Paper: arXiv 2310.04907

Authors: Emanuele Citera, Francesco De Pretis

Abstract

We study the stochastic structure of cryptocurrency rates of returns as compared to stock returns by focusing on the associated cross-sectional distributions. We build two datasets. The first comprises forty-six major cryptocurrencies, and the second includes all the companies listed in the S&P 500. We collect individual data from January 2017 until December 2022. We then apply the Quantal Response Statistical Equilibrium (QRSE) model to recover the cross-sectional frequency distribution of the daily returns of cryptocurrencies and S&P 500 companies. We study the stochastic structure of these two markets and the properties of investors’ behavior over bear and bull trends. Finally, we compare the degree of informational efficiency of these two markets.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper employs advanced statistical equilibrium models and information theory (Shannon entropy) to analyze market structure, indicating high mathematical complexity. It also demonstrates strong empirical rigor with a detailed, multi-year dataset of 46 cryptocurrencies and S&P 500 companies, including data processing, cross-sectional analysis, and comparisons of market efficiency.

Research Flowchart

  flowchart TD
  A["Research Goal: Compare Stock & Crypto Markets<br>via Cross-Sectional Distributions & Efficiency"] --> B["Data Collection: Stocks S&P 500 & 46 Major Cryptos<br>Jan 2017 - Dec 2022"]
  B --> C["Methodology: Apply Quantal Response<br>Statistical Equilibrium (QRSE) Model"]
  C --> D["Process: Recover Cross-Sectional<br>Return Frequency Distributions"]
  D --> E["Analysis: Stochastic Structure &<br>Investor Behavior in Bull vs. Bear Trends"]
  E --> F["Outcome: Comparative Analysis of<br>Informational Efficiency"]
  F --> G["Key Findings:<br>Structural differences in markets<br>Efficiency disparities"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper applies the Quantal Response Statistical Equilibrium (QRSE) model to recover the cross‑sectional frequency distribution of daily returns for cryptocurrencies and S&P 500 companies.

Source sentence
We then apply the Quantal Response Statistical Equilibrium (QRSE) model to recover the cross-sectional frequency distribution of the daily returns of cryptocurrencies and S&P 500 companies.

Assumes 4

  • Return series are truncated at eight standard deviations from the mean to limit extreme observations. data
    Source sentence
    We truncate the series of returns at eight standard deviations from the mean.
  • A balanced‑panel approach is used to handle large non‑random turnover in the cryptocurrency universe. data
    Source sentence
    To construct our dataset, we followed Bhambhwani et al. [2023], who adopted a balanced-panel approach due to a large non-random turnover in the universe of cryptocurrencies.2 Our sample comprises forty-six cryptocurrencies, which we can divide into three groups: 11 Proof-of-Work, 7 Non-Proof-of-Work, and 28 additional.
  • Time‑reversibility of the return processes implies statistical equilibrium. model
    Source sentence
    Since time reversibility implies equilibrium, a reversible transformation must be quasi-static, even if the reverse is not neces-sarily true.
  • Stationarity (or trend‑stationarity) of the return series is required for equilibrium analysis. statistical
    Source sentence
    Stationarity and trend-stationarity tests The Augmented Dickey-Fuller (ADF) and Kwiatkowski-Phillips-Schmidt-Shin (KPSS) tests are commonly used to determine the stationarity or trend-stationarity of a time series.

Data

  • Cryptocurrency price and return dataset · CoinMarketCap (and Bittrex for additional assets) market-prices public
    January 2017 to December 2022 · 46 major cryptocurrencies (11 Proof‑of‑Work, 7 Non‑Proof‑of‑Work, 28 additional)
    Source sentence
    Data Collection and Analysis To conduct our analysis, we used two samples spanning six years, from January 1, 2017, to December 31, 2022. The first data source is a set of cryptocurrencies, whose details are collected in Appendix A. We collected the respective daily prices from CoinMarketCap and then computed logarithmic daily returns.
  • S&P 500 daily price dataset · Yahoo Finance market-prices public
    January 2017 to December 2022 · All companies listed in the S&P 500
    Source sentence
    Data Collection and Analysis To conduct our analysis, we used two samples spanning six years, from January 1, 2017, to December 31, 2022. The second data set collects daily closing adjusted prices of the companies listed in the S&P 500, which we retrieved from Yahoo Finance.

Recognised sources: CoinMarketCap Yahoo Finance

Costs

Not applicable The paper reports no trading or execution result.

Finds 5

  • The cryptocurrency market exhibits more and longer bear‑trend periods than the S&P 500 market. statistical
    Source sentence
    Whereas the S&P 500 records few instances of bear markets, the cryp-tocurrency market generates more occurrences of bear trends over prolonged periods.
  • Buying frequencies decrease before bear markets for both asset classes, with a more pronounced intensity in the cryptocurrency market. statistical
    Source sentence
    respectively.5 In terms of trading frequencies (top and bottom left panes), our results show a general decreasing trend in the buying frequencies before a bear market which is common to both asset classes. However, the intensity of trading is more remarkable in the cryptocurrency market, though bounded within a narrower range of values.
  • Higher trading activity is associated with positive expectation‑fulfillment values for S&P 500 companies, while the opposite holds for cryptocurrencies. statistical
    Source sentence
    In detail, our results show a higher trading activity for positive values of ζ in the case of the S&P 500 companies, whereas the opposite is true for the cryptocurrency market.
  • In the limit of zero temperature (T→0), the QRSE conditional choice becomes deterministic, yielding a single‑peaked Laplace‑like return distribution. theoretical
    Source sentence
    In turn, the conditional distribution f[a|r] approaches the Heavyside step function, and the QRSE distribution ( ˆf[r]) approximates a single-peaked Laplace distribution.
  • Cross‑sectional distributions of returns differ markedly: cryptocurrencies have a larger standard deviation and heavier tails than S&P 500 companies. statistical
    Source sentence
    Min Mean Median StDev Skew Kurt Max Cryptocurrencies -86.33 0.04 -0.09 8.56 0.58 13.27 88.87 S&P 500 Companies -17.99 0.05 0.09 2.11 -0.14 8.41 17.98 3Given that we are comparing two markets with completely different scales of fluctuations, we do not follow the standard definition provided by the Securities and Exchange Commission, which defines bear markets as declines of 20% or more over at least two months.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 21 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.