R14.03.01.01 · TASK CLASS

Risk & actuarial models

Price, reserve or stress-test with actuarial and risk models to verified figures.

20 project-authored proposals · taxonomy v1.1 · 0 new benchmark runs. Verifiers and partial-credit rules below are designs, not executed results.

Previous class address: R05.03.01.01 · Original proposal IDs are retained.

Typical applications: Excel · R · actuarial software
Class GUI-need label: HY · Hybrid

Task proposals · 20

Expand to inspect the complete original wording and proposed checks. Original proposal text remains English in both interfaces.

R05.03.01.01-T01 Estimate IBNR with a volume-weighted chain ladder in Excel

Original instruction

Excel workbook input/paid_triangle_auto.xlsx holds a 10x10 cumulative paid triangle (AY2016-2025, 12-month lags) and incurred-to-date by AY. On a new sheet 'CL', compute volume-weighted age-to-age factors, CDFs (no tail), ultimates, and IBNR = ultimate - incurred, with a total. Use formulas referencing the triangle.

Platforms: mac · macOS win · Windows
Applications: Excel
Interfaces: desk · Desktop GUI
GUI need · task setting: HY · Hybrid
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: click · Point & click type · Text & forms
Work product: WP07 · Simulation result
Conditions: None declared
Proposed difficulty: 2 / 5

Proposed verifier

Factors, ultimates and total IBNR within 0.01% of key; cells are formulas (grader perturbs one triangle cell and checks propagation).

Proposed partial credit

30% LDFs; 30% ultimates; 30% IBNR total; 10% formula propagation

R05.03.01.01-T02 Compute one-day 99% historical VaR for a five-asset portfolio

Original instruction

Excel workbook input/returns_2y.xlsx: 500 daily returns for 5 assets and position values on 'Book' as of 2026-09-30. On 'VaR', compute daily P&L under each historical scenario, then 1-day 99% historical VaR (Excel PERCENTILE.INC on losses) and 97.5% expected shortfall. Report both in output/answer.txt.

Platforms: mac · macOS win · Windows
Applications: Excel
Interfaces: desk · Desktop GUI
GUI need · task setting: HY · Hybrid
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: click · Point & click type · Text & forms
Work product: WP07 · Simulation result
Conditions: None declared
Proposed difficulty: 2 / 5

Proposed verifier

VaR and ES equal key (computed with the specified percentile convention) to 1 USD; P&L vector formulas present.

Proposed partial credit

30% scenario P&L; 40% VaR; 30% ES

R05.03.01.01-T03 Compute loss and combined ratios by line from an underwriting summary

Original instruction

Excel workbook input/uw_q3_2026.xlsx lists earned premium, incurred losses, LAE and underwriting expenses by line of business for Q3 2026. Add loss ratio, LAE ratio, expense ratio and combined ratio columns and a total row (premium-weighted), formatted as 0.0%.

Platforms: mac · macOS win · Windows
Applications: Excel
Interfaces: desk · Desktop GUI
GUI need · task setting: HY · Hybrid
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: click · Point & click type · Text & forms
Work product: WP02 · Records/table
Conditions: None declared
Proposed difficulty: 1 / 5

Proposed verifier

Ratios per line and total equal key to 0.1pp; total computed on aggregated amounts, not averaged ratios; formulas present.

Proposed partial credit

50% per-line ratios; 30% total row correct; 20% formatting

R05.03.01.01-T04 Fit a Mack chain ladder in R and report the 75th-percentile reserve

Original instruction

RStudio with the ChainLadder package. Load input/gl_incurred_triangle.csv (10x10, cumulative). Fit MackChainLadder with a log-linear tail, and write output/mack.csv (AY, ultimate, IBNR, Mack S.E.) plus output/answer.txt with total reserve and its 75th percentile assuming a lognormal with the Mack total S.E.

Platforms: lin · Linux mac · macOS win · Windows
Applications: R · RStudio
Interfaces: term · Terminal desk · Desktop GUI
GUI need · task setting: CL · CLI-sufficient
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: code · Code & commands
Work product: WP07 · Simulation result
Conditions: None declared
Proposed difficulty: 3 / 5

Proposed verifier

Per-AY ultimates and S.E. within 0.1% of key run of the same package version; total reserve and 75th percentile within 0.1%.

