Methodology
Explicit assumptions.
Versioned research.
QuantMetrics Lab develops quantitative indicators and research frameworks across market structure, liquidity and macro-financial questions. Different models require different methods, but they share a common research discipline: define the question, document the information set, preserve model vintages and keep interpretation separate from prediction.
Common research principles
A framework is useful only if its assumptions remain visible.
Define what is being measured
Each framework starts from a specific research question before selecting data, transformations or model structure.
Avoid hidden look-ahead
Where historical vintages matter, later information should not be silently inserted into earlier model states.
Separate model and software changes
Methodological revisions, implementation updates and historical outputs are treated as distinct objects.
Explain what the number means
Outputs are accompanied by definitions, scale interpretation and limitations rather than presented as unexplained scores.
Limit the claim
Small samples, structural breaks, measurement error and model uncertainty remain part of the interpretation.
Model output is not a trade signal
An index value, similarity measure or projected model window is kept conceptually separate from investment advice.
Research workflow
From question to production output.
QuantMetrics Lab does not impose a single statistical template on every project. A cycle-timing model, a historical-similarity framework and a global monetary aggregate require different data and different validation logic. What remains common is the sequence used to turn a research idea into a public production output.
Define the object, unit of analysis and claim the framework is allowed to make.
Select sources, timestamps, transformations and coverage rules appropriate to the question.
Specify the calculation, assumptions, parameters and information available to each vintage.
Check internal consistency, sensitivity, historical behaviour and known failure modes.
Generate stable machine-readable and public-facing outputs from a defined software version.
Introduce new methodology only through an explicit version rather than silently rewriting history.
Historical vintages
The model should not know the future.
Preserving the information available at the time
Some QuantMetrics frameworks depend on historical observations that accumulate over time. In those cases, reconstructing the past with today’s complete dataset can create an artificial impression of foresight.
QuantMetrics Lab therefore treats the information set available to a model version as part of the methodology itself. When new observations justify a different model, the preferred approach is to introduce a new methodological vintage and preserve the earlier one.
Data handling
Source choice is part of the model.
Data availability, frequency, definitions and revision policies differ across research areas. For that reason, every framework can define its own source hierarchy and transformation rules.
- Temporal alignment: observations are assigned to explicitly defined dates or periods.
- Units and transformations: conversions, normalisation and aggregation are documented where they materially affect interpretation.
- Missing observations: interpolation or carry-forward rules are used only when justified by the specific framework.
- Source revisions: revised public data may update future production outputs without retroactively changing a preserved methodological vintage unless stated.
- Third-party data: provider-specific limitations remain part of the data-quality assessment.
Validation & robustness
Evidence should survive reasonable alternatives.
Validation depends on the research question. It may include sensitivity analysis, alternative parameterisations, historical reconstruction, multiple seeds or simulations, percentile comparisons, robustness checks, agreement statistics or comparisons with alternative data definitions.
A robustness exercise is not used to manufacture certainty. Its purpose is to understand which conclusions are stable, which depend on assumptions and where the framework may fail.
Indicator-specific documentation
Shared principles, different methodologies.
Cycle position
Time-based cycle framework with explicit historical model vintages and empirical ATH/ATL timing windows.
Open BCPI →Historical similarity
Multi-timeframe comparison of current return structure with homologous historical windows.
Open BCSI →Liquidity compression
Visible liquidity framework combining exchange and OTC reserve estimates within a monthly production model.
Open BLCI →…and more
Additional monetary, macro-financial and quantitative frameworks are documented individually as they enter production.
Research library →Reproducibility & disclosure
Document enough to inspect the claim.
Public documentation is intended to make the research question, interpretation, material assumptions and limitations inspectable. The level of public code and raw-data availability can differ by project because of licensing, source restrictions or the stage of the research.
Where repositories, datasets, papers or technical notes are publicly available, they are linked from the Research library. The absence of a public code repository should not be interpreted as a claim of full reproducibility from the website alone.