IAI Actuarial Core Principles · Paper CS2

CS2 Risk Modelling and Survival Analysis for IAI Actuarial

CS2 Risk Modelling and Survival Analysis is the IAI Core Principles subject covering risk distributions, time series, stochastic processes, survival models and machine learning. You solve it by learning each method, practising written derivations and calculations, and doing Paper B in R. Paper A is written; Paper B is computer-based.

CS2 tests whether you can build and use models for risk and for survival. The 2026 syllabus has five topic areas: Risk modelling distributions 20%, Time series 20%, Stochastic processes 25%, Survival models 25% and Machine learning 10%. Your 16 chapters map onto these areas. Loss distributions, compound distributions, copulas and extreme value theory cover risk modelling. Time series has two chapters. Stochastic processes, Markov chains and Markov processes form the third block. Survival models, estimation, transition intensities, graduation and mortality projection form the fourth. Machine learning is the last chapter.

The subject has two examinations. Paper A is written and lasts 3 hours 15 minutes. Recent Paper A question papers carry 100 marks and open with 15 multiple-choice questions of 2 marks each (30 marks), then written questions. IAI does not fix the number of MCQs, so it can change between sessions. Paper B is a 1 hour 45 minute computer-based exam, where you work in R. Paper A and Paper B are weighted 70:30.

From the November 2025 session, the pass rule is at least 30% in each of Paper A and Paper B and 50% in aggregate. IAI may amend this session by session and confirms it with results. So a weak Paper B cannot be ignored, and a strong Paper B cannot rescue a weak Paper A. Students usually lose marks not on ideas but on method: unstated assumptions, missing steps, wrong notation, and R code they never practised. Treat CS2 as a subject you do, not one you read.

Risk Modelling and Survival Analysis: chapters and topics

Random variables and distributions for risk modelling

Loss distributions, with and without risk sharing

Random variables and distributions for risk modelling

Compound distributions and their applications in risk modelling

Random variables and distributions for risk modelling

Introduction to copulas

Random variables and distributions for risk modelling

Introduction to extreme value theory

Time series

Core concepts of time series models

Time series

Applications of time series models

Stochastic processes

Markov chains

Stochastic processes

Markov processes

Survival models

Concepts of survival models

Survival models

Estimation procedures for lifetime distributions

Survival models

Maximum likelihood estimators for transition intensities

Survival models

Transition intensities dependent on age (exact or census)

Survival models

Graduation and graduation tests

Survival models

Mortality projection

Machine learning

Elementary principles of machine learning

How to prepare Risk Modelling and Survival Analysis

Plan for steady, repeated practice on both papers. Many of you study alongside work, so use short daily sessions for recall and longer weekend sessions for full questions and R.

Map the syllabus first. List all 16 chapters against the five topic areas and their weightings. Give the heavier areas, stochastic processes and survival models at 25% each, more hours, but do not skip the 10% machine learning area.
Study chapter by chapter in order within each block. Learn risk distributions before copulas and extreme value theory. Learn time series concepts before applications. Learn Markov chains before Markov processes. Learn survival concepts before estimation and graduation.
Build a formula and definition sheet in standard actuarial notation. For each chapter, write the key definitions, the conditions under which each result holds, and one worked example. Revise it daily on your phone.
Practise derivations and calculations by hand. Cover the solution, attempt the question, then compare. Write full working with assumptions stated, as the written paper rewards method.
Start R early, not in the last month. For each chapter, write the code that fits a distribution, simulates a process, fits a time series, or estimates a survival curve. Run it, read the output, and write one sentence interpreting it.
Practise MCQs separately. Do timed sets to build speed on definitions, properties and quick calculations. Note why each wrong option is wrong.
Attempt past papers under exam conditions. Do Paper A in 3 hours 15 minutes and Paper B in 1 hour 45 minutes. Mark yourself strictly and log every lost mark by cause.
In the last two weeks, revise from your error log and formula sheet. Redo the questions you got wrong, and rehearse R tasks until the code comes without looking it up.

Time management in the exam

Paper A carries 100 marks in 3 hours 15 minutes, so you have roughly 1.9 minutes per mark. Use this as a rough guide to judge how long any question deserves.
Do the MCQs first and fast. They are worth 2 marks each, so do not spend long on one. Mark doubtful ones, move on, and return at the end.
Read the whole written paper for two or three minutes before starting. Begin with the questions you are most sure of to secure marks and confidence.
If you are stuck, write down the setup, the notation and the formula you would use, then move on. Partial marks come from method.
In Paper B, which is 1 hour 45 minutes, read all tasks first. Save your code often, comment briefly, and do not lose time on one error. Skip and return.
Keep the last ten minutes for checking units, signs, probabilities between 0 and 1, and that every answer is stated clearly.

Mistakes that cost marks in Risk Modelling and Survival Analysis

Neglecting Paper B until late
Fix: Practise R weekly from the start. Remember you need at least 30% in Paper B under the current rule, as well as 50% in aggregate.
Memorising formulas without conditions
Fix: Next to each formula, write its assumptions, such as independence, stationarity or the Markov property, and check them in every answer.
Skipping stated assumptions and working in written answers
Fix: Write the model, notation, steps and result. Marks are given for method even when the final value is wrong.
Ignoring machine learning because of its low weighting
Fix: Learn the core ideas and terms well enough to pick up easy MCQ and short-answer marks. It is a short chapter.
Reading R output without interpreting it
Fix: After each run, state what the result means for the model, such as fit, significance or the shape of the curve.
Doing past papers untimed or without review
Fix: Time every attempt, mark it strictly, and keep an error log. Review the log before the next paper.

Risk Modelling and Survival Analysis: frequently asked questions

What is the exam format for CS2?

CS2 has Paper A, a written exam of 3 hours 15 minutes, and Paper B, a computer-based exam of 1 hour 45 minutes. Recent Paper A papers carry 100 marks and open with 15 MCQs of 2 marks each. The MCQ count is not fixed and has varied between sessions.

What is the pass mark for CS2?

From the November 2025 session, you need at least 30% in each of Paper A and Paper B and 50% in aggregate. The papers are weighted 70:30. IAI may amend this by session and confirms it with results, so check the latest notice.

Which CS2 topics should I give the most time?

Stochastic processes and survival models each carry 25% in the 2026 syllabus, so they deserve the most time. Risk modelling distributions and time series carry 20% each. Machine learning carries 10%, but it should still be covered.

Do I need R for CS2?

Yes. Paper B is computer-based and uses R. Practise it alongside each chapter, and learn to interpret the output, not only run the code.

When can I sit CS2 in 2026?

The May 2026 session runs from 19 to 29 May 2026. The November 2026 session runs from 24 October to 3 November 2026, centre-based online. Results come 50 to 70 days after the last exam day.