ACT Collab

Actuarial Collaborative Technology Ltd.

Modern actuarial tools built in Python

Flexible, powerful, and intuitive — developed by actuaries, used by actuaries

Built on open-source foundations
Designed for real-world actuarial workflows
Supports rapid prototyping or production-ready models

CaPytal Platform

The CaPytal ecosystem provides a comprehensive suite of tools for modern actuarial workflows. Whether you use the core CaPytal package to enhance your own bespoke model development, or use our prebuilt solutions to accelerate your projects, we have products that should be of interest to any company using Python for their capital or risk modelling.

🐍 CaPytal Package

A Python package that provides a comprehensive and elegant syntax for expressing stochastic capital modelling logic, with integrated functionality for controlling simulations and correlations.

Learn More →

📊 CaPytal Modeller

Turn a simple function into an interactive model with configurable inputs and results dashboards, making models accessible to non-technical stakeholders.

Learn More →

🛡️ CaPytal Reinsurance Components

Comprehensive modelling of complex reinsurance structures, from simple XoL layers to multi-layered inuring programs, with rich interactive dashboards for analysis and reporting.

Learn More →

CaPytal Package

The foundation of our technology stack, providing intuitive syntax for complex actuarial calculations.

Loss Simulation

📊

Loss Simulation

Easily simulate aggregate or frequency severity loss distributions for insurance risk modelling

Intuitive syntax for complex distributions
Built-in parameter targeting
Controlled random seeding for recreatable and stable results

Simple Implementation

Generate loss distributions with just a few lines of code

# Simulate large frequency large_loss_frequency = simulate_values( "Poisson", parameters=mean_frequency, name="Large Loss Frequency", ) # Simulate individual large severity based on the simulated frequency large_losses = simulate_values( "Gamma", parameters={"target mean": 2000, "target std": 1000}, adjustment_parameters={"shift": 1000}, frequency=large_loss_frequency, name="Large Losses", )

Advantages of using CaPytal

CaPytal brings all your simulation functionality inside one convenient package

  • Generates a CaPytal Variable ready to use in downstream analysis
  • Simulate values for one or more elements (class, class/year etc) in a single function call
  • Inputs match up by dimension and index and produces a properly indexed result Variable
  • Named Variables allow control of simulation error and audit
  • Automatic sense checking of simulated outputs
  • Specify parameters directly or in terms of moments or percentiles
  • Apply adjustments to distributions like shifts, truncations and caps

Aggregating Data

Σ

Aggregating Data

Combine multi-dimensional arrays of data with ease

Automatically match on dimensions and indices
Simple inline operations
Works across aggregate, frequency/severity, deterministic and stochastic data

Intuitive Syntax

Add, subtract and multiply Variables with automatic alignment of indices across common dimensions

# Apply inflation by class to individual large losses by class and year inflated_large_losses = large_losses * inflation # Net down gross losses by class net_losses = gross_losses - reinsurance_recoveries # Combine losses by class across risk types total_losses_by_class = total_non_cat_losses + total_cat_losses # Collapse dimenions through sum, count, mean type operations total_losses = total_losses_by_class.sum("class", retain=False)

Advantages of using CaPytal

CaPytal removes the need to think about and manipulate dimension and index order when combining Variables

  • Automatically joins across common dimensions
  • Automically alligns elements
  • Implicit outer join for + allows intuitive use as "union or add"
  • Many in-line operations supported: power, modulus, divide, and, or etc.
  • Generic apply_operation function to combine Variables with specified join types and any operation, including "exact" join
  • Makes code easy to read, write and maintain

Correlation Modelling

🔗

Correlation Modelling

Apply sophisticated dependency structures between risk classes and risk areas using copulas with managed dependency groups

Multiple copula types
Hierarchical structures supported
Tracks dependency groups internally

Apply copula based correlation structures easily

Simple functions for appling correlations between elements within a Variable or between corresponding elements from different Variables

# Apply correlation between attritional and large loss frequency by class apply_pairwise_correlation( variable1=large_loss_frequency, variable2=attritional_losses, copula_type="Gumbel", parameters=2, ) # Apply copula to total losses between classes apply_element_correlation( variable=aggregate_losses, copula_type="t", parameters=0.5, dof=7, )

Advantages of using CaPytal for Correlations Modelling

CaPytal allows flexible copula based correlation modelling modelling using two simple functions

  • Allows a range of copula options, like Gaussan, Student's T, and Gumbel
  • Simple functions for imposing correlations between elements within a Variable or between Variables
  • Correlation inputs match up by dimension and index against Variables being acted on
  • Options to coerce input correlations into positive semi definite form or not
  • Automatic validation for convergene to target correlations
  • Checks for existing dependencies between Variables being correlated to avoid over specification

CaPytal Modeller

Easily turn code into an interactive model with inputs and results dashboards, allowing for easy exploration and communication of results.

CaPytal Modeller

📊

CaPytal Modeller

Turn static code into an interactive model

Works with any existing CaPytal function
Easy drop in wrapper to convert into an interactive model
List inputs and outputs of interest to generate an interactive dashboard

No Code Dynamic Dashboard Creation

Decorator provides additional functionality with no need to change the existing code

# Import the CaPytal Modeller model Decorator from capytal.modeller import model # Decorate any existing function as a model and list out inputs and outputs @model( inputs=["Distribution Choice", "Distribution Parameters"], outputs = ["Total Simulated Losses"] ) def my_existing_function(): # Existing function logic here

Advantages of using CaPytal Modeller

Drop in component to turn any CaPytal function into an interactive model

  • Make models interactive with minimal effort
  • Present results to models in a rich web based format making them more engaging and accessible
  • Include searchable and filterable tables of statistics and graphs
  • Update inputs and rerun models live within the dashboard to drive engagement and business use
  • Configure the layout and content of the dashboards with simple dictionary format config specification

CaPytal Reinsurance Components

Specialized tools for modeling and analyzing complex reinsurance structures.

Reinsurance Modelling

🛡️

CaPytal Reinsurance Components

Model complex reinsurance structures with XoL, QS, inuring layers, deemed layers, complex programs simply

Interactive dashboards for understanding and editing structure
Wide range on prebuilt functionality
Easily extendable to custom program and contract types

Advantages of using CaPytal Reinsurance Components

Prebuilt components for modelling and analysing outwards RI, from simple contracts to complex structures with hundreds of programs

  • Allows fully data driven modelling from a database of programs/contracts and parameters
  • Allows for interactive setup of structures one contract at a time though notebooks with minimal code
  • Object orientated approach allows programs and contracts to be interacted with and reused within an overall structure
  • Interactive dashboad for inspecting structure and underlying programs and contracts, editing parameters and rerunning
  • Handles complex inuring relationships easily
  • Rich diagnostics and visualisations to help with understanding and validation of modelled stuctures
  • Easily extesible to custom program and contract types as subclasses of the core program and contract base classes

See What You Can Build with Python

Request a demo or consultation to learn more about how our products could enhance your risk or capital modelling.

Request a Demo