Paper: arXiv 2504.19050

Authors: Bruno Giorgio

Abstract

This dissertation investigates the ability of the Ising model to replicate statistical characteristics, or stylized facts, commonly observed in financial assets. The study specifically examines in the S&P500 index the following features: volatility clustering, negative skewness, heavy tails, the absence of autocorrelation in returns, and the presence of autocorrelation in absolute returns. A significant portion of the dissertation is dedicated to Ising model-based simulations. Due to the lack of an analytical or deterministic solution, the Monte Carlo method was employed to explore the model’s statistical properties. The results demonstrate that the Ising model is capable of replicating the majority of the statistical features analyzed.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 6.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper uses advanced statistical mechanics and agent-based modeling with Monte Carlo simulations, requiring high mathematical sophistication, but its empirical evaluation relies heavily on theoretical model fitting and visual comparisons to stylized facts rather than robust out-of-sample backtesting or trading performance metrics.

Research Flowchart

  flowchart TD
  A["Research Goal: Investigate Ising model<br>to replicate S&P 500 stylized facts"] --> B["Methodology: Monte Carlo Simulation"]
  B --> C["Data Input: S&P 500 Historical Data"]
  C --> D{"Computational Process:<br>Simulate Ising Model Phase Transitions"}
  D --> E["Key Outcomes: Model successfully replicates<br>Volatility Clustering, Heavy Tails, Negative Skewness,<br>Absence of Return Autocorrelation,<br>Presence of Absolute Return Autocorrelation"]

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 proposes using the Ising model, simulated via Monte Carlo methods, to replicate the main stylized facts observed in financial assets such as the S&P 500.

Source sentence
This dissertation investigates the ability of the Ising model to replicate statistical characteristics, or stylized facts, commonly observed in financial assets. The results demonstrate that the Ising model is capable of replicating the majority of the statistical features analyzed.

Assumes 6

  • Each agent is represented by a spin variable Si(t) that can take values ±1. model
    Source sentence
    Let us consider Si(t) = ±1 as the spin orientation for each i-th agent in the lattice.
  • For simplicity, the dynamics of each agent are updated according to the probabilistic rule Si(t+1)=+1 with p = 1/(1+e^{−2βhi(t)}) and Si(t+1)=−1 with 1−p. model
    Source sentence
    For simplicity, we assume that the dynamics of each agent are updated according to the following equations: Si(t + 1) = +1 with p = 1 1 + e−2βhi(t) Si(t + 1) = −1 with 1 −p In the equations, β is a coupling coefficient, hi(t) rep-resents the local field acting on agent i at time t, and p is the probability.
  • β is a coupling coefficient that determines the influence of the local field on agents. model
    Source sentence
    For simplicity, we assume that the dynamics of each agent are updated according to the following equations: Si(t + 1) = +1 with p = 1 1 + e−2βhi(t) Si(t + 1) = −1 with 1 −p In the equations, β is a coupling coefficient, hi(t) rep-resents the local field acting on agent i at time t, and p is the probability.
  • When β = 0 (infinite temperature) all spin configurations occur with equal probability and spins behave independently. model
    Source sentence
    In con-trast, when β = 0 (i.e., infinite temperature), all configurations occur with equal probability, and the spins behave independently.
Show 2 more
  • The lack of an analytical or deterministic solution for large systems necessitates the use of Monte Carlo simulation. model
    Source sentence
    Due to the lack of an analytical or deterministic solution, the Monte Carlo method was employed to explore the model's statistical properties.
  • Daily adjusted‑close prices of the S&P 500 index are an appropriate proxy for market returns. data
    Source sentence
    We collected daily financial data spanning the past forty years, specifically from February 1982 to Febru-ary 2022. The dataset consists of the Adjusted Close prices of the S&P 500 index.

Data

  • S&P 500 Adjusted Close prices market-prices public
    Feb 1982 - Feb 2022 · S&P 500 index
    Source sentence
    We collected daily financial data spanning the past forty years, specifically from February 1982 to Febru-ary 2022. The dataset consists of the Adjusted Close prices of the S&P 500 index.

