Evidence first.
Application in mind.

We start with a clear economic or analytical hypothesis and the decision the research is intended to inform. The scope, assumptions and evaluation criteria are defined before interpreting results. This helps distinguish meaningful evidence from patterns that may reflect chance or choices made during the analysis.

We examine where the data comes from, what it covers and how it has been collected and transformed. Checks consider missing observations, inconsistent definitions and potential biases, including the use of information that would not have been available at the time. Material limitations are documented so conclusions stay within what the data can reasonably support.

We test whether findings remain credible across alternative assumptions, parameter choices and relevant market conditions. Where appropriate, validation uses out-of-sample data and comparisons with simpler benchmarks or competing explanations. The aim is to understand both the strength of the evidence and the circumstances in which it may fail.

We assess findings against the constraints of their intended use, including transaction costs, liquidity and execution requirements for trading applications. Data availability, system reliability and operational controls also shape what can be implemented and maintained. These considerations help determine whether a research result is practical, and where its use may need to be limited.

We identify the assumptions on which the conclusion depends and the evidence that would weaken or overturn it. Review criteria can include changes in data quality, market conditions, observed performance or implementation costs. When those conditions change, the research should be reassessed rather than treating the original conclusion as permanent.