What does CAS Exam MAS-II cover?

MAS-II (Modern Actuarial Statistics II) covers four topic areas: credibility theory (classical limited-fluctuation, Buhlmann, Buhlmann-Straub, Bayesian credibility), linear mixed models (random effects, ICC, BLUP, REML, hierarchical structures), statistical learning (KNN, decision trees and ensembles, PCA, K-means and hierarchical clustering, neural networks, predictive-accuracy measures including AUROC, Gini, lift, and double lift charts), and time series with constant variance (AR, MA, ARIMA, ACF and PACF diagnostics, stationarity, seasonality, forecasting). The CAS-published 2026 Content Outline allocates approximately 15-25% to credibility, 10-20% to LMM, 40-50% to statistical learning, and 15-25% to time series.

What is the MAS-II pass rate?

MAS-II historical pass rates from CAS-published statistics typically run in the high 40s to mid 50s, similar to MAS-I and other CAS preliminary exams. The CAS publishes pass rate by sitting on casact.org. Most candidates take 200 to 300 study hours to pass; candidates with prior statistics or machine-learning coursework often succeed at the lower end of that range.

How does MAS-II differ from MAS-I?

MAS-I focuses on probability models, statistical inference, and generalized linear models (GLMs) for P&C ratemaking. MAS-II picks up where MAS-I leaves off and covers more advanced statistical-learning methods (tree ensembles, PCA, clustering, neural networks) plus credibility theory and time series. Where MAS-I is 50% GLMs, MAS-II is roughly 45% statistical learning. The two exams together cover the bulk of modern statistical methods used in P&C actuarial practice. Most candidates take MAS-I first because credibility theory in MAS-II builds on the Bayesian / posterior thinking introduced in MAS-I.

What is the difference between MAS-II and SOA Exam SRM?

Both MAS-II and SOA Exam SRM cover statistical-learning foundations, but the calibration and coverage differ. MAS-II is calibrated to P&C insurance (ratemaking, classification, loss modeling) and includes credibility theory and time series, both absent from SRM. SRM is calibrated to broader actuarial applications and covers more decision-tree / random-forest material than MAS-II. A candidate who has passed SRM has a head start on the K-means, PCA, and tree sections of MAS-II but still needs targeted study on credibility theory (the largest novel section by topic-area) and on the LMM and time-series sections that SRM does not cover.

What free MAS-II resources does FreeFellow offer?

616 original practice questions across the four MAS-II topic areas, with detailed step-by-step solutions and per-choice notes. 21 written lessons covering the syllabus with charts, decision-tree diagrams, K-means scatter plots, hierarchical clustering dendrograms, ACF / PACF plots, and worked examples. A dedicated MAS-II formula sheet (downloadable PDF). Plus 12 verbatim CAS Sample Questions and 126 questions reproduced from past CAS exam papers (Fall 2018, Spring 2019, Fall 2019), all with per-question source attribution. The full bank is free with no trial period, no question cap, and no credit card. FreeFellow is not affiliated with the Casualty Actuarial Society.

CAS Exam MAS-II Study Guide (Modern Actuarial Statistics II, 2026)

CAS Exam MAS-II (Modern Actuarial Statistics II): 2026 Study Guide

MAS-II is the fourth preliminary exam in the CAS credentialing path and the second of the two CAS-only statistics exams. Where MAS-I focuses on probability models, statistical inference, and GLMs, MAS-II picks up with credibility theory, linear mixed models, statistical learning methods (KNN, trees and ensembles, PCA, clustering, neural networks), and time series. Together MAS-I and MAS-II cover the modern statistical toolkit used in P&C actuarial practice.

FreeFellow now hosts the full MAS-II stack free: 616 original practice questions across all four topic areas, 21 written lessons with charts and worked examples, a dedicated formula sheet, 12 verbatim CAS Sample Questions, and 126 questions from past CAS exam papers (Fall 2018, Spring 2019, Fall 2019), with per-question source attribution. No trial, no signup required to browse, no credit card.

Start practicing MAS-II (616 free questions)

Note

FreeFellow is not affiliated with the Casualty Actuarial Society. CAS does not endorse, promote, review, or warrant the accuracy or quality of FreeFellow products or services. The CAS Sample Questions and past-paper items reproduced on FreeFellow are presented verbatim under the de-facto industry posture established by major P&C prep providers, with attribution to the CAS-published source.

