Paper: arXiv 2503.18199
Authors: Guillaume Maitrier, Grégoire Loeper, Jean-Philippe Bouchaud
Abstract
This paper introduces a novel algorithm for generating realistic metaorders from public trade data, addressing a longstanding challenge in price impact research that has traditionally relied on proprietary datasets. Our method effectively recovers all established stylized facts of metaorders impact, such as the Square Root Law, the concave profile during metaorder execution, and the post-execution decay. This algorithm not only overcomes the dependence on proprietary data, a major barrier to research reproducibility, but also enables the creation of larger and more robust datasets that may increase the quality of empirical studies. Our findings strongly suggest that average realized short-term price impact is not due to information revelation (as in the Kyle framework) but has a mechanical origin which could explain the universality of the Square Root Law.
Complexity vs Empirical Score
- Math Complexity: 6.5/10
- Empirical Rigor: 6.0/10 (capped from 7.2: transaction costs not modelled; no out-of-sample test; see penalty flags)
- Quadrant: Holy Grail — high math complexity, high empirical rigor (effective empirical rigor capped from 7.2)
Why this score: The paper involves algorithmic complexity and statistical modeling, requiring moderate mathematical sophistication, while being heavily data-driven with specific backtesting methodology and validation against established stylized facts.
Research Flowchart
flowchart TD
A["Research Goal: Develop algorithm<br>to generate realistic metaorders<br>from public trade data"] --> B["Methodology: Constrained Optimization<br>to recover metaorders"]
B --> C["Inputs: Public trade data<br>without labels"]
C --> D["Computation: Constrained optimization<br>to find non-overlapping metaorders"]
D --> E{"Verification against<br>stylized facts?"}
E -- Yes --> F["Key Findings: Model replicates<br>Square Root Law & impact profiles"]
E -- No --> B
F --> G["Conclusion: Impact is mechanical<br>not informational<br>(refuting Kyle model)"]
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 novel algorithm to generate realistic metaorders from publicly available trade data.
Source sentence
This paper introduces a novel algorithm for generating realistic metaorders from public trade data
Assumes 5
- A metaorder is defined as a sequence of trades of the same sign initiated by a given trader within the same trading session. other
Source sentence
We define a metaorder as a sequence of trades of the same sign initiated by a given trader within the same trading session.
- The mapping function can be specified with only two degrees of freedom (number of traders and their trading frequency) and still recover the correct impact. model
Source sentence
We gave this mapping function only two degrees of freedom: the number of different traders in a given day and the distribution of their trading frequency
- Public trade data (e.g., from a single exchange) contain enough information to reconstruct synthetic metaorders that preserve real‑world stylized facts. data
Source sentence
Our method effectively recovers all established stylized facts of metaorders impact
- Daily traded volume (VD) and intraday volatility (σD) can be computed and used as proxies for market liquidity and risk. market
Source sentence
Compute daily traded volume VD and intraday volatility σD, which we define for
Show 1 more
- Impact is independent of metaorder duration because the volatility contribution scales as √T while volume scales as T, cancelling T in the impact formula. other
Source sentence
the volatility contribution scales as √T whereas the total traded volume scales as T, which means that the explicit T dependence cancels from I(Q)
Data
- public trade data order-book public
Source sentence
Our method effectively recovers all established stylized facts of metaorders impact
- BNP Paribas share price data market-prices public
Source sentence
Synthetic metaorders were generated with BNP Paribas share price
Costs
Unclear - sources disagree
Source sentence (the authors' words)
Neglecting such costs can easily turn a profitable strategy on paper into a money losing machine
Finds 4
- The synthetic metaorders generated by the algorithm reproduce all major stylized facts: the Square Root Law, its independence from duration, concave execution profiles, and slow power‑law decay. statistical
Source sentence
Using this algorithm, we were able to recover all the salient stylized facts reported in the existing literature, specifically: the Square Root Law and its independence with respect to metaorder duration; the concave profile during metaorder execution; and the slow power-law decay after execution.