Proposed partial credit

40% ultimates/IBNR; 30% Mack S.E.; 30% percentile

R05.03.01.01-T05 Apply Bornhuetter-Ferguson with a priori loss ratios

Original instruction

Excel workbook input/wc_triangles.xlsx with a reported triangle, earned premium by AY, and a priori loss ratios in input/apriori.pdf. Compute BF ultimates and IBNR by AY using selected reported CDFs from the 'Selected' sheet, compare with chain-ladder ultimates, and flag AYs where the methods differ by more than 15%.

Platforms: mac · macOS win · Windows
Applications: Excel · Adobe Acrobat Reader
Interfaces: desk · Desktop GUI
GUI need · task setting: HY · Hybrid
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: click · Point & click type · Text & forms
Work product: WP07 · Simulation result
Conditions: None declared
Proposed difficulty: 3 / 5

Proposed verifier

BF ultimates/IBNR per AY within 0.01% of key; a priori ratios equal PDF values; flags equal key set.

Proposed partial credit

20% a priori inputs; 50% BF ultimates/IBNR; 30% comparison flags

R05.03.01.01-T06 Fit a Poisson frequency GLM with exposure offset and report relativities

Original instruction

R (or RStudio). input/motor_policies.csv has claim counts, exposure years and rating factors (age band, vehicle group, region). Fit a Poisson GLM with log(exposure) offset, base levels per input/base_levels.txt, and write output/relativities.csv (factor, level, relativity, p-value) plus the model AIC.

Platforms: lin · Linux mac · macOS win · Windows
Applications: R · RStudio
Interfaces: term · Terminal desk · Desktop GUI
GUI need · task setting: CL · CLI-sufficient
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: code · Code & commands
Work product: WP07 · Simulation result
Conditions: None declared
Proposed difficulty: 3 / 5

Proposed verifier

Relativities = exp(coef) within 0.1% of key; base levels exactly 1.000; AIC within 0.01; p-values to 3 sig. figs.

Proposed partial credit

50% relativities; 20% base levels; 15% AIC; 15% p-values

R05.03.01.01-T07 Stress a loan book under three supervisory scenarios and report CET1 impact

Original instruction

Excel workbook input/loan_book.xlsx (segments, EAD, PD, LGD) and input/scenarios.pdf (baseline, adverse, severely adverse PD multipliers and LGD add-ons by segment). Compute expected loss per scenario, after-tax CET1 impact at 25% tax, and CET1 ratio after stress given RWA and capital on 'Capital'.

Platforms: mac · macOS win · Windows
Applications: Excel
Interfaces: desk · Desktop GUI
GUI need · task setting: HY · Hybrid
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: click · Point & click type · Text & forms
Work product: WP07 · Simulation result
Conditions: None declared
Proposed difficulty: 3 / 5

Proposed verifier

EL per segment and scenario, CET1 ratios within 0.01pp of key; PD capped at 100% where multipliers exceed it (planted case).

Proposed partial credit

40% EL per scenario; 20% PD cap handling; 40% CET1 ratios

R05.03.01.01-T08 Run a seeded Monte Carlo loss model with @RISK

Original instruction

Excel with Palisade @RISK. Workbook input/opsrisk.xlsx defines 4 loss categories. Model frequency as Poisson and severity as lognormal per input/params.csv, correlate categories with the given correlation matrix (@RISK Define Correlations), run 10,000 iterations with fixed seed 42 and Latin Hypercube, and report mean annual loss and 99.9% percentile in output/answer.txt.

Platforms: win · Windows
Applications: Excel · @RISK
Interfaces: desk · Desktop GUI
GUI need · task setting: G2 · GUI-essential
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: click · Point & click type · Text & forms
Work product: WP07 · Simulation result
Conditions: licensed · Licensed pro app
Proposed difficulty: 3 / 5

Proposed verifier

Workbook contains RiskPoisson/RiskLognorm and RiskCorrmat definitions; saved simulation settings (iterations, seed, sampling) as specified; results within 1% of key run with same seed. Excel VBA and the @RISK XDK are disabled (macro security 'disable all') and the VM has no Python; workbook must be saved by Excel (docProps/app.xml checked).

