Model M can be numerically correct and still be presented dishonestly. This note decides what the researcher may and may not do with data and with failed trials — as professional duties tied to this case, not as a statute catalogue. It links multiple testing and researcher degrees of freedom and survivorship discipline without inventing legal paragraph numbers or fines.
Learning objectives. By the end of this note you should be able to:
- refuse MNPI-tainted features for M's cross-section;
- keep losing trials in the register;
- refuse survivorship-biased history dressed as live performance;
- limit personal or client data to its stated purpose;
- recognise that a true backtest can still be an unethical presentation.
In this lesson
- The decision: duties on this desk
- Artifact: duty checks applied to M
- Limitation: ethics is not a substitute for validation
- Applying the four duties to concrete M choices
- Worked refusal scripts
- Desk checkpoint: ethics gate before slides
- Pitfall: true numbers, false frame
- Decision card: may / may not for this week
Before you start
- Communicating to traders, risk managers and investment committees
- Multiple Testing and Researcher Degrees of Freedom
- Corporate Actions and Survivorship
The full lesson, with worked examples and exercise solutions, is for members.
Read the full lessonNew to SelfMadeQuant.com? See membership options or browse the Professional Quant Practice syllabus.