Educational content only — not financial advice

Most edges fail validation. That is the process working.

PROVENT5 · 1.5h · Technical Track

Reading results honestly — metrics, decay, when to kill

A validation report is only useful if you know how to read it without self-deception. The metrics that matter, how much data you actually need, equity-curve forensics, and clear rules for killing an edge versus promoting it.

In this module

  1. Why this module exists
  2. The metrics that actually matter
  3. Sample size and statistical significance
  4. Equity curve forensics
  5. Interpreting decay (IS → OOS)
  6. When to kill an edge
  7. When to promote toward PROVEN / DEPLOYED
  8. Where FreqEdge and the Quant Agent help
  9. Key takeaways
Practice

Read one validation report without bias

Open Edge Lab and pick an edge that has completed a Validate idea / held-out run. Use Quant Agent Narrate on the result. Then answer three questions in writing: (1) What is the IS-to-OOS decay? (2) Is the sample large enough to trust? (3) Kill, keep in VALIDATED, or promote toward PROVEN? Come back with your decision.

See how this runs on a private server

Also in Technical Track

Share this lessonShare on X
Sign in free to read the full lessonSee how this runs on a private server

Sign in free to read the full lesson

Email unlocks the whole path immediately. No drip.

Rather start at the beginning? Lesson F1 is free.

Educational content only — not financial advice.

Educational content only — not financial advice.