On this page
A useful FRM® flashcard isolates the decision or step that you missed, asks you to retrieve it without hints, and carries a short answer you can judge. For a calculation error, that often means testing the choice of formula, units, divisor or interpretation before testing arithmetic. A card that reproduces an entire worked solution is usually too broad to tell you what you actually know.
Decide what the card is for
GARP describes Part I as covering quantitative tools alongside foundations, markets and valuation; Part II applies risk tools in more specialised areas on its FRM exam information page. Its study materials page says the learning objectives identify what candidates should learn for each reading and that chapter-linked practice questions are available in GARP Learning. A flashcard is a personal retrieval prompt built from your own gap. It is not a substitute for working full problems or checking the official learning objective.
Start with the error, not the card format. After a practice attempt, identify the first point where your reasoning went wrong. Was the problem a missing definition, a choice between two methods, a units conversion, a sign, or an assumption about the data? Write that point in one sentence. If you cannot name it, rework the example before making a card. Otherwise the card may rehearse a wrong diagnosis. The best prompt lets you see whether that precise failure has been repaired.
This makes the FRM version of a one-idea card more specific than a generic definition deck. GARP’s FRM FAQ notes that formula sheets are not supplied and points candidates to learning objectives that use verbs such as calculate, compute and derive when deciding which formulas merit attention. A card may therefore ask for the condition under which a formula applies, the first setup step or the reason a tempting alternative is wrong. It should connect a quantitative rule to use, rather than store an isolated string of symbols.
Turn a broad prompt into a diagnostic one
Suppose an invented practice calculation asks for sample covariance and you divide by the number of paired observations when the requested convention requires one fewer. A weak card front says, “Explain covariance, correlation, sample variance and portfolio risk.” Its back would need several paragraphs, and a partial answer would be hard to score. A sharper front says, “For observed paired returns with estimated sample means, which divisor gives the conventional unbiased sample covariance?” Its back says, “Use the number of paired observations minus one; the means were estimated from the same sample.”
The sharper card tests a choice, not a full calculation. It also includes the condition that makes the answer true. Without “observed paired returns” and “estimated sample means,” the card could be confused with a probability-weighted scenario model. If you later discover that your real difficulty was pairing dates, make a different card: “When estimating covariance from two historical series, which observations are multiplied together?” Answer: “The deviations from each series’ mean for the same date.” These two errors require different repairs even though they occur in the same topic.
For another invented error, imagine you used a variance directly where a standard deviation was needed. A useful front asks, “A question gives covariance and the two variances. What conversion must precede calculation of correlation?” The answer is, “Take the square root of each variance to obtain standard deviations, then divide covariance by their product.” An overfull back that includes the derivation of every related risk measure makes the card harder to check. Keep the explanation sufficient to prevent the same wrong turn, then do a separate numerical problem for fluency.
Give the front enough context, but no giveaway
A card should be answerable without the page that produced it. “What is the denominator?” has no stable answer because the denominator depends on the estimator. “Sample covariance of observed pairs using estimated means: divisor?” is compact and unambiguous. Include the data setting, target quantity and any convention that matters. Exclude details that merely reveal the answer. If the front contains a miniature essay, you have probably bundled several decisions.
The back needs a grading rule. Write the answer you would accept, plus one brief reason or warning where confusion is likely. For a sign-convention card, the answer might identify both the sign and its meaning. For a units card, it might state which conversion is required and when. Avoid a back that says only “review chapter” or “be careful”; neither tells you whether retrieval succeeded. Conversely, do not make the back a copied page of curriculum text. Your own concise explanation will expose whether you understand the rule.
Test the card cold, without opening the source. If you can answer by recognising the wording but cannot explain why the method applies, revise the front to ask for the condition or a contrast. If your answer is mostly correct but misses a required qualifier, keep the qualifier explicit on the back. If the card needs three independent answers, split it. Repeated failure can also signal that the underlying concept needs a fresh worked example rather than more card repetitions.
Cards for decisions, calculations and interpretations
Some material belongs on a card as a choice between methods. For example, a prompt can ask whether a table of stated scenario probabilities calls for sample covariance with a degrees-of-freedom adjustment or a probability-weighted covariance. The answer should specify the feature that decides it: the probabilities describe the distribution in the model. This tests reading the question, which is often the real difficulty. It should not be phrased as a universal statement that all tables use one convention.
Other material is better represented as a compact setup step. A front might ask, “Before computing a return correlation from two series reported in different units, what must be checked?” The back can say, “Keep each series’ covariance and standard deviation calculations internally consistent; correlation is unitless, but mixing a percent-based covariance with decimal-based standard deviations is wrong.” A short numeric exercise can then test execution. A card alone cannot show whether you can complete a multistep calculation under time pressure.
Interpretation cards should include the limitation that matters. “What does near-zero correlation rule out?” is too vague and may invite an overclaim. Better: “If sample correlation is near zero, what conclusion about independence is justified?” Answer: “None; near-zero correlation describes little observed linear association and does not establish independence.” This card checks the specific false inference. It can be followed by a separate problem that illustrates nonlinear dependence if that is within the learning objective.
What to leave out of the deck
Do not turn an entire reading into one card or every sentence into its own card. A deck made only of definitions may feel easy while leaving the setup of quantitative questions untouched. Do not transcribe practice questions or actual exam content. Build original prompts from the concept and from your own mistake. GARP states that its chapter practice questions are not representative of actual FRM exam questions, so a flashcard based on a practice error should be treated as a learning check, not as a prediction of future exam wording.
Leave out answers that cannot be judged. “Understand market risk” is a study intention, not a prompt. “Explain everything about VaR” is too large for a single retrieval check. If a concept genuinely needs an extended derivation, use a worked exercise or a blank-page explanation, then create cards for the bottlenecks you discover. This keeps the deck manageable and gives each card a reason to exist.
Do not let a card preserve an error in the source solution you wrote yourself. Before adding it, check the relevant official learning objective and redo any arithmetic independently. For a numerical card, mark the numbers as invented and make the answer computable from the front alone. If a rounding convention matters, state it. If multiple reasonable answers fit, clarify the question rather than memorising one arbitrary wording. This quality check is especially useful for formulas, where a missing assumption can change the answer.
A small review loop
After a practice session, select the errors that reflect a reusable rule. For each, write a plain-language diagnosis, a single prompt, and an answer with a reason. On the next review, answer before turning the card over and say why the rule fits. If you fail, inspect whether the problem is recall, understanding or execution. Rewrite ambiguous cards; revisit the reading for understanding gaps; use a worked problem for execution gaps. Keep successful cards in circulation only while they are useful for retrieval.
The FRM programme page provides the wider programme context, while Omni’s study methodology explains how to organise practice. A card earns its place when it helps you identify a specific decision you can now make correctly. The worked problem remains the test of whether that decision survives a full calculation.
Sources
- FRM Program and Exams | GARP · garp.org · Retrieved
- FRM Study Materials | GARP · garp.org · Retrieved
- FRM Exam FAQs | GARP · garp.org · Retrieved



