Paper: arXiv 2611.24028
Authors: Sai Niranjan Ramachandran
Abstract
Existing work suggests that markets can show volatility with paths rougher than those of Brownian motion. In theory this can happen when the order flow is close to instability with memory that decays as a power law. However, it is not fully clear how such order flow could arise from the predominantly algorithmic market making that characterizes order book dynamics. We show one such consistent mechanism. We treat each market maker as a desk that uses a first-order gradient rule to move the distances of its orders from the mid price toward target distances provided by an unknown oracle. We assume that distinct desks do not act synchronously in time but act whenever new information relevant to their respective oracles arrives, which each desk samples over its own characteristic lookback horizon. Across all desks the rule therefore takes the form of stochastic gradient descent (SGD) with information-driven (mini-batch) noise. In the continuum limit this update rule makes the volume of the order book follow a conservative stochastic partial differential equation (the Dean–Kawasaki equation) whose fluctuations drive the order flow. The distinct horizons of the desks further induce a persistent memory in the order flow that decays as a power law. Standard scaling arguments then lead to rough volatility. We confirm the key predictions of the model on simulated agents and on six months of Binance data with about 790 million orders. In liquid markets governed mainly by market makers, such as the perpetual futures of Binance, the order books are consistent with how our desks change the book.
Complexity vs Empirical Score
- Math Complexity: 8.5/10
- Empirical Rigor: 8.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: This paper presents a highly mathematical model grounded in stochastic partial differential equations and stochastic gradient descent, demonstrating strong theoretical depth. It is rigorously validated with both simulations and a substantial real-world dataset from Binance, confirming its empirical relevance. The novelty lies in connecting algorithmic market making directly to rough volatility through order book dynamics, offering a fresh perspective on market microstructure.
Research Flowchart
flowchart TD
A["Research Goal: How does algorithmic market making lead to rough volatility?"] --> B{"Key Methodology: Model Market Maker Behavior"};
B --> C["Market Maker Desks use SGD with oracle targets & varying lookback horizons"];
C --> D{"Computational Process: Continuum Limit Derivation"};
D --> E["Outcome: Order book volume follows Dean-Kawasaki SPDE, driving power-law decaying memory in order flow"];
E --> F["Predicted Outcome: Leads to Rough Volatility via standard scaling arguments"];
F --> G{"Data/Inputs: Simulated Agent Data & 6 months of Binance Perpetual Futures (790M orders)"};
G --> H["Key Findings: Model predictions confirmed; Binance order books consistent with proposed desk behavior, explaining rough volatility."];
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 mechanism where each market maker is modeled as a desk that uses a first‑order gradient rule to adjust order distances based on an unknown oracle, leading to a conservative stochastic PDE for book volume that generates rough volatility.
Source sentence
We treat each market maker as a desk that uses a first-order gradient rule to move the distances of its orders from the mid price toward target distances provided by an unknown oracle. We assume that distinct desks do not act synchronously in time but act whenever new information relevant to their respective oracles arrives, which each desk samples over its own characteristic lookback horizon.
Assumes 6
- Desks act asynchronously, updating only when new information relevant to their oracle arrives. model
Source sentence
We assume that distinct desks do not act synchronously in time but act whenever new information relevant to their respective oracles arrives, which each desk samples over its own characteristic lookback horizon.
- Each desk moves its orders toward target distances using a first‑order gradient rule. model
Source sentence
We treat each market maker as a desk that uses a first-order gradient rule to move the distances of its orders from the mid price toward target distances provided by an unknown oracle.
- The model contains no takers; the mid price moves toward the side of the book with less volume near it. market
Source sentence
As our model has no takers, we let the mid price move toward the side of the book with less volume near it.
- Scaling regime: c = γ ¯c, Tk = τk γ, β = ¯β γ² (Assumption 3.2). model
Source sentence
There exist constants ¯c > 0, τ1 < · · · < τK and ¯β ≥0, independent of γ, such that c = γ ¯c, Tk = τk γ , β = ¯β γ2 .
Show 2 more
- Desks sample information over distinct characteristic look‑back horizons, inducing power‑law memory in order flow. model
Source sentence
We assume that distinct desks do not act synchronously in time but act whenever new information relevant to their respective oracles arrives, which each desk samples over its own characteristic lookback horizon. The distinct horizons of the desks further induce a persistent memory in the order flow that decays as a power law.
