Read the foundations, examine the supporting evidence, and understand the challenges. These works motivate research; they do not validate this engine or promise investment returns.
LITERATURE UPDATE · SEPTEMBER 25, 2026
The research continues. The inference still needs testing.
Three recent peer-reviewed journal articles have been added below: cross-asset evidence, a fractal-feature forecasting model, and a study showing why efficiency conclusions depend on the diagnostic. Together they support investigating multiple scales—not a claim that this engine has an investment edge.
Our selection is a targeted reading list, not a systematic review. We separate the authors’ reported findings from our interpretation, identify abstract-only reviews, and retain counterevidence. The relevant product question remains whether these features add reliable out-of-sample value over simpler baselines after costs.
Yes, broadly. Our synthesis of the works below is that both encourage studying market complexity across scales. Mandelbrot supplies mathematical and statistical foundations; Peters puts heterogeneous investment horizons and liquidity at the center of a market hypothesis.
Question
Mandelbrot’s work
Peters’ FMH
What is the focus?
Roughness, heavy tails, scaling and dependence.
Participants with different horizons and their contribution to liquidity.
How are markets studied?
Statistical evidence and stochastic models, including stable and multifractal processes.
A market framework connecting horizon diversity, information and stability.
What connects them?
Attention to behavior across scales and to the limits of simple Gaussian descriptions. These ideas are compatible, but one does not prove the other.
Scaling does not automatically imply heavy tails, and heavy tails do not establish long memory or profitable predictability. Volatility dependence can coexist with uncorrelated signed returns. Market efficiency also does not, by itself, require Gaussian returns.
Journal articles, discussion papers and central-bank research are identified separately. Recent additions reviewed September 25, 2026; the foundational list was reviewed September 6, 2026. Access and review limits are listed on each recent entry.
Faruk Temel & Osman Tuğay Computational Economics · published online December 16, 2025
Contribution
The publisher abstract reports multifractality across eight assets using daily returns from 2018–2022, with different memory behavior across markets and changing Hurst estimates. This motivates comparing scales and market environments.
Interpretation boundary
Abstract reviewed, not the subscription full text. We do not adopt its stronger claim that these patterns establish achievable abnormal returns. Statistical scaling alone does not demonstrate an executable, cost-adjusted strategy or validate this engine.
Read source ↗Publisher abstract and metadata reviewed September 25, 2026 · full text requires access
02
2025Journal articleRecent methods · fractal features and forecasting
Yuanyuan Du & Ye Tian PLOS One 20(12), e0335554 · December 2, 2025
Contribution
Combines multiscale fractal descriptors with a deep-learning forecasting architecture and reports improved prediction metrics in its experiments. It provides a recent example of testing fractal features as model inputs.
Interpretation boundary
Reported prediction accuracy is not demonstrated net trading performance. The authors identify extreme-market stability as needing further verification. Independent replication, chronological validation and realistic costs are needed before adoption. The current engine does not implement this neural architecture.
Read source ↗Open peer-reviewed article · publisher text reviewed September 25, 2026; experiments not independently replicated
Meghna Jayasankar Computational Economics 66, 5033–5054 · first published February 7, 2025
Contribution
Studies Tether, USD Coin and Bitcoin using multifractal detrended fluctuation analysis and other tests. The abstract reports multifractality, while the variance-ratio test identifies Bitcoin as efficient. Different diagnostics can answer different questions.
Interpretation boundary
Publisher abstract reviewed, not the subscription full text. Stablecoins and Bitcoin have different mechanisms, and conclusions depend on the sample and test. Multifractality is not sufficient evidence of a tradable inefficiency.
Read source ↗Publisher abstract and metadata reviewed September 25, 2026 · full text requires access
Benoît Mandelbrot The Journal of Business 36(4), 394–419
Contribution
Develops a non-Gaussian, stable-Paretian model for speculative price changes. A foundational challenge to treating large market moves as negligibly rare Gaussian events.
Interpretation boundary
A proposed model is not a universal distribution for every asset. Mandelbrot published a correction in 1972 concerning part of the original cotton-price evidence.
Read source ↗Publisher record · access may require a subscription
Benoît B. Mandelbrot & John W. Van Ness SIAM Review 10(4), 422–437
Contribution
Develops fractional Brownian motion and fractional noise as a framework for self-similarity and dependence across scales.
Interpretation boundary
This is a mathematical model, not a finding that financial prices follow it. Fractional Brownian motion is Gaussian: self-similarity and heavy tails are separate properties.
Read source ↗Publisher record · access may require a subscription
Benoît Mandelbrot, Adlai Fisher & Laurent Calvet Cowles Foundation Discussion Paper 1164, Yale University
Contribution
Combines heavy tails with dependence in absolute price increments, while signed increments can remain uncorrelated. Introduces multiscaling of return-distribution moments.
Interpretation boundary
A model construction, with empirical investigation discussed separately. Unlike the earlier stable model, it need not imply infinite variance. The engine does not implement or fit this MMAR model.
Synthesizes recurring properties of financial returns, including distributional tails and nonlinear dependence, and discusses the statistical problems these create for models.
Interpretation boundary
Stylized facts summarize patterns across datasets; they are not laws that every series must obey, and they do not uniquely identify a fractal explanation.
Read source ↗Publisher record · access may require a subscription
08
2013Journal articleInvestment horizons · empirical support
Ladislav Kristoufek Scientific Reports 3, article 2857
Contribution
Finds increased short-scale wavelet power during turbulent periods in six stock indices, consistent with the FMH prediction of short-horizon dominance during market stress.
Interpretation boundary
Wavelet power is a proxy for activity across scales, not a direct census of investors’ holding periods. Retrospective consistency does not establish causation or a reliable advance crash warning.
Nicola Anderson & Joseph Noss Bank of England Financial Stability Paper 23
Contribution
Offers a quantitative model connecting agents with different investment horizons to fractal price behavior and market stability. Examines disruptions to their interaction.
Interpretation boundary
The causal mechanism is a conjecture explored through a model. A central-bank research paper does not establish official endorsement of FMH or of this product.
Develops a modified R/S test robust to short-range dependence. Finds no evidence of long-range dependence in the stock-return indices and sample periods studied after that adjustment.
Interpretation boundary
This challenges particular long-memory claims, not every form of scaling or dependence in volatility. It motivates testing short-range alternatives before interpreting an estimated Hurst exponent.
The (Mis)Behavior of Markets: A Fractal View of Risk, Ruin, and Reward
Benoît B. Mandelbrot & Richard L. Hudson
An accessible synthesis of Mandelbrot’s market research: roughness, extreme events, and the limits of familiar risk assumptions. A general-audience book, not a peer-reviewed validation of a trading system.
Fractal Market Analysis: Applying Chaos Theory to Investment and Economics
Edgar E. Peters
The foundational book for Peters’ Fractal Market Hypothesis, with investment horizons, fractal time series and R/S analysis. Read alongside later empirical and methodological challenges.
Research inspiration is not product validation. The current engine implements experimental R/S, modified R/S, DFA and Haar wavelet diagnostics. It does not fit Mandelbrot’s multifractal asset-return model, qualify Hurst-based trading signals, or enable recommendations. No cited author, publisher or institution is affiliated with or endorses this product. See implementation status →