Proposed partial credit

40% distributions and correlation set up; 20% simulation settings; 40% outputs within tolerance

R05.03.01.01-T09 Optimize a reinsurance retention with Excel Solver

Original instruction

Excel workbook input/xl_reinsurance.xlsx models net losses and reinsurance cost as a function of retention and limit. Use Solver (GRG Nonlinear) to minimize the cost of capital plus reinsurance premium subject to 1-in-200 net loss <= 25m and retention between 2m and 15m. Keep the Solver model saved on the sheet.

Platforms: mac · macOS win · Windows
Applications: Excel
Interfaces: desk · Desktop GUI
GUI need · task setting: G2 · GUI-essential
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: click · Point & click type · Text & forms
Work product: WP07 · Simulation result
Conditions: None declared
Proposed difficulty: 3 / 5

Proposed verifier

Retention and limit within 1% of key optimum; constraints satisfied; workbook stores Solver model names (solver_opt, solver_lhs*) with GRG engine; objective within 0.5% of key. Excel VBA (SolverOk/SolverSolve) is disabled and the agent VM has no Python/openpyxl; workbook must be saved by Excel (docProps/app.xml checked).

Proposed partial credit

30% Solver model saved correctly; 50% optimum within tolerance; 20% constraints hold

R05.03.01.01-T10 Stage 500 loans and compute IFRS 9 expected credit loss

Original instruction

Excel workbook input/ifrs9_portfolio.xlsx with 500 loans (origination and current PD, days past due, watchlist flag, EAD, LGD, maturity). Apply staging rules from input/sicr_policy.pdf (relative PD threshold, 30 DPD backstop, credit-impaired at 90 DPD), compute 12-month or lifetime ECL per stage, and summarize by stage on 'Summary'.

Platforms: mac · macOS win · Windows
Applications: Excel
Interfaces: desk · Desktop GUI
GUI need · task setting: HY · Hybrid
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: click · Point & click type · Text & forms
Work product: WP07 · Simulation result
Conditions: None declared
Proposed difficulty: 3 / 5

Proposed verifier

Stage per loan equals key (accuracy >= 99% for full credit); ECL by stage within 0.1% of key; summary formulas reference the loan table.

Proposed partial credit

40% staging accuracy; 40% ECL by stage; 20% summary

R05.03.01.01-T11 Build a 2026 generational mortality table with an improvement scale

Original instruction

Excel workbook input/base_table_2012.xlsx (base q_x by age and sex, 2012) and input/scale_mp2021.xlsx (improvement rates by age and year). Project q_x for calendar year 2026 for ages 20-100 by sex with year-specific improvement, then compute life expectancy at 65 for each sex.

Platforms: mac · macOS win · Windows
Applications: Excel
Interfaces: desk · Desktop GUI
GUI need · task setting: HY · Hybrid
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: click · Point & click type · Text & forms
Work product: WP07 · Simulation result
Conditions: None declared
Proposed difficulty: 3 / 5

Proposed verifier

Projected q_x within 1e-6 of key for all ages/sexes; e_65 within 0.01 years. Using a single long-term rate instead of year-specific rates fails the q_x check.

Proposed partial credit

60% projected q_x; 40% life expectancies

R05.03.01.01-T12 Compute EVE sensitivity to parallel rate shocks for a bank book

Original instruction

Excel workbook input/irrbb_book.xlsx (repricing gap buckets for assets, liabilities and non-maturity deposits with behavioral assumptions) and the 2026-09-30 base curve on 'Curve'. Compute economic value of equity in the base case and under +200bp and -200bp parallel shocks (floor at zero per policy), and the change as % of Tier 1.

Platforms: mac · macOS win · Windows
Applications: Excel
Interfaces: desk · Desktop GUI
GUI need · task setting: HY · Hybrid
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: click · Point & click type · Text & forms
Work product: WP07 · Simulation result
Conditions: None declared
Proposed difficulty: 3 / 5

Proposed verifier

Base and shocked EVE within 0.1% of key; floor applied in -200bp; % of Tier 1 within 0.1pp.

