Feature add quant plugin enhancements

[edetterman]

No description provided.

[wshobson]

Thanks for the comprehensive PR, @edetterman! You clearly have deep domain expertise in quantitative trading - the analysis of overfitting, survivorship bias, and multiple testing issues is excellent. However, this PR needs significant revisions before it can be merged.

Critical Issues

1. Missing Marketplace Integration (Blocking)

The PR doesn't update .claude-plugin/marketplace.json, which means none of your agents, skills, or commands will actually be registered and available. This must be fixed.

Required:

// Update the quantitative-trading entry in marketplace.json:
{
  "name": "quantitative-trading",
  "version": "1.3.0",
  "commands": ["./commands/paper-validation.md"],
  "agents": [
    "./agents/quant-analyst.md",
    "./agents/risk-manager.md",
    "./agents/paper-analyzer.md",
    "./agents/statistical-validator.md",
    "./agents/economic-viability-analyst.md",
    "./agents/implementation-feasibility-analyst.md",
    "./agents/complexity-assessor.md",
    "./agents/research-orchestrator.md"
  ],
  "skills": [
    "./skills/sharpe-ratio-validation",
    "./skills/transaction-cost-modeling",
    "./skills/capacity-estimation",
    "./skills/implementation-planning",
    "./skills/infrastructure-sizing",
    "./skills/alternative-data-processing",
    "./skills/embeddings-for-finance",
    "./skills/market-microstructure-analysis"
  ]
}

2. Planning Document Doesn't Belong in Codebase

The 2,082-line quantitative-trading-expansion-plan.md is essentially a research paper/design doc. No other plugin in this repo has similar documents checked in - they're work-in-progress materials that belong in personal notes or a separate design doc repo.

Required: Remove this file entirely (or condense to <200 line executive summary).

3. Agents Too Thin for Repository Standards

Your agents average 30-40 lines. Repository standards are 100-200+ lines with comprehensive guidance.

Comparison:

• Your statistical-validator.md: 38 lines
• Repo standard (julia-pro from PR feat: Add Julia development plugin #110): 187 lines
• Repo standard (python-pro): 150+ lines

What's needed in agents:

• Comprehensive capabilities lists (not just bullets)
• Specific methodologies and formulas
• Example interactions showing realistic usage
• Best practices and anti-patterns
• Technical depth appropriate to the domain

Example expansion for statistical-validator.md:

Instead of:

## Focus Areas
- Sharpe ratio significance and minimum track record length
- Sample size adequacy assessment

Provide:

## Statistical Validation Methodologies

### Sharpe Ratio Significance Testing
- **Minimum Track Record Length (Bailey-López de Prado 2014):**
  - Formula: minTRL = (1 + (1-γ)SR/γ)² × (Z_α)² / SR²
  - Where SR = Sharpe ratio, γ = skewness, Z_α = Z-score for confidence level
  - Interpretation: Years of data needed for statistical significance
  
- **Probabilistic Sharpe Ratio (PSR):**
  - Formula: PSR = Φ[(SR - SR_ref) / sqrt((1-γSR+((κ-1)/4)SR²)/(T-1))]
  - Target: PSR ≥ 0.95 for high confidence
  - Accounts for higher moments (skewness, kurtosis)

### Multiple Testing Corrections
- **Bonferroni correction:** p_adj = p × n_tests
  - Example: Testing 100 variations? Threshold becomes 0.05/100 = 0.0005
- **Benjamini-Hochberg (FDR control):**
  - Ranks p-values, adjusts threshold: p_i ≤ (i/m) × α
  - Less conservative than Bonferroni for exploratory research

[Continue with more methodologies, code examples, decision trees, etc.]

4. Skills Need Implementation Guidance

Skills are currently educational/explanatory. They need actionable workflows, code patterns, and step-by-step procedures.

Example: embeddings-for-finance/skill.md should include:

• Python code snippets for Word2Vec on 13F data
• PCA implementation with sklearn
• Clustering algorithms with parameter recommendations
• Common pitfalls and debugging approaches

Required Actions

Must-fix (blocking):

1. ✅ Update .claude-plugin/marketplace.json with all new components
2. ✅ Remove the 2,082-line planning document
3. ✅ Expand at least 3 core agents to 100-150+ lines:
   • paper-analyzer.md
   • statistical-validator.md
   • research-orchestrator.md
4. ✅ Add technical depth: formulas, methodologies, code examples, decision trees
5. ✅ Test the /paper-validation workflow end-to-end

Recommended:

• Enhance skills with code examples and implementation workflows
• Add example interactions to each agent
• Include references to key papers (Bailey-López de Prado, Harvey-Liu-Zhu, etc.)

Path Forward

I recommend focusing on quality over quantity. Start with 3-4 agents done to high standards rather than 6 agents at MVP level. The conceptual foundation is strong - now it needs implementation depth to match repository standards.

Looking forward to seeing the revised version! Happy to discuss any questions.

[wshobson]

Comprehensive review complete. This PR has excellent conceptual foundation and domain expertise, but requires significant revisions to meet repository standards.