- Impact of synthetic metaorders is independent of metaorder duration, with an average prefactor Y ≈ 0.6 for durations longer than ~30 seconds. statistical
Source sentence
The blue dots represent the impact divided by the volume contribution as a function of duration, which is approximately constant for T ≳30 seconds, and on average equal to Y = 0.6.
- The post‑execution decay of synthetic metaorders matches the propagator model with decay exponent β ≈ 0.2. theoretical
Source sentence
Figure 5: Price relaxation post metaorder execution, well fitted by the impact predicted by the propagator model, with the value β ≈0.2.
- Synthetic metaorders show no correlation with short‑term predictive signals, supporting a mechanical origin of price impact. theoretical
Source sentence
By construction, a synthetic metaorder has no connection to a trader’s intention to trade based on a predictive signal. Yet, its impact is indistinguishable from that of a real metaorder
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 16 related papers). Coverage is partial, so this is neither replication nor refutation.
7 papers in the archive : Metaorder modelling and identification from public data ; When AI Trading Agents Compete: Adverse Selection of …
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 (44)
- Inferring microscopic financial information from the long memory in market-order flow: A quantitative test of the Lillo-Mike-Farmer model (2023)
- A fully consistent, minimal model for non-linear market impact (2015)
- Market impact with multi-timescale liquidity (2018)
- The two square root laws of market impact and the role of sophisticated market participants — → 2311.18283
- The Square-Root Impace Law Also Holds for Option Markets (2016)
- Institutional Investors and Stock Market Volatility* (2006) — link
- Universal scaling and nonlinearity of aggregate price impact in financial markets (2018)
- How efficiency shapes market impact (2013)
and 36 more
- A Bayesian theory of market impact — → 2303.08867
- Market impacts and the life cycle of investors orders (2015)
- Market impact: A systematic study of limit orders (2017)
- The" double" square-root law: Evidence for the mechanical origin of market impact using Tokyo Stock Exchange data — → 2502.16246
- Does the square-root price impact law belong to the strict universal scalings?: quantitative support by a complete survey of the Tokyo stock exchange market — → 2411.13965
- Anomalous price impact and the critical nature of liquidity in financial markets (2011)
- Trades, quotes and prices: financial markets under the microscope (2018)
- Slow decay of impact in equity markets: insights from the ANcerno database (2018)
- Exactly solvable model of the square-root price impact dynamics under the long-range market-order correlation — → 2502.17906
- Crossover from Linear to Square-Root Market Impact (2019) — link
- Why is the estimation of metaorder impact with public market data so challenging? — → 2501.17096
- “Microstructure Modes” – Disentangling the Joint Dynamics of Prices & Order Flow — → 2405.10654
- in preparation (2025)
- Continuous auctions and insider trading (1985)
- Online learning of order flow and market impact with Bayesian change-point detection methods — → 2307.02375
- Quantitative statistical analysis of order-splitting behavior of individual trading accounts in the Japanese stock market over nine years (2023) — link
- A million metaorder analysis of market impact on the Bitcoin (2015)
- Direct estimation of equity market impact (2005)
- Trading cost: the critical link between investment information and results (1983)
- Active portfolio management (2000)
- Handbook of price impact modeling (2023)
- The Cost of Misspecifying Price Impact — → 2306.00599
- Theory for long memory in supply and demand (2005)
- Limit order book simulation with generative adversarial networks (2023)
- Learning to simulate realistic limit order book markets from data as a world agent (2022)
- Generative ai for end-to-end limit order book modelling: A token-level autoregressive generative model of message flow using a deep state space network — → 2309.00638
- Crossover from linear to square-root market impact (2019)
- Trading costs (2018)
- The inelastic market hypothesis: a microstructural interpretation (2022)
- Market impact and trading profile of hidden orders in stock markets (2009)
- Empirical market microstructure: The institutions, economics, and econometrics of securities trading (2007)
- Anomalous Impact in Reaction-Diffusion Financial Models (2014) — link
- From Walras’ auctioneer to continuous time double auctions: A general dynamic theory of supply and demand (2016)
- Price Impact (2010)
- In search of the origins of financial fluctuations: The inelastic markets hypothesis (2021)
- Fluctuations and response in financial markets: thesubtle nature ofrandom’price changes (2003)
10 of 44 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.