How to defend a QR research project
A five-part evidence checklist for methods, results, bias, limitations and clear communication.
Lead with a claim sized to the evidence
In a research interview, explain the question, data and target in one sentence. Then say what observation would count against your hypothesis. A small synthetic example can establish that code runs; it cannot establish a market edge.
Bring the baseline and timeline
Show why your comparison is meaningful. An unconditional mean forecast, a no-signal strategy, or an analytic option price may be suitable depending on the project. Explain which observations were available when each choice was made. Record any holdout reveal that affected later decisions.
Discuss failure without hiding it
Report model and baseline metrics together, including costs when a trading strategy is actually simulated. For a yield-change forecast or model option price, a trading-cost figure is not meaningful. Sensitivity, sampling error, missing timestamps and model assumptions give the interviewer useful places to probe.
Export a reproducible project, take a timed practice, and use the text defense for up to five follow-up questions about your own saved brief. The feedback is a practice tool, not evidence that a hiring panel would make the same judgment.