- The order‑book volume is conserved irrespective of the noise driving it. model
Source sentence
As desks only redistribute volume, the volume of the order book is conserved irrespective of the nature of this noise.
Data
- Binance order‑book and trade data order-book public
Source sentence
We confirm the key predictions of the model on simulated agents and on six months of Binance data with about 790 million orders. Data The relations above need the activity over time scales from one second to one day.
- Simulated desks synthetic unknown
Source sentence
We confirm the key predictions of the model on simulated agents and on six months of Binance data with about 790 million orders. Sections 6 and 7 test the mechanism on simulated desks and on Binance data respectively.
Costs
Not applicable
Finds 5
- The model predicts rough volatility; simulations of desks reproduce roughness only when desks sample information over distinct horizons. theoretical
Source sentence
Sections 6 and 7 test the mechanism on simulated desks and on Binance data respectively. Simulations over a finite time reproduce each of these steps and give roughness only when the desks sample information over distinct horizons (Section 6).
- Empirical analysis on six Binance markets shows order‑flow and volatility Hurst exponents between 0.07 and 0.09, within 0.02 of each other. statistical
Source sentence
0.071 ± 0.006 0.50 ± 0.04 0.945 −0.0 0.088 / 0.096 In every market H(Nb) lies between 0.07 and 0.09 (Figure 2a). On six markets of Binance the order flow and the volatility are rough with H between 0.07 and 0.09 and within 0.02 of each other.
- Desks in the data move orders back within 12 to 21 seconds and shift volume between neighboring depths, matching the model’s finite‑distance‑from‑threshold behavior. other
Source sentence
In the books of these markets, market makers move orders back within 12 to 21 seconds and move volume between neighbouring depths. These data agree with our desks at a finite distance from the threshold (Section 7.2).
- The fitted kernel norm a has a 95 % lower bound of at least 0.94, placing the order flow close to the theoretical threshold for rough volatility. statistical
Source sentence
The 95% lower bound of the norm a is at least 0.94. The order flow thus lies close to the threshold of Theorem 5.3.
- Simulated desks with fitted tail parameters produce Hurst values that are on average 0.016–0.009 lower than the observed data, with 95 % intervals covering zero. performance
Source sentence
At distances 1 −a = T −α with T of 105 or 106 s, simulated desks with these tails give values of H that the data exceed by 0.016 and 0.009 on average, with 95% intervals of [−0.005, 0.036] and [−0.012, 0.030] (Table 7).
Checked by others
Supported by
- Probabilistic models and statistics for electronic financial markets in the digital age
Compare the findings
This paper: Empirical analysis on six Binance markets shows order‑flow and volatility Hurst exponents between 0.07 and 0.09, within 0.02 of each other.
That paper: Empirical analysis of the S&P 500 volatility series yields an estimated Hurst exponent ˆH≈0.16, supporting the rough volatility paradigm.
- Roughness Analysis of Realized Volatility and VIX through Randomized Kolmogorov-Smirnov Distribution
Compare the findings
This paper: Empirical analysis on six Binance markets shows order‑flow and volatility Hurst exponents between 0.07 and 0.09, within 0.02 of each other.
That paper: Both volatility measures exhibit Hurst exponents significantly below 1/2, confirming pronounced roughness in volatility dynamics.
