Independent Quantitative Finance Research Laboratory

Advancing quantitative finance through mathematics, statistics, stochastic modelling, scientific computing, and artificial intelligence.

Alpha Stochastic Research (ASR) is an independent quantitative finance research laboratory dedicated to advancing rigorous, transparent, and reproducible research at the intersection of financial markets, mathematics, statistics, stochastic modelling, scientific computing, and artificial intelligence.

ASR develops research projects, technical articles, educational resources, numerical experiments, open-source software, and reproducible publications designed to bridge academic theory and real-world financial applications.

Our work focuses on transforming complex financial data into robust models, measurable insights, reusable scientific tools, and evidence-based financial decisions.

Our mission is to advance quantitative finance through open, rigorous, and reproducible scientific research.

We aim to build a research ecosystem where mathematical theory, statistical inference, computational modelling, and financial engineering converge to support a better understanding of markets, risk, investment, and decision-making.

ASR follows a simple research framework:

Research → Modelling → Analysis → Impact

ASR is built around a set of core scientific principles.

Publication status: ASR currently publishes and archives research outputs through Zenodo. Software is released through GitHub and, where applicable, distributed through PyPI. SSRN dissemination is planned for selected working papers, but no SSRN identifier is listed until a public SSRN record exists.

asr-deep-hedging is an auditable NumPy research framework for discrete-time neural option hedging under transaction costs.

The project combines:

• exact finite-sample empirical CVaR estimation and tail weighting;
• Geometric Brownian Motion and full-truncation Heston simulation;
• proportional and quadratic transaction-cost models;
• state-only and inventory-aware neural hedging policies;
• manually implemented NumPy forward and backward passes;
• pathwise hedging-loss gradients;
• Black–Scholes and cost-aware benchmark strategies;
• evaluation metrics, paired-bootstrap comparisons, tests, and reproducible experiment configurations.

Version 0.2.0 aligns the public Python API with the documented examples. Core training, optimization, benchmark, evaluation, risk, simulation, feature, and inventory-policy utilities can now be imported directly from deep_hedging, while the original module-level imports remain available for backward compatibility.

python -m pip install --upgrade asr-deep-hedging

from deep_hedging import (
    Adam,
    TanhMLP,
    black_scholes_delta,
    evaluate_positions,
    simulate_gbm,
    train_step,
)

Scope: The software is intended for methodological research, numerical validation, reproducibility, and education. It is not a live-trading system or investment recommendation.

ASR projects are organized around research, implementation, documentation, publication, and education.

Alpha Stochastic Research
│
├── Research Reproductions
├── Quantitative Finance Libraries
├── Scientific Computing Projects
├── Educational Notebooks
├── Technical Reports
├── Literature Reviews
├── Open-Source Implementations
└── Research Publications

The shared open-science ecosystem is coordinated through:

• asr-open-sc
• asr-deep-hedging
• the ASR Zenodo community
• the ASR GitHub organization

ASR publishes research, software, and educational content through several channels.

ASR is currently focused on building a coherent open-source research foundation in quantitative finance.

ASR repositories are designed to follow consistent open-source research standards.

Each project should include, when applicable:

.
├── .github/
│   ├── ISSUE_TEMPLATE/
│   └── workflows/
├── docs/
├── examples/
├── figures/
├── notebooks/
├── paper/
├── src/
├── tests/
├── CITATION.cff
├── CONTRIBUTING.md
├── LICENSE
├── README.md
├── REPRODUCIBILITY.md
├── pyproject.toml
└── requirements.txt

ASR welcomes thoughtful contributions that improve the scientific, educational, or technical quality of its projects.

Contributions may include:

• Mathematical corrections
• Code improvements
• Documentation enhancements
• Numerical experiments
• Literature references
• Reproducibility improvements
• Research proposals
• Educational notebooks

Please read the contribution guidelines before opening a pull request.

Our vision is to build a trusted independent research laboratory where mathematics, statistics, computer science, and financial engineering converge to transform data into insight and insight into better financial decisions.

ASR aims to contribute to a more transparent, reproducible, and intellectually rigorous quantitative finance ecosystem.

Alpha Stochastic Research
Independent Quantitative Finance Research Laboratory

Research. Modelling. Analysis. Impact.

© 2026 Alpha Stochastic Research