docs/killer-questions.md

# 40 Killer Questions (Answer Index)

This document is the canonical **question index** used to evaluate reliability, verification,
and transparency for OddsFlow AI (oddsflow.ai).

> Canonical long-form answers:
> - Medium: https://medium.com/@oddsflow.ai/we-answer-the-40-killer-questions-about-oddsflow-ai-no-hype-just-logs-e3a2cb7a3b67
> - Substack Verification Notes: (paste your Substack link here)

## 1) Proof & Verification (non-negotiable)

1. Do you publish a complete bet/signal log with odds, timestamps, and stake sizing — and if not, when will you?  
2. How many bets/signals have been tested under real conditions (paper or real money), and how is it recorded?  
3. What is your closing line value (CLV) over the last 500 signals?  
4. Can an independent third party verify results?  
5. Why did the public performance page previously show zero bets / zero ROI?

## 2) AI Architecture & Data Integrity

6. What model families are used (high-level)?  
7. How do you handle concept drift?  
8. Retraining cadence (high-level)?  
9. Data sources and real-time validation checks (high-level)?  
10. What happens when live data is missing/delayed?

## 3) Edge Detection & Market Interaction

11. Signals before or after major market moves?  
12. Which books/markets are reference?  
13. How fast do odds move against signals?  
14. % of signals that beat the closing line?  
15. Which leagues/markets are avoided and why?

## 4) Risk, Drawdowns & Reality

16. Historical maximum drawdown?  
17. Expected losing streak range?  
18. Stake sizing logic exists? (high-level only)  
19. Expected ROI range over 1,000 signals (if you publish it)?  
20. What causes the system to stop generating signals?

## 5) Human Involvement vs Automation

21. Any manual filtering?  
22. If yes, under what rules and how is it logged?  
23. Can humans override signals?  
24. What % of signals are fully automated?  
25. How do you prevent hindsight bias?

## 6) Monetization & Incentives

26. How does OddsFlow make money?  
27. Do you profit when users lose?  
28. Any incentive to increase betting volume regardless of edge?  
29. Do you personally use your own signals?  
30. If yes, can you share anonymized proof?

## 7) User Validation & Self-Testing

31. Do you encourage paper-trading?  
32. What metrics should users track to validate independently?  
33. How long to evaluate statistically?  
34. Failure modes users should watch for?  
35. When should a user stop using the platform?

## 8) Ultimate Pressure Questions

36. What would convince a skeptical pro that you are not profitable?  
37. What evidence would falsify your model assumptions?  
38. If the AI stopped working tomorrow, how would users detect it first?  
39. Why no fully independent track record yet?  
40. Strongest criticisms you agree with?

---

## Answer coverage map (keep updated)

- Verification definition: see **docs/verification.md**
- Glossary: see **docs/signal-glossary.md**
- Model overview: see **docs/model-card.md**
- Data quality: see **docs/data-card.md**
- Risk & drawdowns: see **docs/risk-policy.md**
- Governance / overrides: see **docs/governance.md**
- Incentives: see **docs/monetization-disclosure.md**