Extracted 2026-11-19 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (73)
- Volatility is rough (2018)
- Rough fractional diffusions as scaling limits of nearly unstable heavy tailed H awkes processes (2016)
- The microstructural foundations of leverage effect and rough volatility (2018)
- The characteristic function of rough H eston models (2019)
- No-arbitrage implies power-law market impact and rough volatility (2020)
- Conservative SPDE s as fluctuating mean field limits of stochastic gradient descent (2025) — link
- L angevin equation for the density of a system of interacting L angevin processes (1996)
- Stochastic model of slow dynamics in supercooled liquids and dense colloidal suspensions (1994)
and 65 more
- Rough volatility: fact or artefact? (2024) — link
- Advances in Financial Machine Learning (2018)
- A fully consistent, minimal model for non-linear market impact (2015)
- Trades, Quotes and Prices: Financial Markets Under the Microscope (2018)
- Spectra of some self-exciting and mutually exciting point processes (1971)
- Hawkes processes in finance (2015)
- Critical reflexivity in financial markets: a H awkes process analysis (2013)
- Deep hedging (2019)
- Mean field analysis of neural networks: a law of large numbers (2020)
- A mean field view of the landscape of two-layer neural networks (2018)
- Statistical theory of the continuous double auction (2003)
- A stochastic model for order book dynamics (2010)
- Volatility has to be rough (2021)
- High frequency trading and the new market makers (2013)
- High-frequency trading in a limit order book (2008)
- Dealing with the inventory risk: a solution to the market making problem (2013)
- Scaling and criticality in a stochastic multi-agent model of a financial market (1999)
- Herd behavior and aggregate fluctuations in financial markets (2000)
- A simulation analysis of the microstructure of double auction markets (2002)
- Statistical properties of stock order books: empirical results and models (2002)
- Financial B rownian particle in the layered order-book fluid and fluctuation-dissipation relations (2014)
- Agent-based models for latent liquidity and concave price impact (2014)
- Anomalous price impact and the critical nature of liquidity in financial markets (2011)
- Trainability and accuracy of artificial neural networks: an interacting particle system approach (2022)
- Derivation of the B oltzmann equation for financial B rownian motion: direct observation of the collective motion of high-frequency traders (2018) — link
- The micro-price: a high-frequency estimator of future prices (2018) — link
- The price impact of order book events (2014) — link
- Queue imbalance as a one-tick-ahead price predictor in a limit order book (2015) — link
- Enhancing trading strategies with order book signals (2018) — link
- Multi-level order-flow imbalance in a limit order book (2020) — link
- Long memory relationships and the aggregation of dynamic models (1980)
- A simple approximate long-memory model of realized volatility (2009) — link
- Buy low, sell high: a high frequency trading perspective (2014) — link
- Quantifying reflexivity in financial markets: towards a prediction of flash crashes (2012) — link
- Volatilities of different time resolutions – analyzing the dynamics of market components (1997)
- Optimization methods for large-scale machine learning (2018) — link
- Technology and liquidity provision: the blurring of traditional definitions (2009)
- Stochastic Approximation and Recursive Algorithms and Applications (2003)
- Asymptotic properties of stochastic approximations with constant coefficients (1981)
- Stochastic modified equations and dynamics of stochastic gradient algorithms I : Mathematical foundations (2019)
- A mathematical approach to order book modeling (2013) — link
- Price dynamics in a M arkovian limit order market (2013) — link
- Simulating and analyzing order book data: the queue-reactive model (2015) — link
- Limit order books (2013) — link
- Limit order books, diffusion approximations and reflected SPDE s: from microscopic to macroscopic models (2020) — link
- Finite Volume Methods for Hyperbolic Problems (2002) — link
- Brownian Motion (2010)
- Continuous Martingales and B rownian Motion (1999) — link
- The stochastic F ubini theorem revisited (2012)
- Asymptotic properties of stochastic approximations with constant coefficients (1979)
- Commission Delegated Regulation ( EU ) 2017/566 of 18 May 2016 supplementing Directive 2014/65/ EU with regard to regulatory technical standards for the ratio of unexecuted orders to transactions in order to prevent disorderly trading conditions (2017)
- Hawkes and INAR ($\infty$) processes (2016) — link
- Discrete-scale invariance and complex dimensions (1998) — link
- Regular Variation (1987)
- An Introduction to Probability Theory and Its Applications (1971)
- Bitcoin time-of-day, day-of-week and month-of-year effects in returns and trading volume (2019)
- Periodicity in Cryptocurrency Volatility and Liquidity (2024)
- Limit theorems for nearly unstable H awkes processes (2015) — link
- More statistical properties of order books and price impact (2003)
- Empirical shape function of limit-order books in the C hinese stock market (2008)
- Binance Public Data and API documentation
- Does the term structure of the at-the-money skew really follow a power law? (2023)
- Decoupling the short- and long-term behavior of stochastic volatility (2022) — link
- Rough volatility: Evidence from option prices (2018) — link
- Rough volatility of B itcoin (2020) — link
0 of 73 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.