Exam P · September 2026 syllabus
Use 16 recorded sessions: four for General Probability, eight for Univariate Random Variables, and four for Multivariate Random Variables. That 25%–50%–25% allocation fits all three published ranges without pretending to forecast one test form.
16-session allocation
The percentages allocate study time. They do not convert into a guaranteed number of questions because a form can depart from the ranges and one question can touch more than one learning outcome.
| Topic | Week 1 | Week 2 | Week 3 | Week 4 | Total |
|---|---|---|---|---|---|
| General Probability | 2 | 0 | 0 | 2 | 4 (25%) |
| Univariate Random Variables | 2 | 3 | 2 | 1 | 8 (50%) |
| Multivariate Random Variables | 0 | 1 | 2 | 1 | 4 (25%) |
Change the split only with diagnostic evidence. If conditional probability is the dominant setup failure, move a session toward it and record which session moved. Keeping the total visible prevents a familiar area from quietly taking half the calendar.
Weekly work products
A session ends with a problem record and a repair decision, not with a page-count target.
Use two General Probability and two Univariate sessions. Name the event, support, or distribution before calculating. Record every case where a keyword suggested the wrong model.
Use three Univariate sessions for support, tails, moments, and transformations. Add one Multivariate session that starts from a joint model and derives a marginal or conditional quantity.
Use two Univariate sessions for insurance payment variables and two Multivariate sessions for covariance, independence, and linear combinations. Check the requested variable before using a memorized moment formula.
Use two General, one Univariate, and one Multivariate session in mixed timed sets. Reopen only failed setup or check records and test whether the repair transfers to changed inputs.
Session record
| Field | What to record | Repair evidence |
|---|---|---|
| Recall | Could you name the definition or model before opening notes? | State the rule from memory on the next attempt. |
| Setup | Did you identify support, conditioning information, and the requested variable? | Rebuild the setup after changing one boundary or parameter. |
| Calculation | Was the algebra, summation, or integration correct after the model was fixed? | Reproduce the value without copying an intermediate line. |
| Check | Did probabilities, moments, and distributions satisfy basic bounds? | Use a complement, normalization, bound, or second route. |
“Careless” is not a useful category. “Used a density value as a probability” or “computed the loss instead of the insurer payment” names a repairable decision.
Final-week rule
During week four, rework failed setup and checking records before starting another source. A correct answer reached by an unexplained guess remains open. An incorrect first attempt followed by a clean reconstruction is useful evidence because it identifies what changed.
Use the univariate topic audit when the eight-session block still feels too broad. Use the active-reference ledger to keep official source numbers stable while you select practice.
Optional independent explanations
The free SOA sample material is enough to run this schedule. TheProbability Proof Manual is an optional indexed set of 718 independent explanations and 718 original variants. It is not an SOA publication and cannot predict a live exam form or guarantee an outcome.
Before relying on an exam date or registration deadline, check theSOA registration page.
Sources and accountability
ActuaryProof is a Fermion Co., Ltd. publication. Official exam facts on this page come from the sources above; schedules, study ledgers, and checking protocols are ActuaryProof editorial tools unless stated otherwise. Read about the publisher or report a suspected error.
Primary sources last checked: August 31, 2026.