ActuarialForge

A .NET library for actuarial reserving, built with a focus on clarity, composability, and domain correctness.

---

Overview

ActuarialForge provides a structured and extensible framework for actuarial reserving methods, including:

• Chain Ladder
• Additive Method
• Bornhuetter-Ferguson
• Residual analysis
• Validation utilities

The library is designed around explicit domain modeling and strong typing, ensuring that actuarial logic remains transparent, reproducible, and robust.

---

Design Goals

• Domain-driven: Clear separation between model, methods, workflow, and validation
• Composable: Methods can be reused and combined
• Safe by design: Guards against invalid actuarial states (e.g. currency mismatches, zero denominators)
• Extensible: Easy to add new reserving methods or validation approaches

---

Getting Started

1. Build a Triangle from Claim Histories

var triangle = new TriangleWorkflowBuilder()
    .FromClaimHistories(claims)
    .WithTimeGranularity(ReservingTimeGranularity.Yearly)
    .UsingClaimDateBasis(ClaimDateBasis.AccidentDate)
    .Build();

---

2. Apply a Reserving Method

var chainLadder = new ChainLadder(triangle);

var projection = chainLadder.ComputeProjection();
var ultimates = chainLadder.ComputeUltimates();

---

3. Compute Loss Ratios

var lossRatios = chainLadder.ComputeUltimateLossRatios(premiums);

---

4. Use Bornhuetter-Ferguson

var bf = new BornhuetterFerguson(
    triangle,
    chainLadder,
    premiums
);

var bfUltimates = bf.ComputeUltimates();

---

Architecture

The library is structured into clearly separated modules:

ActuarialForge.Reserving.Model

Core domain objects:

• Triangle representations
• Money & currency handling
• Period abstractions

---

ActuarialForge.Reserving.Methods

Implementation of reserving methods:

• ChainLadder
• AdditiveMethod
• BornhuetterFerguson

Includes:

• Development patterns (multiplicative / additive)
• Run-off factors
• Variance estimation

---

ActuarialForge.Reserving.Workflow

Transformation layer from raw data to triangles:

• ClaimHistory
• ClaimHistoryEvent
• TriangleWorkflowBuilder

---

ActuarialForge.Reserving.Validation

Validation and diagnostics:

• Residual analysis
• Correlation structures
• Method comparison

---

ActuarialForge.Utils

Mathematical utilities:

• Linear regression
• Decimal-based numerical helpers

---

Example: End-to-End Workflow

var triangle = new TriangleWorkflowBuilder()
    .FromClaimHistories(claims)
    .WithTimeGranularity(ReservingTimeGranularity.Yearly)
    .UsingClaimDateBasis(ClaimDateBasis.AccidentDate)
    .Build();

var method = new ChainLadder(triangle);

var projection = method.ComputeProjection();
var ultimates = method.ComputeUltimates();

var residuals = new Residuals(method);
var correlation = AccidentPeriodCorrelationCalculator.Compute(residuals);

---

Key Concepts

Development Patterns

• Multiplicative (Chain Ladder)
• Incremental / additive

Run-off Factors

Encapsulated via IRunOffFactors, allowing:

• unified access
• method-specific interpretation

Strong Typing

• AccidentPeriod
• DevelopmentPeriod
• Money
• Currency

---

Validation Philosophy

The library enforces correctness early:

• Currency mismatches throw immediately
• Invalid triangles are rejected
• Division-by-zero situations are guarded explicitly

---

Extending the Library

You can easily add:

• New reserving methods via IPatternBasedReservingMethod
• Custom validation logic via IReservingMethodSelector
• Alternative workflows or data sources

---

Contributing

Contributions are welcome.
Please ensure:

• Domain consistency is preserved
• XML documentation is included
• Edge cases are handled explicitly

---

License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Author

Jan Braunschmidt