ASTAM Excel Question Strategy
ASTAM has one Excel workbook question and five written-answer questions. The Excel question rewards clean formulas, traceable tables, distribution-function fluency, and enough notes for graders to see what each cell is doing.
Quick Answer
The Spring 2026 syllabus says ASTAM is a three-hour, six-question, 60-point exam. One question is answered in the Excel workbook; the other five are answered in written booklets. The workbook question is graded from the workbook, not from scratch paper or the written booklet.
The notation note says candidates should know simple row and column operations, probability and quantile functions for common distributions, Goal Seek, and simple plots. Sumproduct and matrix functions may help, but the core expectation is much smaller than full spreadsheet engineering.
Excel Functions To Practice
Practice Normal, Binomial, Poisson, Chi-Squared, Negative Binomial, Gamma, F, and Lognormal probability and inverse functions. The point is not to memorize every Excel argument in isolation. The point is to know which distribution the actuarial model calls for and to enter the parameters consistently.
Negative binomial deserves special care because Excel's parameterization differs from the Loss Models beta notation. Translate before using the function, and label the translated parameter in the workbook.
Goal Seek Pattern
A common ASTAM Excel shape is: define a parameter cell, calculate fitted probabilities or fitted values from that parameter, calculate a target expression, then use Goal Seek to make the target equal zero or a specified value.
A grader-friendly workbook leaves the trail visible. Keep the parameter cell separate, label the target cell, avoid hard-coded final answers, and use formulas down the table. If you must type a numeric starting value, label it as a starting value rather than as the answer.
Original Practice Drill
Create a five-row frequency table with counts for 0, 1, 2, 3, and 4 claims. Fit a geometric model by computing fitted probabilities from one beta cell. Then add a zero modified version with a separate p0 cell and use Goal Seek to solve the beta score equation.
After the workbook calculates both maximum log-likelihoods, add a one-line likelihood-ratio-test decision cell. This drill mirrors the skill stack in released ASTAM Excel materials without reproducing any official question.
Workbook Mistakes To Avoid
Do not paste values where formulas should be. The ASTAM instructions prefer formulas because formulas help graders assign credit. Do not hide calculations on scratch sheets that are not submitted or in comments that may not be readable.
Do not overbuild. A simple table with clear labels, visible formulas, and a final interpretation cell is better than a clever workbook that only you can understand.
Original Source-Backed Practice
3 questions built from syllabus outcomes and released-exam patterns. The prompts and answers are original, so they train the skill without copying official exam text.
ASTAM Excel Workbook Drill
A compact workbook planning set for frequency fits, Goal Seek, and likelihood-ratio decisions.
- Question 1/Spreadsheet
Zero-modified frequency layout
You have counts for 0, 1, 2, 3, and 4 claims. Sketch the columns you would build to compare a geometric model with a zero-modified geometric model.
Solution And Grading Points
Use columns for k, observed count, fitted geometric probability, fitted zero-modified probability, observed times log fitted probability for each model, and a total log-likelihood cell for each model.
- Keeps observed counts separate from fitted probabilities.
- Uses one parameter cell for the geometric model and a separate zero-probability cell for the zero-modified model.
- Builds log-likelihood totals with visible formulas rather than pasted values.
- Question 2/Spreadsheet
Goal Seek target cell
In a one-parameter fit solved by Goal Seek, what should the changing cell and target cell represent?
Solution And Grading Points
The changing cell should hold the parameter being estimated. The target cell should hold the score equation, or another stated estimating equation, arranged so the desired target is zero.
- Identifies the changing cell as the parameter.
- Identifies the target as an estimating equation.
- States the target value clearly, usually zero.
- Question 3/Calculation
Likelihood-ratio decision cell
Two nested models have maximized log-likelihoods ell_0 and ell_1, with the larger model adding one parameter. Write the likelihood-ratio statistic and reference distribution.
Solution And Grading Points
The statistic is 2(ell_1 - ell_0). Under the usual regularity conditions for nested models, compare it to a chi-square distribution with 1 degree of freedom.
- Uses the larger model log-likelihood minus the smaller model log-likelihood.
- Multiplies by 2.
- Uses 1 degree of freedom because one parameter was added.