README.md

metadata

license: apache-2.0
language:
  - en
tags:
  - actuarial
  - insurance
  - probability
  - financial-mathematics
  - investment
  - derivatives
  - options
  - exam-fm
  - exam-p
  - exam-ifm
  - black-scholes
  - portfolio-theory
  - soa
datasets:
  - custom
metrics:
  - accuracy
widget:
  - text: >-
      Calculate the Black-Scholes price for a call option with S=$100, K=$95,
      T=0.25, r=5%, σ=20%
  - text: >-
      Stock has beta 1.5. Risk-free rate 3%, market return 9%. What's the
      required return under CAPM?
  - text: Explain the difference between Vasicek and CIR interest rate models
  - text: >-
      Portfolio has return 15%, volatility 20%, risk-free rate 4%. Calculate the
      Sharpe ratio.

MORBID-Actuarial v0.0.7 📈

🎯 Triple Exam Coverage: FM + P + IFM!

MORBID-Actuarial v0.0.7 expands to cover THREE major actuarial exams:

• ✅ Exam FM (Financial Mathematics)
• ✅ Exam P (Probability)
• 🆕 Exam IFM (Investment and Financial Markets)

📊 Model Statistics

Training Data

• Total Examples: 18,794 (+37 IFM examples)
• Training Set: 15,037 examples
• Validation Set: 1,877 examples
• Test Set: 1,880 examples

Performance Benchmarks

• FM Exam: 92.7% accuracy
• P Exam: 75.5% accuracy
• IFM Exam: 58.5% accuracy (new content)

🆕 IFM Coverage (NEW in v0.0.7)

Options & Derivatives

• Black-Scholes Formula: European option pricing
• Binomial Trees: American option valuation
• Put-Call Parity: Arbitrage relationships
• Option Strategies: Straddles, strangles, butterflies, collars

Option Greeks

• Delta (Δ): Price sensitivity to underlying
• Gamma (Γ): Delta sensitivity
• Theta (Θ): Time decay
• Vega (ν): Volatility sensitivity
• Rho (ρ): Interest rate sensitivity

Portfolio Theory

• Modern Portfolio Theory: Markowitz optimization
• CAPM: Capital Asset Pricing Model
• APT: Arbitrage Pricing Theory
• Efficient Frontier: Risk-return optimization
• Sharpe Ratio: Risk-adjusted returns

Interest Rate Models

• Vasicek Model: Mean-reverting rates
• Cox-Ingersoll-Ross (CIR): Non-negative rates
• Hull-White Model: Time-dependent parameters
• Duration & Convexity: Bond price sensitivity

Financial Derivatives

• Forward Contracts: Custom OTC agreements
• Futures Contracts: Standardized exchange-traded
• Interest Rate Swaps: Fixed-for-floating exchanges
• Currency Swaps: Cross-currency exchanges

Risk Management

• Value at Risk (VaR): Maximum loss estimation
• Conditional VaR (CVaR): Expected shortfall
• Stress Testing: Extreme scenario analysis
• Monte Carlo Simulation: Risk modeling

💻 Quick Start

Installation

pip install transformers torch

Example Usage

Black-Scholes Pricing

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("MorbidCorp/MORBID-Actuarial-v007")
tokenizer = AutoTokenizer.from_pretrained("MorbidCorp/MORBID-Actuarial-v007")

prompt = "Calculate Black-Scholes call price: S=$50, K=$48, T=0.25 years, r=5%, σ=25%"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_length=300)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)

Portfolio Optimization

prompt = """
Two stocks: A has E(r)=12%, σ=20%; B has E(r)=8%, σ=15%; correlation=0.3.
Find the minimum variance portfolio weights.
"""

CAPM Analysis

prompt = """
A stock has beta of 1.4. The risk-free rate is 3% and market return is 10%.
Calculate the required return using CAPM and explain the result.
"""

📈 Training Process

Data Sources

• SOA exam syllabi and materials
• Generated synthetic problems
• Financial engineering textbooks
• Options pricing literature

Model Architecture

• Base Model: LLaMA-2-7B or similar
• Fine-tuning: LoRA/QLoRA
• Context Length: 2048 tokens
• Training: 3 epochs

📊 Benchmark Results

IFM Topics Performance

Topic	Score
Forward Pricing	77.5%
Put-Call Parity	76.0%
Interest Rate Models	76.0%
Value at Risk	76.0%
Black-Scholes	70.0%
Option Greeks	70.0%
CAPM	70.0%

By Difficulty

• Easy: 71.88%
• Medium: 65.00%
• Hard: 23.33%

🎯 Roadmap

Completed

• ✅ v0.0.5: FM (Financial Mathematics)
• ✅ v0.0.6: P (Probability)
• ✅ v0.0.7: IFM (Investment & Financial Markets)

Upcoming

• 📅 v0.0.8: LTAM (Long-Term Actuarial Mathematics)
• 📅 v0.0.9: STAM (Short-Term Actuarial Mathematics)
• 📅 v0.1.0: SRM (Statistics for Risk Modeling)
• 📅 v0.2.0: Fellowship track specializations

⚠️ Important Notes

1. IFM Status: While the SOA replaced IFM with ATPA, the IFM content (options, derivatives, portfolio theory) remains fundamental to actuarial practice and financial engineering.

2. Limitations:

   • Complex multi-step calculations should be verified
   • Newer exam formats may differ
   • Not a substitute for official study materials

3. Best Use Cases:

   • Concept explanation and understanding
   • Practice problem assistance
   • Quick reference for formulas
   • Study companion

📚 Dataset

The training dataset is available at MorbidCorp/actuarial-fm-p-ifm-dataset

📖 Citation

@model{morbid-actuarial-v007,
  title={MORBID-Actuarial v0.0.7: Triple-Exam Actuarial AI},
  author={MORBID AI Team},
  year={2024},
  version={0.0.7},
  publisher={HuggingFace},
  url={https://huggingface.co/MorbidCorp/MORBID-Actuarial-v007}
}

🤝 Contributing

We welcome contributions for:

• Additional exam coverage
• Practice problems
• Performance improvements
• Bug fixes

📜 License

Apache 2.0 - See LICENSE file for details

📧 Contact

• GitHub: MorbidCorp/morbid-actuarial
• Discord: MORBID AI Community

---

Note: This model is for educational purposes. Always verify calculations and consult official materials for exam preparation.