Research systems should preserve both what survived testing and what did not.
Kelex Quant Lab develops structures for defining claims, declaring decision criteria, preserving evidence and recording why an investigation continued, changed or stopped.
The initiative focuses on the validity of the process used to reach a conclusion—not only whether the result appears positive.
Quant Lab is not presented as a trading product, investment fund or financial-advice service.
A research and decision-systems initiative within Kelex.
Kelex Quant Lab focuses on how claims are formed, tested, revised, evaluated and preserved. Its work is concerned with the structure of a dependable investigation and the quality of the record left behind.
The initiative treats research as an operating system: a connected relationship between hypotheses, methods, criteria, evidence, changes, decisions and retained records.
The objective is not to make every claim survive. It is to make the conclusion examinable.
Quant Lab is currently developing its research structures and public record. The website should communicate the discipline of the initiative without inventing outputs that have not been published.
The result of a claim and the validity of the process are different questions.
A claim may be rejected after a valid investigation. An apparently favourable result may still be unusable when the method was invalid, the criteria changed after the result or the evidence cannot be traced.
Was the investigation valid enough to justify a conclusion?
- Was the claim explicit?
- Was the method declared?
- Were decision criteria established before results were known?
- Was the test operated as described?
- Were changes and exceptions recorded?
- Is the evidence traceable?
- Can another reviewer understand the reasoning?
What does the available evidence support?
- Does the evidence support the claim?
- Does it require revision?
- Does it justify rejection?
- Is the claim unresolved?
- Should the opportunity close?
- Is another test justified?
A dependable investigation leaves connected records.
A conclusion should remain connected to the claim, method, criteria, evidence and changes that produced it.
The proposition being investigated and the conditions under which it is expected to hold.
The workflow, participants, environment and constraints surrounding the investigation.
The process used to operate the test and collect evidence.
The declared conditions for continuation, revision, rejection, closure or further testing.
The observations, behaviour, outcomes, exceptions and limitations preserved during the work.
Any revision to the claim, method, criteria or operating conditions.
An assessment of whether the investigation was conducted validly enough to support a conclusion.
The conclusion supported by the available evidence under the declared conditions.
The selected next action: continue, revise, reject, close or replicate.
The reasoning that connects the evidence and criteria to the decision.
Four states clarify what the result can—and cannot—mean.
Process validity and claim outcome must remain visible as separate dimensions.
The declared method was followed and the available evidence supports the tested claim under the stated conditions.
The declared method was followed, but the evidence does not sufficiently support the tested claim.
The result may appear favourable, but the process cannot justify a dependable conclusion.
The available record is insufficient to support acceptance or rejection.
Decision boundaries should exist before the result.
Criteria established after a result is known can turn an investigation into a search for justification. Quant Lab treats declared boundaries as part of the research record.
- What evidence would be sufficient to support the claim under the declared conditions?
- What evidence would show that the problem may remain valid but the claim, method or workflow must change?
- What result would show that the tested claim is not sufficiently supported?
- What condition would show that the opportunity no longer justifies active investigation?
- What result requires another test before the conclusion can be treated as dependable?
- What missing record, process failure or evidence limitation prevents a conclusion?
A result should be separated from the conditions that produced it.
Replication is not required for every simple operational question. It becomes important when a claim depends on repeatability, stability or transfer across conditions.
- Can the method be operated again?
- Are the relevant conditions defined?
- Can the evidence be collected consistently?
- Were material changes recorded?
- Does the result survive a credible repeat?
- Does a changed context require a revised claim?
Revision is not the same as quietly changing the original claim. The prior version, the reason for change and the new decision criteria should remain visible.
A revised investigation should preserve the path that made revision necessary.
Work should not disappear because the outcome was inconvenient.
A serious registry retains investigations across different outcomes so that claims, changes and decisions remain available for later review.
- The available evidence supports the claim under the declared conditions.
- The claim, method, criteria or operating context changed and the prior version remains preserved.
- The tested claim was not sufficiently supported.
- The opportunity no longer justifies active investigation under the current conditions.
- The evidence or process record is insufficient for a dependable conclusion.
- Another test is required before the conclusion can be treated as sufficiently dependable.
A registry should preserve relationships between claims, versions, evidence, decisions and later replications.
The initiative is developing its structures and public record.
Kelex Quant Lab is currently building the systems required to define, operate, review and preserve investigations consistently.
The website should communicate this stage honestly. It should not imply that a mature publication archive, completed experiment portfolio or established performance record already exists.
- Claim and hypothesis records
- Decision-criteria templates
- Method and change records
- Evidence-preservation structures
- Process-validity review
- Decision records
- Registry relationships
- Replication records
- Public-record standards
The absence of published outputs should not be filled with invented results.
Relevant collaboration begins with capability, access or a difficult research question.
Quant Lab may be relevant to researchers, technical collaborators and operators who can strengthen the design, execution, evaluation or preservation of a credible investigation.
Methods, experimental structure, validity review or replication design.
Registries, evidence structures, data integrity, tooling or reproducible workflows.
Specialist understanding required to define a credible claim or interpret operating conditions.
A real workflow, participants, records or pilot environment in which an investigation can be conducted.
Independent challenge, methodological review or examination of the decision record.
Structures that make claims, methods, evidence and revisions easier to inspect.
Bring a research capability, a credible workflow or a difficult claim.
A useful proposal should explain what can be investigated, what access exists and how the collaboration may improve the quality of the process or record.