Start with a question, not a position
Our proposed research process begins with a narrowly framed question about global liquid markets. We would specify the assets, the time horizon and the condition being examined before choosing a test. The memo should explain why an opportunity might exist, who may bear its cost and what would make the explanation implausible. An attractive narrative alone would not justify a decision.
We would separate an observation from an explanation and from a proposed allocation rule. These are different claims. A relationship in a dataset may be worth investigating without yet supporting either a causal explanation or an implementable investment approach.
Make the evidence inspectable
For each input, we plan to record its source, retrieval date, licence, revision history and the point at which it became available. Missing observations, changes in definitions and adjustments should be visible in the research record. The intended standard is that another reviewer can reconstruct both the dataset and the choices made around it.
The test plan would be written before interpreting its results. It would specify the comparison, sample boundaries, assumed costs and conditions under which the hypothesis should be rejected. Data reserved for evaluation should remain separate from data used to develop the rule. Any subsequent change would create a new version rather than overwrite the original.
Look for reasons to stop
We intend to ask a reviewer to develop the strongest alternative explanation and identify evidence that conflicts with the thesis. A proposed test might examine a different market condition, a more conservative cost assumption or the removal of a single influential period. The purpose is to understand how dependent a conclusion is on its supporting choices.
An inconclusive or failed test would remain part of the record. We would not erase an unfavourable result, quietly redefine the question or treat repeated searching as independent confirmation. If essential data cannot be checked, the appropriate research status would be unresolved.
Preserve a decision trail
The closing memo would connect the question, data version, test settings and result to a specific next step: stop, revise or investigate further. It would name the person responsible for review and set a review date. This is a proposed working discipline, not a description of completed investment tests.
Backtesting is hypothetical, as the Investor.gov performance bulletin explains. Our inference from that distinction is procedural: a promising historical test would support further scrutiny, not a claim of realised performance. No investment recommendation or completed result is presented here.
THE ESSENTIAL POINT
A hypothesis becomes useful when its rejection conditions are as clear as its supporting argument.
Sources & context
Supports the distinction between hypothetical backtests and actual performance. The research workflow above is Arccerno’s proposed methodology.