bull-bear-thesis

bull-bear-thesis

Purpose

Force a symmetric treatment of upside and downside. The skill output is two arguments of comparable length, each one trying its hardest to make its case. Confirmation-bias antidote: a one-sided thesis is a sign the analysis is incomplete.

Inputs

• ticker (required).
• All prior skill outputs for this ticker.

Methodology

1. Build the bull case — 3 to 5 numbered pillars. Each pillar:
   • Claim (one sentence).
   • Evidence (cites specific outputs from fundamental-research, historical-baseline, earnings-analysis, news-intelligence).
   • Milestones — what the next 12 months should produce if the thesis is right.
   • Invalidation — the specific datapoint or event that would break this pillar.
2. Build the bear case — same structure, equal length. The pillars draw on risk-assessment, news-intelligence, the qa_pressure_points from earnings-analysis, and the anomalies from historical-baseline.
3. Stress-test both. For each side, list one piece of evidence the other side would weaponize. Acknowledge it honestly.
4. State your view. One paragraph, with explicit weights on bull vs bear pillars. This is the only place a directional view is allowed in the harness. It is still framed as a labeled assumption.

Dependencies

• Consumes: every other skill's output for this ticker.

Output schema

Markdown at ${CLAUDE_PLUGIN_DATA}/companies/<TICKER>/bull-bear-thesis.md:

# Bull case
1. **Pillar.** Claim. Evidence: ... (source). Milestones: .... Invalidation: ....
2. ...

# Bear case
1. **Pillar.** Claim. Evidence: ... (source). Milestones: .... Invalidation: ....
2. ...

# Stress test
- The bull's strongest point against the bear: ...
- The bear's strongest point against the bull: ...

# Working view
Assumption-weighted stance: bull X% / bear Y%. Rationale: ....

Source citation policy

Every pillar cites the specific prior-skill output it draws on, with a path under ${CLAUDE_PLUGIN_DATA}/companies/<TICKER>/. The working view explicitly labels itself assumption:.