Paper: arXiv 2507.09863
Authors: Ryuji Hashimoto, Kiyoshi Izumi
Abstract
We investigate the mechanisms behind the power-law distribution of stock returns using artificial market simulations. While traditional financial theory assumes Gaussian price fluctuations, empirical studies consistently show that the tails of return distributions follow a power law. Previous research has proposed hypotheses for this phenomenon – some attributing it to investor behavior, others to institutional demand imbalances. However, these factors have rarely been modeled together to assess their individual and joint contributions. The complexity of real financial markets complicates the isolation of the contribution of a single component using existing data. To address this, we construct artificial markets and conduct controlled experiments using optimal transport (OT) as a quantitative similarity measure. Our proposed framework incrementally introduces behavioral components into the agent models, allowing us to compare each simulation output with empirical data via OT distances. The results highlight that informational effect of prices plays a dominant role in reproducing power-law behavior and that multiple components interact synergistically to amplify this effect.
Complexity vs Empirical Score
- Math Complexity: 7.0/10
- Empirical Rigor: 6.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced mathematical concepts like optimal transport, power-law exponents, and Hill estimators, while also detailing a controlled simulation pipeline with quantitative metrics like OT distances to compare synthetic data against real market data.
Research Flowchart
flowchart TD
A["Research Goal<br>Disentangle factors behind<br>power-law distribution of stock returns"] --> B
B["Methodology<br>Artificial Market Simulations"] --> C
subgraph C ["Data & Inputs"]
direction LR
C1["Empirical Equity Data<br>Power-law tails"]
C2["Incremental Agent Models<br>Behavioral components"]
C3["Optimal Transport OT<br>Quantitative similarity measure"]
end
C --> D
subgraph D ["Computational Process"]
direction LR
D1["Controlled Experiments<br>Intro behavioral components"]
D2["Simulate Market Dynamics<br>Price impact & information flow"]
D3["Compare with Empirical Data<br>Using OT distance metrics"]
end
D --> E["Key Findings/Outcomes"]
E --> F["Informational effect of prices<br>dominant in power-law reproduction"]
E --> G["Multiple components interact<br>synergistically to amplify effect"]
E --> H["Interpretable framework<br>isolates individual contributions"]
style A fill:#e1f5fe,stroke:#01579b,stroke-width:2px
style B fill:#fff3e0,stroke:#e65100,stroke-width:2px
style C fill:#f3e5f5,stroke:#4a148c,stroke-width:2px
style D fill:#e8f5e8,stroke:#1b5e20,stroke-width:2px
style E fill:#fff9c4,stroke:#f57f17,stroke-width:2px
style F fill:#ffebee,stroke:#c62828,stroke-width:1px
style G fill:#ffebee,stroke:#c62828,stroke-width:1px
style H fill:#ffebee,stroke:#c62828,stroke-width:1px
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 a method to factorize financial stylized facts using artificial market simulations with quantitative evaluation based on optimal transport, and demonstrates it on the universal cubic law.
Source sentence
To summarize, our contributions are as follows: • We propose methods for 1) factorizing financial stylized facts through artificial market simulations with hypotheses evalu-ation cycles of agent models, and 2) calibrating simulations based on quantitative evaluation of financial synthetic data using OT. • We demonstrate these methods by applying them to uni-versal cubic law.
Assumes 4
- The artificial market uses a single market with tick size 1.00×10⁻⁴, initial price 300.00 and constant fundamental price. market
Source sentence
We recorded ev-ery order and execution event with several statistics such as market/mid prices and order/execution volumes at each time step to form synthetic tick data. • Market: We set the number of artificial markets to 1, tick size to 1.00 × 10−4, initial market price to 𝑝0 = 300.00,
- Agent parameters are set as n=200, λ_f=10.00, λ_n=1.00, σ_n=0.01, etc., following Chiarella et al. model
Source sentence
and fundamental price to a constant value of ∀𝑡𝑝𝑓 𝑡= 300.00. • Agent: We set the value of following parameters regarding our agent model as follows with reference to Chiarella et al. [9]. 𝑛= 200, 𝜆𝑓= 10.00, 𝜆𝑛= 1.00, 𝜎𝑛= 0.01, ∀𝑗𝑃𝑟[𝑀𝑗 0 = 𝑂] = 0.5, 𝜏𝑗,𝑓= 200,
- Real data are taken from FLEX-FULL historical tick data provided by the Japan Exchange Group. data
Source sentence
Common settings for simulations 0 to 7 are as follows. • Real data: We used FLEX-FULL historical tick data provided by the Japan Exchange Group [27].
- Each simulation consists of 2,110 time steps, with order placement only for the first 100 steps and no execution. execution
Source sentence
In our experiment, each simu-lation consisted of 𝑇𝑠𝑖𝑚= 2, 110 time steps. For the first 100 steps and from step 1, 100 to 1, 110, only order placement was allowed; no order execution occurred.
Data
- FLEX-FULL historical tick data order-book commercial
Source sentence
Common settings for simulations 0 to 7 are as follows. • Real data: We used FLEX-FULL historical tick data provided by the Japan Exchange Group [27]. It includes order and exe-cution series data, called tick data.
