300 practice problems, and you still don't know which mistake you keep making.

The problem with answer keys at this level

Exam P and FM questions are long. A single problem can run a page: identify the distribution, set up an integral or a cash-flow equation, manipulate it, evaluate, and convert to what was asked. One bad line and the final number is wrong — and the answer key gives you exactly one bit of information about which of the ten lines it was.

Worked solutions help, but they show you a correct path, not where yours diverged. When your approach differs legitimately from the manual's, comparing them tells you almost nothing.

Where Exam P candidates actually lose it

• Wrong distribution identified. Everything downstream is immaculate and irrelevant. The most expensive error and the hardest to spot alone.
• Integration limits. Joint distributions over non-rectangular regions — the algebra is fine, the region is wrong.
• Conditional vs joint vs marginal confusion, especially in Bayes-flavoured questions.
• Forgetting the Jacobian on a transformation of variables.
• Variance of a sum when the terms aren't independent and the covariance term is quietly dropped.
• Deductibles and limits in loss questions — the payment function is piecewise and the pieces get mishandled.

Where Exam FM candidates lose it

• Timing. Annuity-immediate vs annuity-due, off by exactly one period. Endemic.
• Rate conversion. Nominal vs effective, and a compounding frequency that doesn't match the payment frequency.
• Valuation date. Discounting to the wrong point and comparing quantities that aren't comparable.
• Amortisation vs sinking fund structures being conflated.
• Duration and convexity — Macaulay and modified mixed up under time pressure.
• Sign conventions on cash flows in and out, which the calculator will happily carry through to a confident wrong number.

How to review so that 300 problems actually compounds

1. Work it fully on paper before looking at anything. Partial attempts teach partial lessons.
2. Attempt it a second time before checking if you're unsure. Whether the second attempt lands tells you if this is a knowledge gap or an execution slip — and those get different study time.
3. Check the working, not the answer. Find the first line that doesn't follow. That line names the category.
4. Log categories, not problems. "Missed covariance term" three times is a study plan. "Got 14 of 20" is a mood.
5. Re-derive from the category, not from the specific question. You will not see that question again.

The habit that separates people who pass on the second sitting from people who keep sitting: they track error types and drill those, rather than grinding fresh problem sets and hoping the rate improves on its own.

What this tool does and doesn't do

It reads handwritten working — including the notation, which is most of the difficulty — and checks it line by line, naming the first step that doesn't follow and the rule involved. It works on the algebra, the calculus and the setup.

It is not a study manual and not a question bank. Use ASM, Coaching Actuaries or the SOA sample questions for those — they're calibrated to the exam and this isn't. It also won't replace a financial calculator for BA II Plus keystroke fluency, which you need independently and which is worth drilling to reflex.

Related: integration by parts and derivatives of trig functions cover two techniques Exam P leans on constantly, and checking answers without an answer key covers the self-verification habits that catch errors during the exam rather than after it.