How to Study for Finance Exams With ChatGPT (2026)

Used well, a $0 to $20 per month chatbot replaces the two most expensive parts of a $2,000 prep course: on-demand explanation and personalized drilling. Used badly, it quietly teaches you wrong numbers and inflates your confidence. This post is the practical playbook: six workflows that work, with copy-paste prompts, and the four jobs you should never give a chatbot.

I am Jeffrey Ting, FSA, CFA, the founder of FreeFellow. My bias runs the other way here: FreeFellow competes with nothing in this post, because a chatbot used as a tutor makes any question bank more valuable, including a free one.

The One Rule

A chatbot is a tutor, not an examiner. Everything that works below uses the model to explain, probe, and critique. Everything that fails asks it to be the source of truth: generating your question bank, verifying arithmetic, or judging your readiness. I catalogued why the examiner role fails in the seven tells of an unaudited AI question bank; this post is the other half.

Workflow 1: Explain It Back

The single highest-yield prompt. After studying a topic, close the book and write your own explanation from memory, then paste it:

I am studying for [exam]. Here is my from-memory explanation of [concept]: [your explanation]. Find every error, gap, or imprecision, ranked by how likely it is to cost me points. Do not restate the concept; critique my version.

This is retrieval practice plus targeted correction, which learning science consistently ranks above re-reading and re-watching. The model's advantage over a study partner is that it never lets a subtle imprecision slide to be polite, as long as you ask for ranked errors rather than encouragement.

Workflow 2: Interrogate Every Miss

When you miss a practice question, the correct answer is the least interesting part. The valuable part is which specific misconception produced your choice:

I answered this practice question wrong. Question: [stem and choices]. I chose [X] because [your actual reasoning]. The correct answer is [Y]. Name the specific misconception in my reasoning, state the general rule I violated, and give me one new scenario where the same mistake would trap me again.

The "one new scenario" clause is the important part: it tests whether the fix generalizes. FreeFellow's per-choice notes are built on the same idea (each wrong choice is annotated with the mistake that produces it), so on FreeFellow questions you can paste the note and go one level deeper.

Workflow 3: Socratic Drilling on a Named Topic

For a topic you have already learned but need to keep warm:

Drill me on [learning objective] for [exam]. Ask one question at a time, wait for my answer, then tell me if I am right and why before the next question. Escalate difficulty as I get answers right. Mostly conceptual; make me do no arithmetic longer than one step.

The constraints matter. One question at a time forces retrieval. The no-long-arithmetic clause keeps the model away from its weakest skill, so a wrong "correct answer" from the model is unlikely to slip in.

Workflow 4: Unpack the Formula

Memorized formulas fail under exam-day rephrasing; understood ones do not:

Take [formula]. Walk through what each term is doing economically, what happens to the result when each input rises, and what breaks if each term is removed. Then give me the two most common ways exam questions disguise this formula.

Check anything numerical against your formula sheet or textbook, but the structural intuition is where models are genuinely strong.

Workflow 5: Translate the Dense Passage

Accounting standards, regulatory text, and actuarial study notes are written for practitioners, not learners. Paste the passage and ask for a plain-language translation plus a worked micro-example. Then, and this is the step people skip, ask the model to quiz you on the original passage's terms, so the plain-language version becomes a bridge rather than a substitute.

Workflow 6: Critique Your Written Answers

For essay and written-answer formats (CFA Level III, actuarial written-answer exams, CPA simulations), a chatbot with the actual rubric is the closest thing to a personal grader most candidates will ever have. The rubric is the condition; without it, models grade on plausibility and flattery. This deserved its own post: Can AI Grade Your Essay Exam Answers? covers what works, what fails, and how FreeFellow ships the rubric-embedded prompt free.

The Four Jobs You Should Never Give a Chatbot

  • Your question bank. Unreviewed generated questions carry statistical tells and unverified answer keys. Practice against a fixed, audited bank so your score means something.
  • Arithmetic verification. Models produce confident wrong numbers. Verify with the worked solution or your calculator, never by asking the same model again.
  • Current-year figures. Tax thresholds, contribution limits, and updated standards blend across training years. Check the exam body or the current law.
  • Readiness judgment. "Am I ready?" needs data on how you perform against calibrated questions relative to other prepared candidates. A model has neither your history nor the calibration. Use a readiness measure built on real attempt data, or a full timed mock, and let the chatbot explain the gaps it reveals.

The Stack, Assembled

The 2026 setup that costs almost nothing: a fixed audited question bank for measurement (FreeFellow's is entirely free, no signup), the exam body's materials as the source of truth, and a chatbot running the six workflows above. That combination covers explanation, drilling, feedback, and measurement, the four things a $2,000 course actually sells.