Where MAS-II Sits in the CAS Path

MAS-II is the fourth preliminary exam, taken after MAS-I but before the upper-level written-answer exams. The full CAS path:

Exam	Administered by	Format
Exam 1/P (Probability)	Joint SOA & CAS	3 hours, 30 MCQ
Exam 2/FM (Financial Mathematics)	Joint SOA & CAS	3 hours, 35 MCQ
MAS-I (Modern Actuarial Statistics I)	CAS only	4 hours, 45 MCQ
MAS-II (Modern Actuarial Statistics II)	CAS only	4 hours, 45 MCQ
Exam 5 (Basic Techniques for Ratemaking and Estimating Claim Liabilities)	CAS only	Written-answer
Exam 6 (Regulation and Financial Reporting)	CAS only	Written-answer
Exam 7 (Estimation of Policy Liabilities, Insurance Company Valuation, and ERM)	CAS only	Written-answer
Exam 8 (Advanced Ratemaking)	CAS only	Written-answer
Exam 9 (Financial Risk and Rate of Return)	CAS only	Written-answer

MAS-II is the last all-MCQ exam in the CAS path. After passing it, candidates move into the upper-level written-answer exams where the format and study patterns shift substantially. Many candidates view MAS-II as the natural pivot point to commit to the CAS specialty: by this stage you have invested significant time in P&C-calibrated material and the upper-level exams reward that depth.

What MAS-II Tests

The CAS-published 2026 Content Outline breaks MAS-II into four sections with these approximate weights:

Section	Weight	Topics
A. Introduction to Credibility	15-25%	Classical limited-fluctuation credibility, Buhlmann credibility, Buhlmann-Straub, Bayesian credibility with conjugate priors, credibility-weighted estimation for frequency, severity, and aggregate loss
B. Linear Mixed Models	10-20%	LMM assumptions, fixed vs random effects, REML vs ML estimation, BLUP, intraclass correlation (ICC), hierarchical / nested grouping, residual and Q-Q diagnostics, variance-component identification
C. Statistical Learning	40-50%	KNN, decision trees and pruning, tree ensembles (random forest, boosting), PCA, K-means and hierarchical clustering, neural network output interpretation, predictive-accuracy measures (lift, Gini, AUROC), model comparison via double lift chart
D. Time Series with Constant Variance	15-25%	AR / MA / ARIMA model framework, ACF and PACF diagnostics, stationarity conditions, deterministic vs stochastic trends, seasonality via regression and seasonal differencing, forecast construction, prediction-interval interpretation

Statistical Learning is the dominant section at 40-50%. Credibility theory is the conceptual entry point and the section that introduces the Bayesian thinking that recurs throughout MAS-II. LMM is a smaller section by weight but conceptually dense; expect to spend disproportionate time on lmerMod-style output interpretation. Time series rounds out the exam with framework recognition and applied forecasting.

MAS-II vs SOA Exam SRM: How They Compare

For candidates choosing between the SOA and CAS paths, MAS-II and SOA Exam SRM are the natural comparison point on statistical learning. Both cover machine-learning foundations applied to actuarial problems, but the coverage differs.

Dimension	MAS-II (CAS)	SRM (SOA)
Credibility theory	Heavy (15-25%)	Not covered
LMM (random effects)	Covered (10-20%)	Not covered
KNN, trees, PCA, clustering	Heavy (~40%)	Covered (~40-50%)
Time series	Covered (15-25%)	Covered (~15-20%)
Neural networks	Output interpretation only	Output interpretation only
Calibration	P&C insurance	Broader actuarial
Path	CAS only	SOA only

A candidate who has passed SRM has a head start on KNN, PCA, clustering, and time series, but still needs targeted study on credibility theory (an entirely novel section) and LMM. A candidate who has passed MAS-II has the same observation in reverse: statistical learning and time series are familiar, but SRM-specific topics (broader calibration, deeper tree-ensemble coverage) need fresh study.

