Evidence before conclusions

A traceable process from question definition to validation, risk review and communication

A disciplined process

A working framework for research; no completed model or project is implied by its description

01 /

Define the question

Identify the instrument, observation period and intended purpose. Separate mechanism explanations, market observations and investment hypotheses. State what the analysis can and cannot address

02 /

Validate the data

Record sources, timestamps, currencies and field definitions. Check missing values, duplicates and abnormal prices before combining datasets

03 /

Document assumptions

Explain variables, parameters and rationale. Include funding, premiums, holding periods, liquidity and costs where relevant

04 /

Test and compare

Use alternative scenarios and out-of-sample observations where appropriate. Check sensitivity, execution constraints and backtest bias

05 /

Review risk

Include market, liquidity, counterparty, technology and operational risks. Describe uncertainties that cannot be quantified reliably

06 /

Communicate and update

Distinguish facts, inferences and views. Record dates, sources, limitations and subsequent changes rather than treating historical results as future promises

What a research report should contain

The question, observation period and information date
Sources, definitions, assumptions and data limitations
Methods, alternatives, costs and implementation constraints
Interpretation, risks and conditions that could invalidate the result
References, review comments and update history

From a question to an analysis

A liquidity study first defines the position size and required exit period. It then examines spreads, depth, withdrawal conditions and stressed observations. Scenarios include execution costs and delayed access. The final report explains both what the evidence supports and where data cannot establish a reliable conclusion

SIX RESEARCH AREAS