Costs

Not applicable The paper reports no trading or execution result.

Finds 5

  • The Ising model is capable of replicating the majority of the statistical features analyzed, including volatility clustering, heavy tails, and negative skewness. statistical
    Source sentence
    The results demonstrate that the Ising model is capable of replicating the majority of the statistical features analyzed.
  • The simulated returns exhibit volatility clustering, with clusters of large fluctuations followed by other large changes, mirroring the empirical S&P 500 behavior. statistical figure from elsewhere in the paper
    Source sentence
    Analysis of Results Volatility Clustering Volatility clustering was the first criterion used to evaluate whether the model exhibited behavior con-sistent with real financial markets. The model exhibits a memory effect, characterized by clusters of large fluctuations followed by other large changes.
  • Autocorrelation of absolute returns in the Ising simulation shows long‑memory behavior that decays according to a power‑law, similar to the real market data. statistical
    Source sentence
    The autocorrelation of absolute returns demon-strated long-memory behavior, decaying according to a power-law—a significant departure from the inde-pendence assumed in random walk models.
  • Both the Ising model returns and the S&P 500 returns display significant skewness and leptokurtic (heavy‑tailed) distributions, rejecting normality in Shapiro‑Wilk and Jarque‑Bera tests. statistical figure from elsewhere in the paper
    Source sentence
    Statistical analysis further confirmed the presence of leptokurtic distributions with heavy tails and nega-tive skewness in both the Ising simulations and actual market data. These findings were supported by infer-ential tests (Jarque-Bera and Shapiro-Wilk), which consistently rejected the null hypothesis of normality, aligning with the heavy-tailed, asymmetric nature of financial return distributions.
  • The Monte Carlo simulation was performed on a 32×32 lattice for 1,000,000 iterations, generating 10,000 return observations with lag Δt = 100. theoretical figure from elsewhere in the paper
    Source sentence
    In our initial simulation, we employ a two-dimensional lattice composed of 32×32 agents. Specifically, the system runs through a loop of 1, 000, 000 iterations.

Checked by others

Supported by

  • Impact of the COVID-19 pandemic on the financial market efficiency of price returns, absolute returns, and volatility increment: Evidence from stock and cryptocurrency markets (2025)
    On the question: Do absolute returns in financial markets display long‑memory (Hurst exponent >0.5) and power‑law autocorrelation decay?
    Compare the findings

    This paper: Autocorrelation of absolute returns in the Ising simulation shows long‑memory behavior that decays according to a power‑law, similar to the real market data.

    That paper: The Hurst exponents of stock returns remain around 0.5 (except VIX), while absolute returns have exponents >0.5, reflecting long‑memory.

    A reports Hurst exponents >0.5 for absolute returns; B observes power‑law decay of absolute‑return autocorrelation, confirming long‑memory.

  • Black-Litterman Asset Allocation under Hidden Truncation Distribution (2023)
    On the question: Do asset return distributions deviate from normality, showing skewness and heavy tails?
    Compare the findings

    This paper: Both the Ising model returns and the S&P 500 returns display significant skewness and leptokurtic (heavy‑tailed) distributions, rejecting normality in Shapiro‑Wilk and Jarque‑Bera tests.

    That paper: Likelihood‑ratio test strongly rejects normality, supporting the skew‑normal assumption for the data (p‑value = 0.0015).

    Finding A rejects normality (p=0.0015) favoring skew‑normal; Finding B also rejects normality via Shapiro‑Wilk and Jarque‑Bera, showing skewness and leptokurtosis.

  • Stylized Facts of High-Frequency Bitcoin Time Series (2024)
    On the question: Do absolute‑return autocorrelations exhibit power‑law decay?
    Compare the findings

    This paper: Autocorrelation of absolute returns in the Ising simulation shows long‑memory behavior that decays according to a power‑law, similar to the real market data.

    That paper: The autocorrelation of absolute returns follows a power‑law initially, with Period 2 showing a slightly higher decay rate than Period 1.

    Both A and B find power‑law decay; B’s simulation confirms A’s real‑data pattern. In A, Period 2 decays slightly faster than Period 1.

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.