- Synthetic data from artificial market simulations synthetic unknown
Source sentence
Artificial markets, a form of multi-agent simulation, offer a con-structive approach to understanding financial phenomena—such as simulation 0 w/o any components simulation 1 w/ component 1 simulation !! −# w/ components 1,...,N ... synthetic data 1 synthetic data !! −# synthetic data 0 real data generate and preprocess quantitative evaluation by optimal transport ... components regarding the origins of stylized
Costs
Not applicable
Finds 2
- The informational effect of prices (chartist trader) plays a dominant role in reproducing power‑law behavior, and multiple components interact synergistically to amplify this effect. theoretical
Source sentence
The results highlight that informational effect of prices plays a dominant role in reproducing power-law behavior and that multiple components interact synergistically to amplify this effect. Additionally, the experimental results suggest that each of the above candidates acts in combination, centering on the chartist trader, that is, the informational effect of prices.
- When both the power‑law of demand size and informational effect of prices are incorporated, the Hill index decreases more significantly than the simple sum of individual effects, indicating a synergistic relationship. statistical
Source sentence
The figure reveals that when both the power law of demand size and informational effect of prices are incorporated in to the agent model, the Hill index decreases more significantly than the simple sum of the individual effects, as suggested by Equation (13).
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 6 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.
Cited References (59)
- The economy needs agent-based modeling (2009)
- Agent-based models for financial crises (2017)
- Complexity theory and financial regulation (2016)
- High-frequency financial market simulation and flash crash scenarios analysis: An agent-based modelling approach (2024)
- A study on the reversal mechanism for large stock price declines using artificial markets (2012)
- How to evaluate trading strategies: Single agent market replay or multiple agent interactive simulation? (2019) — link
- Market making via reinforcement learning (2018)
- Multi-agent reinforcement learning in a realistic limit order book market simulation (2021)
and 51 more
- Deep reinforcement learning in agent based financial market simulation (2020)
- Volatility clustering in financial markets: A micro-simulation of interacting agents (1998)
- Power laws of wealth, market order volumes and market returns (2001)
- Deep calibration of market simulations using neural density estimators and embedding networks — → 2311.11913
- Institutional investors and stock market volatility (2006)
- Cul de sacs and highways: An optical tour of dark pool trading performance (2008)
- Deeper hedging: A new agent-based model for effective deep hedging — → 2310.18755
- A theory of power-law distributions in financial market fluctuations (2003)
- The limiting extremal behaviour of speculative returns: An analysis of intra-daily data from the frankfurt stock exchange (1998)
- Scaling of the distribution of fluctuations of financial market indices (1999)
- A simulation analysis of the microstructure of double auction markets (2002)
- The impact of heterogeneous trading rules on the limit order book and order flows (2009)
- Stock prices and social dynamics (1984)
- Generating synthetic data in finance: Opportunities, challenges and pitfalls (2021)
- Generating realistic stock market order streams (2020)
- Deep Q-learning market makers in a multi-agent simulated stock market (2022)
- Adversarial deep hedging: Learning to hedge without price process modeling — → 2307.13217
- Data augmentation of high frequency financial data using generative adversarial network (2020)
- Financial power laws: Empirical evidence, models, and mechanisms (2016)
- The flash crash: High-frequency trading in an electronic market (2017)
- Historical Data (2025)
- Get real: Realism metrics for robust limit order book market simulations (2021)
- Learning and calibrating heterogeneous bounded rational market behaviour with multi-agent reinforcement learning (2024)
- A noise trader model as a generator of apparent financial power laws and long memory (2007)
- A simple general approach to inference about the tail of a distribution (1975)
- Trade clustering and power laws in financial markets (2020)
- A limit theorem for financial markets with inert investors (2006)
- From KISS to KIDS – An `Anti-simplistic’ modelling approach (2005)
- The complexity of cooperation: Agent-based models of competition and collaboration (1997) — link
- Rank — 1/2: A simple way to improve the OLS estimation of tail exponents (2011)
- Implementation of actual data for artificial market simulation (2022)
- Scaling and criticality in a stochastic multi-agent model of a financial market (1999)
- A method of simulated moments for estimation of discrete response models without numerical integration (1989)
- Applying the method of simulated moments to estimate a small agent-based asset pricing model (2009)
- Reinforcement learning for combining search methods in the calibration of economic ABMs (2023)
- Impact analysis of financial regulation on multi assets markets using artificial market simulations (2020)
- Noise (1986)
- PAMS: Platform for artificial market simulations — → 2309.10729
- Computational optimal transport: With applications to data science (2019)
- Introduction to optimal transport theory (2010) — link
- POT: Python optimal transport (2021)
- Scaling behavior in the dynamics of an economic index (1995)
- Heterogeneity and persistence in returns to wealth (2020)
- Towards a fully rl-based market simulator (2022)
- Empirical properties of asset returns: Stylized facts and statistical issues (2001)
- Understanding intra-day price formation process by agent-based financial market simulation: Calibrating the extended chiarella model (2022) — link
- Effects of price regulations and dark pools on financial market stability: An investigation by multiagent simulations (2016)
- Coherent measures of risk (1999)
- Volatility clustering in financial markets: Empirical facts and agent-based models (2007)
- The relation between price changes and trading volume: A survey (1987)
- Towards realistic market simulations: A generative adversarial networks approach (2022)
4 of 59 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.