Checksums, normalization, coverage and accepted observations.
INDEPENDENT RESEARCH / ENGINEERING NOTES
Work worth
Work worth
examining.
Quantitative research is only as useful as the engineering beneath it. These publications document what we built, how assumptions are tested, and how results become evidence that a client can evaluate.
RESEARCH
PAPER 001EDGAR / QUANT
PAPER 001EDGAR / QUANT
OCT / 2026From Market
Data to
Defensible
Research.
Data to
Defensible
Research.
BacktestApp engineering & CSL research workflow
CAUSALITY / EXECUTION / ACCOUNTING01 — 09
VALIDATED DATA CONTRACTSDETERMINISTIC REPLAYEXECUTION EVIDENCERESEARCH GOVERNANCE
SOFTWARE EVIDENCE / BACKTESTAPPSee the research software
Explore software case study ↗See the research software
behind the publication.
Explore a privacy-safe illustrated interface study informed by five captured screens of the BacktestApp workspace, run configuration, results and library.
FEATURED PUBLICATION / 001PUBLIC EDITION · 10 OCTOBER 2026
QUANTITATIVE SOFTWARE ENGINEERING · TECHNICAL CASE STUDY
From Market Data
From Market Data
to Defensible Research
BacktestApp and CSL are two distinct research workstreams connected by one engineering discipline: make input identity, decision timing, modeled execution, accounting, and evaluation history inspectable.
01 / THE SYSTEMDeterministic historical research application
Verified data handling, causal rule evaluation, versioned execution and ledger-based accounting.
02 / THE THINKINGChallenge hypotheses, preserve failures
Separate what is measured, what is assumed and what a result actually supports.
03 / THE VALUEReviewable engineering deliverables
Data checks, temporal reviews, cost analysis and reproducible reports for real project decisions.
OUR METHOD / VISIBLE AT A GLANCECONCEPTUAL ASSURANCE MODEL
An answer is only as strong
as its weakest assumption.
Closed observations, availability and prefix invariance.
Gaps, costs, activation rules and ambiguity handling.
Cash, holdings, fees, equity and traceable artifacts.
Frozen assumptions, evaluation usage and recorded outcomes.
These are assurance responsibilities, not a disclosure of proprietary feature pipelines or trading signals.
WHY CLIENTS CAREBuilt to make
Built to make
complex work legible.
Trace the decision
Show exactly which assumptions a conclusion depends on.
Reconcile the economics
Make financial arithmetic independently checkable.
Respect the evidence
Distinguish source inspection, archived test records and fresh bounded checks.
Deliver usable work
Turn unclear research problems into scoped reports and software milestones.
ENGINEERING COLLABORATIONHave a research result
Have a research result
worth understanding?
Discuss dataset quality, historical execution assumptions, backtest integrity, or reconciliation. We can start with a bounded review instead of a large commitment.