Define the question and scope
Begin with the question the study is intended to answer. A fund study should identify the fund, share class, mandate and disclosure dates. Market-data analysis should define the instrument, observation window and measurement conventions. A strategy description, a portfolio disclosure and an observed trade each provide different evidence.
Preserve provenance
Record the source, version and reference date of each input, as well as when it was retrieved. Issuer filings and portfolio files may refer to different dates or valuation conventions. For market data, also document missing observations and asynchronous fields; distinguish quote snapshots, incremental updates and complete event histories.
Match test and claim
Descriptive analysis characterises the observed sample. Prediction requires a defined target and out-of-sample evaluation; causal analysis requires an identification strategy. The research design and interpretation should reflect those different objectives.
Include operating conditions
Document the costs and operating conditions relevant to the question. In execution research, these include information delay, order size, fees, queue position and unfilled orders. An evaluation of a decision must use only information available at that time.
Label status and revisions
Each publication identifies its format and status. Substantive changes to inputs, methods or conclusions should be recorded in the version history so readers can identify the result they are citing.
Send a question or correction