Proposed partial credit

30% discounting set-up; 40% shocked EVEs; 15% floor; 15% Tier 1 ratio

R05.03.01.01-T13 Value a defined-benefit plan's obligation with projected unit credit

Original instruction

Excel workbook input/pension_members.xlsx (50 active members: age, service, salary) and input/assumptions.pdf (discount 4.9%, salary growth 3%, retirement 65, accrual 1.5%, mortality table on 'Mort'). Compute the PUC defined benefit obligation and current service cost per member and in total.

Platforms: mac · macOS win · Windows
Applications: Excel · Adobe Acrobat Reader
Interfaces: desk · Desktop GUI
GUI need · task setting: HY · Hybrid
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: click · Point & click type · Text & forms
Work product: WP07 · Simulation result
Conditions: None declared
Proposed difficulty: 4 / 5

Proposed verifier

Total DBO within 0.2% and service cost within 0.2% of key; per-member DBO within 0.5% for >= 48 members; grader perturbs discount rate and rechecks.

Proposed partial credit

50% total DBO; 25% service cost; 25% per-member accuracy

R05.03.01.01-T14 Bootstrap an over-dispersed Poisson reserve distribution in Python

Original instruction

Python with numpy/pandas (chainladder-python allowed). input/paid_triangle_property.csv is a 12x12 triangle. Implement ODP bootstrap with 10,000 resamples, seed 2026, scaled Pearson residuals, process variance via gamma, and write output/odp.csv (AY mean reserve, S.E.) plus total 75th/99.5th percentiles.

Platforms: lin · Linux mac · macOS win · Windows
Applications: Python · chainladder
Interfaces: term · Terminal
GUI need · task setting: CL · CLI-sufficient
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: code · Code & commands
Work product: WP07 · Simulation result
Conditions: None declared
Proposed difficulty: 4 / 5

Proposed verifier

Mean reserves within 0.5% of key; S.E. and percentiles within 2% (seeded run of reference implementation).

Proposed partial credit

30% mean reserves; 30% S.E.; 40% percentiles

R05.03.01.01-T15 Select development factors and a tail in WTW ResQ

Original instruction

WTW ResQ (Windows), project input/resq_property.rqp with paid and incurred triangles. Select age-to-age factors as the 5-year average excluding high and low, fit an inverse power curve tail through ages 36-120, and save the selection plus an ultimate summary report as output/resq_summary.xlsx.

Platforms: win · Windows
Applications: ResQ · Excel
Interfaces: desk · Desktop GUI
GUI need · task setting: G2 · GUI-essential
GUI need · ordinary access: G3 · GUI-only
Interaction primitives: click · Point & click drag · Drag & drop vis · Visual judgment
Work product: WP07 · Simulation result
Conditions: licensed · Licensed pro app
Proposed difficulty: 4 / 5

Proposed verifier

Saved project's selected factors and tail equal key within 0.001; exported ultimates within 0.1%. ResQ's automation API and Excel macros are disabled for the agent, so selections must be made in the ResQ UI; grader reopens the saved project.

Proposed partial credit

40% factor selections; 30% tail fit; 30% exported ultimates

R05.03.01.01-T16 Project term-life liabilities for a model-point file in Moody's AXIS

Original instruction

Moody's AXIS (Windows), dataset input/axis_term.axs. Import input/model_points.csv into the Life cell 'TERM20', set the valuation date 2026-09-30 and the assumption tables in input/assumptions.xlsx (mortality, lapse, discount curve), run a deterministic projection and export the gross premium reserve and present value of cash flows to output/axis_results.xlsx.

Platforms: win · Windows
Applications: Moody's AXIS · Excel
Interfaces: desk · Desktop GUI
GUI need · task setting: G2 · GUI-essential
GUI need · ordinary access: G3 · GUI-only
Interaction primitives: click · Point & click type · Text & forms
Work product: WP07 · Simulation result
Conditions: licensed · Licensed pro app
Proposed difficulty: 4 / 5

Proposed verifier

Exported reserve and PVs within 0.1% of key run; dataset saved with correct tables and valuation date (grader reopens). No scripting/batch interface is available to the agent session.

