Paper: arXiv 2509.17715
Authors: Axel Ciceri, Austin Cottrell, Joshua Freeland, Daniel Fry, Hirotoshi Hirai, Philip Intallura, Hwajung Kang, Chee-Kong Lee, Abhijit Mitra, Kentaro Ohno, Das Pemmaraju, Manuel Proissl, Brian Quanz, Del Rajan, Noriaki Shimada, Kavitha Yograj
Abstract
The estimation of fill probabilities for trade orders represents a key ingredient in the optimization of algorithmic trading strategies. It is bound by the complex dynamics of financial markets with inherent uncertainties, and the limitations of models aiming to learn from multivariate financial time series that often exhibit stochastic properties with hidden temporal patterns. In this paper, we focus on algorithmic responses to trade inquiries in the corporate bond market and investigate fill probability estimation errors of common machine learning models when given real production-scale intraday trade event data, transformed by a quantum algorithm running on IBM Heron processors, as well as on noiseless quantum simulators for comparison. We introduce a framework to embed these quantum-generated data transforms as a decoupled offline component that can be selectively queried by models in low-latency institutional trade optimization settings. A trade execution backtesting method is employed to evaluate the fill prediction performance of these models in relation to their input data. We observe a relative gain of up to ~ 34% in out-of-sample test scores for those models with access to quantum hardware-transformed data over those using the original trading data or transforms by noiseless quantum simulation. These empirical results suggest that the inherent noise in current quantum hardware contributes to this effect and motivates further studies. Our work demonstrates the emerging potential of quantum computing as a complementary explorative tool in quantitative finance and encourages applied industry research towards practical applications in trading.
Complexity vs Empirical Score
- Math Complexity: 7.0/10
- Empirical Rigor: 8.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced quantum algorithms and statistical learning theory, with substantial mathematical formalism in the methodology section, justifying a high math complexity score. It is highly empirical, using real production-scale intraday trade data, a defined backtesting protocol, and reporting specific out-of-sample performance metrics (e.g., 34% gain), warranting a high empirical rigor score.
Research Flowchart
flowchart TD Start["Research Goal<br>Enhance Fill Probability<br>Estimation for Bond Trading"] --> DataPrep["Data Preparation<br>Intraday Trade Event Data"] DataPrep --> QCompute["Quantum Computing<br>IBM Heron Processors<br>Transform Data via VQE"] QCompute --> ModelTrain["Model Training<br>Machine Learning Models<br>Quantum vs. Original vs. Simulator"] ModelTrain --> Backtest["Evaluation<br>Trade Execution Backtesting<br>Out-of-Sample Test Scores"] Backtest --> Findings["Key Outcomes<br>34% Relative Gain with<br>Quantum Hardware Data<br>Noise as Beneficial Factor"]
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 framework that embeds quantum-generated data transforms as a decoupled offline component for low‑latency institutional trade optimization.
Source sentence
We introduce a framework to embed these quantum-generated data transforms as a decoupled offline component that can be selectively queried by models in low‑latency institutional trade optimization settings.
Assumes 5
- Training and test samples are assumed i.i.d. from the same distribution P(X,Y). statistical
Source sentence
Standard learning theory offers generalization guarantees with respect to making in‑distribution predictions, i.e., under the assumption that both observed training and unseen test samples are i.i.d. sampled from the same distribution P(X, Y)
- In practice the i.i.d. assumption does not hold for the market data. data
Source sentence
we cannot assume to have i.i.d. access to the full underlying distribution P at training time
- The quantum feature map ϕ is deterministic for simplicity. model
Source sentence
ϕ assumed deterministic for simplicity.
- There is a hidden distribution of trade inquiries and response windows are on the order of seconds. market
Source sentence
a hidden distribution of trade inquiries to selected market participants, inquiry response time windows at the order of seconds
Show 1 more
- Quantum‑generated features add no additional information beyond the original market event vector. model
Source sentence
ϕ adds no additional information, and is independent of market time and associated market event outcome label y
Data
- real production‑scale intraday trade event data other proprietary
Source sentence
real production‑scale intraday trade event data
- quantum‑generated feature dataset (IBM Heron processors) synthetic proprietary
Source sentence
quantum algorithm running on IBM Heron processors
Costs
Unclear
Finds 3
- Models using quantum hardware‑transformed data achieve up to ~34% relative gain in out‑of‑sample test scores compared to models using original or noiseless‑simulated data. performance
Source sentence
We observe a relative gain of up to ∼34% in out‑of‑sample test scores for those models with access to quantum hardware‑transformed data over those using the original trading data or transforms by noiseless quantum simulation.
- Classical‑input models show a flat AUC profile with an average test AUC of about 0.63 ± 0.04 over up to four trading days of blinding. predictive-accuracy
Source sentence
average test AUC of ∼0.63 ± 0.04 over up to 4 trading days of blinding, which may be interpreted as a lack of time dependence
- Significant uplifts of out‑of‑sample test scores are observed for models with noisy quantum hardware‑generated inputs versus noiseless simulation or classical inputs. performance
Source sentence
The model performance results for respectively generated fill probability predictions, compared to actually realized trade outcomes, show significant uplifts of out‑of‑sample test scores for models with noisy quantum hardware‑generated inputs over those with noise‑less quantum simulation or classical inputs.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 15 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.