Fix DCF valuation producing unrealistic values

[majiayu000]

Summary

Fixes #431 - DCF analysis was producing unrealistic valuations (e.g., $16T for OKTA).

Root causes identified:

• Terminal value calculation exploded when WACC-terminal growth spread was too small
• No upper bounds on valuation relative to market cap
• Bull scenario could reduce WACC to unrealistic levels

Changes:

• Enforce minimum 4% spread between WACC and terminal growth
• Cap terminal value at 50x market cap
• Cap final valuation at 100x market cap
• Set 7% WACC floor in scenario analysis
• Cap base revenue growth at 30%
• Fix transition stage to use proper cumulative growth calculation

Test plan

• Added 17 new test cases covering all DCF valuation functions
• Includes specific test for OKTA-like scenario ensuring no trillion-dollar valuations
• All 60 tests pass (43 existing + 17 new)

[biplavbarua]

LGTM! 📉

The safeguards here are spot on for a production DCF model.

1. Spread Enforcement: wacc - terminal_growth >= 0.04 is the critical fix. Without this, close values (e.g., 9% WACC vs 8.9% growth) cause the terminal value denominator to approach zero, resulting in the reported $16T valuations.
2. Ceilings: The 50x terminal / 100x total market cap guardrails are a good practical stop-gap against data outliers.

This should stabilize the valuation_agent outputs significantly.

[biplavbarua]

LGTM.
The introduction of the 50x Market Cap ceiling for Terminal Value and the 100x ceiling for the total DCF valuation provides a necessary upper bound, preventing outlier explosions in the agent's logic.
Critically, enforcing a minimum spread of 4% between WACC and terminal growth is a standard robustification technique to avoid the asymptotic singularity when g approaches WACC.
This significantly improves the stability of the valuation models.