov Models (HMMs) where the probability of switching states is constant, this model allows the transition probabilities to change based on external variables (covariates). For example, a spike in the NASDAQ might make the market more likely to switch from a "stable" state to a "volatile" state. The authors use Markov Chain Monte Carlo (MCMC) sampling to estimate the model parameters, providing a robust way to handle the non-linear and non-stationary nature of crypto data.
Datasets and Covariates
The study analyzes daily log-returns for Bitcoin (BTC) and Ether (ETH) across multiple time windows. To drive the "non-homogeneous" transitions, they used 14 covariates categorized into exchange rates, stock indices, commodities, market sentiment, and blockchain metrics.
Key Results: The Two Faces of Bitcoin
The model identifies two clear states for BTC. State 1 is high-volatility, accounting for about one-third of the time, and is heavily influenced by miners' activity (Hash Rate) and stock market swings. This represents "noise traders" or speculators. State 2 is more stable, showing "long memory" (autocorrelation) and behaving more like a random walk influenced by traditional exchange rates. This represents "fundamental investors."
Interestingly, the model struggled to find clear state transitions for Ether. ETH prices appeared more isolated from global macroeconomic indicators than BTC, suggesting it is still an emerging, less mature market. Despite BTC and ETH being highly correlated in price, they are influenced by markedly different indicators, suggesting that "crypto" is not a monolithic asset class.
Conclusion
The NHPG model successfully segmented the data into subseries with vastly different statistical properties, proving that a single-regime model is insufficient for crypto. While the study assumes exactly two states, future work could integrate transaction graph analysis and account for the shift from Proof-of-Work to Proof-of-Stake, which will fundamentally change how network activity influences price.
Summary updated