Proposed partial credit

25% model points loaded; 30% assumption tables; 45% results within tolerance

R05.03.01.01-T17 Compute average annual loss and 1-in-200 OEP from an event loss table

Original instruction

input/elt_hurricane.csv has 25,000 events with annual rate, mean loss, std dev and exposure for a property book; input/terms.json gives a 5m deductible and 100m limit per occurrence. Simulate 50,000 years (seed 7, Poisson frequency, beta severity by moment matching), apply terms, and report AAL, 1-in-100 and 1-in-200 OEP and AEP.

Platforms: lin · Linux mac · macOS win · Windows
Applications: Python · Excel
Interfaces: term · Terminal desk · Desktop GUI
GUI need · task setting: HY · Hybrid
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: code · Code & commands
Work product: WP07 · Simulation result
Conditions: None declared
Proposed difficulty: 4 / 5

Proposed verifier

AAL within 1% of key analytic value; OEP/AEP percentiles within 3% of key seeded run; occurrence terms applied per event (not per year).

Proposed partial credit

30% AAL; 35% OEP; 35% AEP

R05.03.01.01-T18 Run a full reserving study with three methods and selected ultimates

Original instruction

Excel. input/reserving_pack/ has paid and incurred triangles, claim counts and premium for commercial auto AY2014-2025, plus last year's selections. Produce chain-ladder (paid and incurred), BF and Cape Cod estimates, select an ultimate per AY following input/selection_guidelines.pdf, and build exhibits on sheets Ex1-Ex5 including an actual-vs-expected test.

Platforms: mac · macOS win · Windows
Applications: Excel
Interfaces: desk · Desktop GUI
GUI need · task setting: HY · Hybrid
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: click · Point & click type · Text & forms nav · Pan/zoom/scroll
Work product: WP07 · Simulation result
Conditions: long · Long-horizon
Proposed difficulty: 5 / 5

Proposed verifier

Method results within 0.1% of key; selected ultimates follow guideline rules (deterministic rule check per AY); AvE values within 0.5%; exhibits present with named ranges.

Proposed partial credit

40% method results; 30% selections per rule; 15% AvE; 15% exhibits

R05.03.01.01-T19 Compute a Solvency II standard-formula SCR for a small non-life insurer

Original instruction

Excel template input/sii_sf.xlsx with premium/reserve volumes by line, market exposures (equity, interest, spread, currency) and counterparty data as of 2026-09-30. Compute non-life premium and reserve risk, market risk sub-modules, counterparty default risk, aggregate with the standard correlation matrices, add operational risk, and report BSCR and SCR.

Platforms: mac · macOS win · Windows
Applications: Excel
Interfaces: desk · Desktop GUI
GUI need · task setting: HY · Hybrid
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: click · Point & click type · Text & forms
Work product: WP07 · Simulation result
Conditions: long · Long-horizon
Proposed difficulty: 5 / 5

Proposed verifier

Sub-module charges within 0.5% of key; aggregation uses the regulation's matrices (grader checks intermediate cells); SCR within 0.5%.

Proposed partial credit

30% non-life risk; 30% market risk; 20% aggregation; 20% final SCR

R05.03.01.01-T20 Calibrate Hull-White to swaption vols with Solver and price a callable bond

Original instruction

Excel. input/hw_calibration.xlsx has the 2026-09-30 discount curve, a co-terminal swaption normal-vol strip and a trinomial-tree template. Calibrate Hull-White mean reversion and sigma by minimizing squared price errors with Solver, then price the 10NC3 callable bond in input/bond_terms.pdf and report its option-adjusted value.

Platforms: mac · macOS win · Windows
Applications: Excel
Interfaces: desk · Desktop GUI
GUI need · task setting: G2 · GUI-essential
GUI need · ordinary access: CL · CLI-sufficient
Interaction primitives: click · Point & click type · Text & forms
Work product: WP07 · Simulation result
Conditions: long · Long-horizon
Proposed difficulty: 5 / 5

Proposed verifier

Calibrated a and sigma within 5% of key optimum with objective <= key x 1.05; tree prices within 0.05 of key; callable value within 0.10 of key. Excel VBA (SolverOk/SolverSolve) is disabled and the agent VM has no Python, so calibration must run in the Solver dialog; workbook must be saved by Excel (Solver model names and docProps/app.xml checked).

Proposed partial credit

40% calibration quality; 20% Solver model saved; 40% callable bond value

Mapped research references

No benchmark reference is assigned to this class in the current mapping.