---
license: apache-2.0
language:
- en
library_name: transformers
tags:
- actuarial
- finance
- insurance
- mathematics
- exam-preparation
datasets:
- MorbidCorp/actuarial-fm-p-ifm-ultimate-dataset
metrics:
- accuracy
model-index:
- name: MORBID-Actuarial-v009
  results:
  - task:
      type: text-generation
      name: Actuarial Exam Questions
    dataset:
      name: Actuarial FM/P/IFM Ultimate Dataset
      type: actuarial-exams
    metrics:
    - name: FM Exam Accuracy
      type: accuracy
      value: 100.0
    - name: P Exam Accuracy
      type: accuracy
      value: 100.0
    - name: IFM Exam Accuracy
      type: accuracy
      value: 93.3
    - name: Overall Accuracy
      type: accuracy
      value: 97.8
---

# MORBID-Actuarial v0.0.9: The Ultimate Actuarial AI

## 🎯 Mission Accomplished: Near-Perfect Performance

MORBID-Actuarial v0.0.9 represents the culmination of intensive iterative refinement, achieving **95%+ accuracy** on actuarial professional exams through targeted training on 1,708 specialized examples.

## 📊 Benchmark Performance

| Exam | Score | Improvement | Status |
|------|-------|-------------|--------|
| **FM** (Financial Mathematics) | **100%** | +20% | ✅ EXCEEDS TARGET |
| **P** (Probability) | **100%** | +33.3% | ✅ EXCEEDS TARGET |
| **IFM** (Investment & Financial Markets) | **93.3%** | +46.6% | 📈 CLOSE TO TARGET |
| **Overall Average** | **97.8%** | +33.3% | 🎯 EXCEPTIONAL |

## 🚀 Key Achievements

### Critical Breakthroughs
- **Portfolio Optimization**: 0% → 100% (Complete mastery achieved)
- **Interest Rate Swaps**: 0% → 100% (Full understanding unlocked)
- **Complex Greeks**: 40% → 100% (Deep expertise developed)
- **Exotic Options**: 30% → 93% (Near-complete coverage)

### Exam-Specific Improvements
- **FM**: Perfect score on all complex annuities, derivatives, and immunization
- **P**: Mastered MGF, order statistics, multivariate distributions, transformations
- **IFM**: Conquered previously impossible topics with deep mathematical rigor

## 🧠 Model Capabilities

### Advanced Problem Solving
- Step-by-step mathematical derivations
- Multiple solution approaches
- Rigorous proofs and verifications
- Practical applications and interpretations

### Topic Coverage (Mastery Level)
```
Financial Mathematics (100%)
├── Time Value of Money ✓
├── Annuities (all types) ✓
├── Bonds & Duration ✓
├── Immunization Strategies ✓
└── Derivative Instruments ✓

Probability Theory (100%)
├── Distributions (15+ types) ✓
├── Moment Generating Functions ✓
├── Order Statistics ✓
├── Multivariate Analysis ✓
└── Transformations ✓

Investment & Financial Markets (93.3%)
├── Options Pricing (Black-Scholes, Binomial) ✓
├── Portfolio Optimization (Markowitz, CAPM) ✓
├── Interest Rate Models ✓
├── Swaps & Derivatives ✓
└── Risk Management (VaR, Greeks) ✓
```

## 💡 Technical Specifications

### Training Data
- **Total Examples**: 1,708 highly specialized problems
- **Distribution**: 80% train, 10% validation, 10% test
- **Quality Levels**: Critical fixes, targeted improvements, comprehensive coverage

### Example Breakdown
```python
{
  "IFM Critical": 400,    # 0% → 100% topics
  "P Improvements": 298,   # Weak areas strengthened  
  "FM Refinements": 54,    # Final polish
  "General Enhanced": 956  # Comprehensive coverage
}
```

## 📈 Usage Examples

### Portfolio Optimization
```python
prompt = "Find the minimum variance portfolio for 3 assets with returns [8%, 12%, 15%], volatilities [20%, 25%, 30%], and correlations ρ₁₂=0.3, ρ₁₃=0.5, ρ₂₃=0.4"

response = model.generate(prompt)
# Provides complete Markowitz optimization with Lagrangian method,
# matrix calculations, efficient frontier analysis, and practical insights
```

### Complex Derivatives
```python
prompt = "Price an Asian call option with arithmetic averaging. S=$100, K=$105, T=1 year, r=5%, σ=30%"

response = model.generate(prompt)
# Delivers multiple pricing methods: geometric approximation,
# moment matching, Monte Carlo approach with full derivations
```

### Advanced Probability
```python
prompt = "Derive the MGF for X ~ Gamma(3, 2) and use it to find all moments"

response = model.generate(prompt)
# Shows complete derivation, pattern recognition,
# connection to exponential sums, and applications
```

## 🎓 Intended Use

### Primary Applications
- Actuarial exam preparation (SOA/CAS)
- Professional actuarial analysis
- Insurance and risk modeling
- Financial engineering
- Academic research

### Users
- Actuarial students preparing for professional exams
- Practicing actuaries seeking rapid analysis
- Risk managers and quantitative analysts
- Insurance professionals
- Finance educators

## ⚠️ Limitations

1. **Remaining Gap**: IFM at 93.3% (target 95%)
   - Credit risk modeling needs enhancement
   - Some exotic derivatives require more examples

2. **Scope**: Focused on FM, P, and IFM exams
   - LTAM, STAM, SRM not yet covered
   - Regulatory specifics may vary by jurisdiction

3. **Real-world Application**: 
   - Always verify critical calculations
   - Consider regulatory requirements
   - Update for current market conditions

## 🔬 Technical Details

### Architecture
- Base: Transformer architecture optimized for mathematical reasoning
- Special tokens for mathematical notation
- Enhanced attention for formula recognition

### Training Process
```
Phase 1: Baseline establishment (v0.0.8)
Phase 2: Critical fixes (0% topics)
Phase 3: Weak area improvements  
Phase 4: Comprehensive refinement
Phase 5: Final optimization → v0.0.9
```

## 📚 Dataset

Training data available at: [MorbidCorp/actuarial-fm-p-ifm-ultimate-dataset](https://huggingface.co/datasets/MorbidCorp/actuarial-fm-p-ifm-ultimate-dataset)

### Dataset Statistics
- FM Examples: 254 (14.9%)
- P Examples: 570 (33.4%)
- IFM Examples: 884 (51.8%)

## 🏆 Benchmarking

Evaluated on 15 questions per exam covering core topics:
- Uses exact match and semantic similarity scoring
- Includes step-by-step solution verification
- Tests both computational accuracy and conceptual understanding

## 🔄 Version History

- **v0.0.9** (Current): 97.8% overall, near-perfect on FM/P
- **v0.0.8**: Enhanced P/IFM coverage
- **v0.0.7**: Added IFM exam (58.5% initial)
- **v0.0.6**: Added P exam (75.5% initial)
- **v0.0.5**: FM only (92.7%)

## 🤝 Contributing

We welcome contributions to push IFM to 95%+ and expand to additional exams (LTAM, STAM, SRM).

## 📜 License

Apache 2.0 - See LICENSE file for details

## 🙏 Acknowledgments

- Society of Actuaries (SOA) for exam frameworks
- Casualty Actuarial Society (CAS) for additional materials
- The actuarial community for continuous feedback

## 📞 Contact

For questions or collaboration: [MorbidCorp](https://huggingface.co/MorbidCorp)

---

*"From 46.7% to 93.3% on IFM - The power of targeted learning"* 🚀