ALPHA STOCHASTIC
RESEARCH

ASR / PUBLISHED RESEARCH

Quantitative Research Under Real-World Constraints

ASRQuant 1.3.0 is an open quantitative-finance research platform developed by Alpha Stochastic Research to connect point-in-time data, market conventions, modelling, validation, risk, portfolio construction, backtesting and reproducible research artifacts within a single workflow. The release strengthens fixed-income and derivatives analytics, curv

25 September 2026Srijan Mishra · Alpha Kabinet TOUREQuantitative research

Abstract

ASRQuant 1.3.0 is an open quantitative-finance research platform developed by Alpha Stochastic Research to connect point-in-time data, market conventions, modelling, validation, risk, portfolio construction, backtesting and reproducible research artifacts within a single workflow. The release strengthens fixed-income and derivatives analytics, curve and credit infrastructure, statistical validation, strategy-selection diagnostics, experiment lineage and research reproducibility.

Research overview

Quantitative research failures often arise not from a single model equation but from breaks between stages: information may enter a backtest before it was available, calibrated structures may not reproduce their market inputs, strategy selection may be evaluated without accounting for repeated trials, and implementation assumptions may become detached from the reported result. This paper evaluates ASRQuant 1.3.0 as an attempt to make those interfaces explicit within one reproducible quantitative-finance research architecture. The release combines point-in-time data contracts, market conventions and calibration diagnostics, quote-space and credit risk, strategy-selection diagnostics, experiment lineage and chronology-aware backtesting. Evaluation uses deterministic fixtures, fixed-seed simulations, analytical identities, round-trip tests, calibration repricing, numerical invariants and regression tests. A separate internal release-candidate review reproduced and corrected three material numerical issues: key-rate DV01 partitioning on irregular tenor grids, scale consistency in Probabilistic and Deflated Sharpe Ratios, and implied-volatility inversion at the intrinsic boundary. Re-execution of the corrected artifact on Linux x86_64 / Python 3.13.5 produced 336 passing tests with no pytest skips and 77.36% statement coverage across src/asrquant; deterministic curve and CDS fixtures reprice to numerical tolerance, and the corrected key-rate construction reconciles aggregate key-rate DV01 with parallel DV01 to approximately 3.91 × 10−6% relative error on the review fixture. These results support software and research-process claims for the tested artifact, not investment performance or third-party model-validation independence. The principal contribution is therefore architectural: ASRQuant makes data availability, market conventions, calibration quality, model-selection risk, implementation assumptions, provenance and artifact dependencies explicit objects that can be inspected and reproduced across the research chain.

This page presents the publicly released research summary and overview. It does not constitute investment advice or a performance guarantee.