How Many Hours You Need

The CAS does not publish an official hour recommendation. Candidate and prep-provider consensus:

Strong stats + ML background (statistics or applied math degree, completed coursework in regression and machine learning): 150 to 200 hours
Solid math background, some stats exposure (engineering, econ, actuarial science undergrad): 250 to 300 hours
Limited stats / ML exposure (most career changers and finance backgrounds): 300 to 400 hours

Most candidates take 16 weeks at 15 to 25 hours per week. Statistical learning rewards iterative practice on output-interpretation questions (PCA scree plots, LMM diagnostic plots, ROC and lift curves, double lift charts), so daily consistent study beats weekend cramming. Candidates who pass on the first attempt typically maintain a 5-day-per-week study habit through the final 4 weeks.

Common Trap

Underestimating the credibility section. Credibility theory is 15-25% of the exam, but it is conceptually unlike anything in MAS-I, and the Bayesian credibility subsection (conjugate priors, posterior derivations) is the single most-missed topic in post-mortems. Candidates who skim credibility because "it's small" lose more points there than in the larger statistical-learning section.

What's Free on FreeFellow

Question bank (entirely free). 616 original practice questions across all four MAS-II topic areas, with detailed step-by-step solutions and per-choice notes. Choices use the CAS interval-bucket format on calculation items ("Less than X / At least X, but less than Y / ... / At least W") matching the real-exam shape. Browsable without an account.

Written lessons with charts and worked examples. 21 lessons covering every learning outcome, with credibility-factor curves, KNN decision boundary scatter plots, decision tree diagrams, PCA scree lollipops, K-means and hierarchical clustering visualizations, ROC and lift curves, double lift charts, and ACF / PACF correlograms. Read on the web or save the lessons offline.

Formula sheet. A dedicated MAS-II formula reference at /free/cas-masii/formula-sheet/, with the credibility factor formulas, ICC, decision-tree summary statistics, PCA loadings, K-means update step, AUROC and Gini relationships, and ARIMA forecast equations. Downloadable as a print-ready PDF.

CAS Sample Question reproduction. All 12 questions from the CAS-published MAS-II Sample Question PDF, reproduced verbatim with attribution. Includes the Systolic Blood Pressure case study referenced by 2 of the sample items.

CAS past-paper reproduction. 126 questions from three published CAS exam papers (Fall 2018, Spring 2019, Fall 2019), reproduced verbatim with FreeFellow-authored worked solutions. Includes the case-study supplemental for Spring 2019 (Systolic Blood Pressure, also reused for Fall 2018 references). The Fall 2019 Warranty Payments case study is not publicly available, and the two Fall 2019 questions referencing it ship with a placeholder vignette explaining the limitation. Spring 2018 is not in the launch (CAS has not released the SPMAS-II-18 paper).

Fellow tier (optional, $59 / quarter or $149 / year per track). Timed mock exams, spaced-repetition flashcards, topic-level performance analytics, personalized study plan. The free tier covers everything needed to pass.

A 16-Week Plan

Weeks 1-3: Credibility Theory. Lessons + 25 to 35 questions per day. Classical credibility, Buhlmann, Buhlmann-Straub, Bayesian. The Bayesian subsection is heaviest; budget extra time on conjugate-prior tables.

Weeks 4-5: Linear Mixed Models. Smaller section by weight but conceptually dense. Cover fixed vs random effects, ICC, REML, BLUP shrinkage. Practice interpreting lmerMod output until variance-component reading is automatic.

Weeks 6-12: Statistical Learning. The largest section (40-50%). Sequence: KNN + trees (6-7), ensembles (8), PCA + clustering (9-10), neural networks (11), predictive-accuracy measures including the double lift chart (12). Lift and Gini computations are most-tested.

Weeks 13-14: Time Series. ARIMA framework, ACF / PACF identification, stationarity, seasonality, forecasting. The section is computational and rewards practice on AR(p) / MA(q) recognition from diagnostic plots.

Weeks 15-16: Mixed Practice + Past Papers. Take both FreeFellow practice exams under realistic conditions. Then work the 126 CAS past-paper reproductions; target 70%+ before sitting.

Start Practicing

CAS Exam MAS-II free practice questions (616 questions, 4 topics)
CAS Exam MAS-I Study Guide for candidates taking MAS-I first
SOA vs CAS: which actuarial path is right for you for candidates still choosing
Best free actuarial exam prep resources for 2026 for the cross-actuarial overview

FreeFellow is not affiliated with the Casualty Actuarial Society. CAS does not endorse, promote, review, or warrant the accuracy or quality of FreeFellow products or services.