CFA Level II common mistakes
CFA Level II · Understand the traps and how to avoid them.
Answering ethics vignettes by picking whichever option sounds most virtuous ('disclose to everyone', 'resign') instead of identifying the specific Standard and its actual requirement. Example: selecting 'report the violation to all affected clients' where the Standards only require informing the employer or dissociating from the activity.
Ethics is studied as general moral reasoning, and the distractors are deliberately written as stricter-sounding but non-required actions. Candidates never anchor the fact pattern to a rule.
What to do instead
Before reading the answer choices, name the Standard and sub-section and write the operative verb ('must', 'should', 'must not'). Memorize the 'must vs should' split, and treat GIPS and the Asset Manager Code as separate rulebooks with their own requirements.
Accepting a multiple regression as valid because R-squared is high or the F-test is significant, then quoting the coefficient t-statistics at face value even when the data show serial correlation, conditional heteroskedasticity, or multicollinearity.
R-squared and the F-statistic are the most familiar outputs and are placed most prominently; assumption diagnostics feel like secondary detail.
What to do instead
Run a fixed checklist on every regression before interpreting coefficients: Durbin-Watson near 0 or 4 (serial correlation), Breusch-Pagan or ARCH cues (heteroskedasticity), condition index or VIF (multicollinearity), RESET (functional form). State the direction of bias - inconsistent standard errors inflate or deflate t-stats - before concluding significance.
Converting an exchange rate the wrong way up: applying a USD/EUR quote to a EUR amount by dividing when multiplication is required, or reporting the direction of a currency move with the wrong sign.
Spot, forward, and implied cross rates appear in several formats in the same question, and the quote's base/price convention is never written down, so the calculation gets inverted under time pressure.
What to do instead
Before any calculation, rewrite every quote in words: 'X units of price currency per 1 unit of base currency', and label the currency of the target cash flow. Sanity-check the sign afterwards - if USD/EUR rises, the euro has strengthened and the dollar has weakened.
Defaulting to the current rate method for a foreign subsidiary without applying the functional-currency test, then placing the translation gain or loss in the wrong place (OCI versus net income).
Candidates memorize the shortcut 'current rate method goes to OCI' and skip the determination of whether the functional currency is the parent's or the local currency.
What to do instead
Write the two-question decision tree first: functional currency equals the parent's reporting currency → temporal method, remeasurement gain/loss in net income; functional currency is the local currency → current rate method, translation adjustment in OCI (CTA). Only then go to the numbers.
In an acquisition analysis, computing post-merger EPS by adding the two earnings figures and dividing by the acquirer's standalone share count, ignoring the new shares issued in a stock deal.
The acquirer's share count is the most salient number in the data, and accretion/dilution mechanics (exchange ratio, ownership split) are often skimmed in favor of the earnings arithmetic.
What to do instead
Build the deal from the share count up: new shares = target shares × exchange ratio; post-deal shares = acquirer shares + new shares; then compare post-deal EPS with the acquirer's standalone EPS to label accretion or dilution, and verify the ownership split as a cross-check.
Valuing a firm with residual income or FCFE using inputs that violate clean surplus - for example projecting book value growth that does not equal ROE × (1 − payout ratio) - so the terminal value and the stated dividend policy contradict each other.
ROE, growth, and payout are presented as separate inputs and each is plugged in independently without a consistency check.
What to do instead
Check g = ROE × b before valuing. If the given growth is inconsistent, recompute the retention ratio or ROE and explain which one the forecast implies; then use a persistence factor in the residual income terminal value that matches the accounting assumption.
Using the Z-spread in place of the option-adjusted spread when a bond is callable or putable, or computing option cost with the sign reversed, which misprices the embedded option.
Both spreads are quoted to the same benchmark curve and differ only by a term, so the two labels get treated as interchangeable.
What to do instead
Use the identity Z-spread = OAS + option cost, where option cost is positive to the investor (so Z-spread exceeds OAS for a callable bond). Before computing, state explicitly whether the bond contains an option and whether the investor is short or long that option.
In a binomial interest rate tree, discounting a node's expected payoff with that node's own rate instead of the rate that applies to the period, or discounting back only one step, and using real-world rather than risk-neutral probabilities.
Backward induction has several steps that are easy to collapse into a single discount, and 'probability' defaults to the intuitive real-world notion.
What to do instead
Write the three steps every time: (1) terminal payoffs; (2) risk-neutral probability p = (1 + r − d)/(u − d) from the risk-free rate; (3) roll back node by node, discounting each expected value at the node's rate. Count the number of discounting steps against the number of periods as a final check.
Comparing a private equity or hedge fund's since-inception IRR directly with a public equity index return and concluding the fund outperformed.
IRR is the headline number in alternatives and is money-weighted; candidates overlook capital call timing, the J-curve, and the fact that unrealized holdings are carried at stale marks.
What to do instead
Judge performance with cash-flow-aware tools: report DPI, TVPI, and a since-inception IRR alongside a public market equivalent benchmark; note fund age and the J-curve; state that IRR assumes reinvestment at the IRR itself.
Recommending the portfolio with the highest Sharpe ratio as the client's final allocation, skipping the step that mixes it with the risk-free asset according to the client's risk aversion.
Sharpe-ratio maximization is heavily drilled, so the tangency portfolio gets treated as the end of the analysis rather than the input to the asset allocation decision.
What to do instead
Present the answer in two explicit steps: identify the tangency (maximum Sharpe) portfolio, then compute the weight y = (E(Rp) − rf)/(A × variance) in the risky portfolio and state the resulting risk-free/leverage position; check the result against the client's stated constraints and behavioral biases.