An eight-week study plan for SOA Exam P

What eight weeks can and cannot buy you

Eight weeks at ten to twelve hours a week is enough for Exam P if you already have a calculus course behind you and you are willing to be honest about what you cannot yet do from memory. It is not enough if you are learning integration by parts for the first time, and no plan fixes that; add four weeks of calculus revision at the front instead of compressing the probability.

The plan below front-loads recall. The single biggest predictor of a comfortable sitting is whether the distribution facts — mean, variance, moment generating function, the shape of the density — arrive without effort. Candidates who spend week one on hard problems and week eight trying to memorise the gamma distribution have the order backwards.

Weeks 1-2: the foundations you will use every day

Cover set theory, the probability axioms, counting, conditional probability and Bayes' theorem. These are cheap marks and they underpin everything after them, so getting them automatic pays compound interest for the rest of the plan.

Finish these two weeks able to state inclusion-exclusion, the definition of conditional probability, and Bayes' theorem without looking. Then do thirty mixed questions across the four topics in one sitting; if fewer than twenty-four come out right, repeat the fortnight rather than moving on.

Weeks 3-5: distributions, one family at a time

Take the discrete families first — binomial, geometric, negative binomial, hypergeometric, Poisson — then the continuous ones — uniform, exponential, gamma, normal, lognormal, beta. For each, learn the model it describes BEFORE the formula. A candidate who knows that the geometric distribution counts failures before the first success can rebuild its mean; a candidate who has only memorised (1−p)/p cannot tell it from the trials version under pressure.

Build a recall habit here: every study session starts with ten minutes of prompts like 'variance of a negative binomial?' answered aloud or on paper before anything else happens. Ten minutes a day for three weeks is roughly three and a half hours, and it is the highest-yield time in the whole plan.

Weeks 6-7: multivariate work and the applications

Joint distributions, marginals, conditional densities, covariance, the conditional variance formula, transformations and order statistics. This is where the exam separates candidates, and almost all of the difficulty is in setting up the region of integration rather than in the calculus.

Sketch every region. Every single one. Candidates who skip the sketch make limit errors that are invisible in their own working, which means they cannot find them when they check.

Week 8: timing, not learning

The last week is for pace. Exam P gives you three hours for thirty questions — six minutes each — and the arithmetic of that is unforgiving: two questions you refuse to abandon can cost you five you would have got right.

Do timed sets of ten questions in sixty minutes and practise skipping. The skill being trained is recognising within ninety seconds that a question is not going to fall, marking it, and moving on. Nothing new should be learned in week eight; if a topic is still weak, accept it and spend the time protecting the topics that are strong.

Where a supplement fits

This site is a supplement, not a replacement. The formula reference and the drills here are built for weeks one to five, where the work is recall and pattern recognition and where a full seminar's adaptive engine is overkill. For the last stretch — realistic full-length exams, difficulty calibration, a readiness score you can trust — keep a full study package. Pretending otherwise would not survive contact with your results.