Paper: arXiv 2407.17401
Authors: Xavier Brouty, Matthieu Garcin, Hugo Roccaro
Abstract
Starting from a basic model in which the dynamic of the transaction prices is a geometric Brownian motion disrupted by a microstructure white noise, corresponding to the random alternation of bids and asks, we propose moment-based estimators along with their statistical properties. We then make the model more realistic by considering serial dependence: we assume a geometric fractional Brownian motion for the price, then an Ornstein-Uhlenbeck process for the microstructure noise. In these two cases of serial dependence, we propose again consistent and asymptotically normal estimators. All our estimators are compared on simulated data with existing approaches, such as Roll, Corwin-Schultz, Abdi-Ranaldo, or Ardia-Guidotti-Kroencke estimators.
Complexity vs Empirical Score
- Math Complexity: 7.5/10
- Empirical Rigor: 7.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper presents advanced stochastic processes (fractional Brownian motion, Ornstein-Uhlenbeck) and moment-based derivations with asymptotic normality proofs, scoring high in math; it rigorously tests estimators on both simulated and real data, comparing against established benchmarks, indicating strong empirical rigor.
Research Flowchart
flowchart TD G["Research Goal: Develop estimators for<br>bid-ask spreads incorporating serial dependence<br>(Fractional Brownian Motion & Ornstein-Uhlenbeck)"] --> M["Methodology: Moment-Based Estimation"] M --> D["Data: Simulated Financial Data<br>(GBM, fBM, OU processes)"] D --> C["Computation: Deriving & evaluating<br>statistical properties (Consistency, Asymptotic Normality)"] C --> E["Evaluation: Comparison with benchmarks<br>(Roll, Corwin-Schultz, Abdi-Ranaldo, Ardia-Guidotti-Kroencke)"] E --> F["Findings: New consistent estimators for<br>serially dependent models validated via simulation"]
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 moment‑based estimators for bid‑ask spreads, including consistent and asymptotically normal estimators under serial dependence.
Source sentence
we propose moment-based estimators along with their statistical properties.
Assumes 5
- The spread is constant over the estimation period (typically one day) and buys and sells are equally likely. market
Source sentence
we assume that the spread is constant over the estimation period, which is typically one day, and that buys and sells are equally likely.
- The mid‑price process p⋆t and the microstructure noise ηt are independent of each other. model
Source sentence
Assumption 1. The processes p⋆t and ηt are independent of each other.
- Non‑overlapping increments of p⋆t are independent (or uncorrelated) unless serial dependence is explicitly modelled. model
Source sentence
we successively examine the case where non‑overlapping increments of p⋆t are independent of each other, and the case where they are correlated, modelled by a log‑price following an fBm.
- When serial dependence of trades is considered, the microstructure noise follows a binarized Ornstein‑Uhlenbeck process with exponentially decaying autocorrelation. model
Source sentence
the microstructure noise ηt is autocorrelated, with an exponentially decaying autocorrelation function, stemming for instance from a mean‑reverting underlying process that follows the Ornstein‑Uhlenbeck model.
Show 1 more
- The observed price deviates from the unobserved mid‑price by half the bid‑ask spread. model
Source sentence
the observed price Pt, since it corresponds to the price of transactions, deviates from P⋆t by half the bid‑ask spread, S.
Data
- CAC 40 constituents (40 French stocks) market-prices public
Source sentence
Our dataset includes open, high, low, and close prices in one‑minute intervals, as well as bid‑ask spreads sampled every five minutes. All the data are in the interval between the 19th September 2022 and the 31st March 2023.
- Simulated data synthetic unknown
Source sentence
All our estimators are compared on simulated data with existing approaches, such as Roll, Corwin‑Schultz, Abdi‑Ranaldo, or Ardia‑Guidotti‑Kroencke estimators.
Costs
Not applicable
Finds 3
- In the real‑data application, the simple estimator bS1,1 achieves the lowest average MAPE (6.06%) and RMSE (0.613) among all estimators. performance
Source sentence
Table 5: Average RMSE and MAPE over the 40 constituents of the CAC 40 index, for various spread estimators.
- Estimators that filter serial dependence (bS2,1 and bS3,1) are more accurate than classical estimators for stocks with smaller market capitalisation. statistical
Source sentence
the estimators AGK1 and to a lesser extent bS1,1 seem to be a good choice whatever the capitalization, but bS2,1, bS3,1, and AGK2 improve their relative accuracy for smaller capitalizations.
- Simulation results show that each estimator performs well in the specific setting for which it was designed, outperforming classical estimators. theoretical
Source sentence
simulations show the good performance of each of these estimators in the context for which it has been built, compared to classical bid‑ask spread estimators.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 30 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.