CFA Level III common mistakes
CFA Level III · Understand the traps and how to avoid them.
When the facts state that a practice is legal under local law but conflicts with the CFA Institute Code and Standards, the candidate selects the answer that follows local law and concludes no violation occurred.
Candidates remember the phrase 'follow local law' from Standard I(A) but drop the qualifier. The exam deliberately sets local law below the Code's standard to test whether the candidate applies the 'stricter of the two' rule without hesitation.
What to do instead
Apply a mechanical rule on every ethics item: comply with the stricter of applicable law/regulation and the Code and Standards. Only if local law is the stricter rule does it govern. Say the rule out loud before reading the options so the distractor about local custom cannot pull you toward the permissive answer.
The candidate correctly senses that the conduct is wrong but names or selects the wrong Standard, e.g. labelling a member's personal trade ahead of client orders as a Fair Dealing breach rather than a Priority of Transactions breach, or treating a gift from a broker as an Independence and Objectivity issue when the question asks about Conflicts of Interest.
Level III ethics vignettes bundle several duties into one narrative, and candidates pattern-match on a single keyword ('trade', 'gift', 'client') rather than identifying which specific duty was breached and who was harmed.
What to do instead
Before looking at the options, answer three questions in writing: who is the member's client or beneficiary, what duty was owed, and which specific Standard names that duty. Note that many Level III items only ask whether a violation occurred and how to remedy it, so the label matters less than the duty analysis - do not let an uncertain label change a correct 'violation, and here is the fix' answer.
When the client's ability to bear risk and willingness to bear risk point in different directions, the candidate adopts the higher of the two risk tolerance levels - often justified by recent strong portfolio performance - and builds the allocation around it.
The questionnaire answers are vivid and recent market returns feel like evidence, so willingness is overweighted; candidates also misread the objective as 'maximise return subject to the client's stated comfort' instead of the binding constraint.
What to do instead
The lower of ability and willingness governs. Treat ability (wealth, time horizon, liquidity needs, human capital) as the hard constraint and willingness as the soft one. If willingness is below ability, the correct answer usually includes client education and counselling, not a shift to the higher risk allocation.
In an individual or institutional asset allocation problem the candidate builds the balance sheet from accounting-style line items only, omitting the present value of future earnings and the present value of planned consumption (or, for an institution, the PV of future contributions and benefit payments), or double counts a pension surplus already reflected in the financial assets.
The economic balance sheet is presented in the curriculum as a separate concept and is easy to skip when the vignette supplies a tidy net-worth table that already looks complete.
What to do instead
Always construct the economic balance sheet explicitly: extended assets = financial assets + human capital (PV of future labour income) + PV of pension/other contractual inflows; extended liabilities = financial liabilities + PV of planned consumption and bequests (or PV of benefit payments). Total economic wealth is the difference, and the allocation percentages must be computed on that base. Deduct an existing pension claim once, not twice.
When the risk-free rate does not coincide with a corner portfolio, the candidate maximises the Sharpe ratio using a single corner portfolio or the full efficient frontier, instead of combining the two adjacent corner portfolios that bracket the risk-free rate.
The corner portfolio table is long and candidates try to shortcut by picking the highest-Sharpe corner portfolio outright; the two-fund separation logic is understood conceptually but not executed under time pressure.
What to do instead
Locate the risk-free rate relative to the corner portfolio expected returns. If it lies between two corner portfolios, the optimal risky portfolio is a weighted combination of exactly those two; solve for the weights that equate the Sharpe ratio (or that hit the required return), then blend with the risk-free asset. Verify the resulting portfolio's expected return meets the required return before selecting it.
In a Brinson-style attribution question the candidate computes the allocation effect using ending weights (or uses the total benchmark return in the selection effect) so that the allocation, selection and interaction effects do not sum to the total excess return.
The beginning-weight convention is stated once in the curriculum and candidates default to 'current' weights; the interaction term is also routinely dropped without checking which attribution model the question has specified.
What to do instead
Write the model down before computing. Brinson-Fachler: allocation = (w_p - w_b)(R_b - R_b,total); selection = w_b(R_p - R_b); interaction = (w_p - w_b)(R_p - R_b). Brinson-Hood-Beebower folds interaction into selection. Use beginning-of-period weights throughout, and finish by adding the effects and checking that they reconcile to the actual minus benchmark total return - if they do not tie, the convention is wrong.
On a GIPS item the candidate selects an answer stating that a firm must be verified, that CFA Institute certifies compliance, or that a single composite can claim compliance on its own.
Words like 'verify' and 'compliance' are conflated with audit and certification, which are familiar from other regulatory regimes; candidates also skim past the distinction between firm-level and composite-level claims.
What to do instead
Anchor three facts: compliance is claimed only at the firm level (a composite is 'presented in a GIPS-compliant firm', it does not claim compliance itself); verification is voluntary, recommended, performed by an independent third party, and covers the whole firm, not one composite; and no certification or endorsement by CFA Institute exists. Any option containing 'must be verified' or 'certified' is wrong.
When asked to change a portfolio's beta with index futures the candidate writes N = (beta_target - beta_current) x Portfolio value / (Futures price x multiplier) and omits the beta of the futures contract; in a cross hedge the candidate divides notional values directly without the correlation and relative-volatility terms.
The simplified textbook version of the formula assumes the futures contract has a beta of one, and candidates memorise that version without noticing the vignette has supplied a contract beta; the minimum-variance hedge ratio also gets reduced to a naive notional match.
What to do instead
Use N = (beta_target - beta_current)/beta_futures x V_P/(P_f x multiplier) and confirm whether the hedge requires a long or short position. For cross hedges use the minimum-variance hedge ratio h* = correlation x (sigma_spot/sigma_futures) applied to the exposure, and round to whole contracts with a sanity check that the resulting hedge is directionally correct.
The candidate compares a private equity or private real estate fund's reported standard deviation, beta, correlation with public equities and Sharpe ratio directly against a public market index and concludes the private fund is superior on a risk-adjusted basis; alternatively, the candidate compares the fund's IRR directly with a time-weighted public index return.
Reported private market returns are appraisal- or model-based and smoothed, which mechanically damps volatility and correlation, and IRR is a money-weighted, reinvestment-assuming measure, so both comparisons look favourable on the page.
What to do instead
State the bias explicitly and correct for it: unsmooth the reported return series (first-order autoregressive/Geltner de-smoothing) before computing volatility, beta or correlation, or substitute transaction-based and comparable-listed indices. Compare like with like - IRR against an IRR-based or public-market-equivalent (PME-style) benchmark - and report MOIC/DPI/TVPI alongside IRR because IRR assumes reinvestment at the IRR itself.
Asked for the return an individual's portfolio must earn, the candidate divides the annual spending need by the portfolio value, producing a pre-tax, real (or arbitrarily chosen) figure, and then selects the allocation whose expected return just clears that number; tax drag and fees are either ignored or applied to the whole return rather than to the taxable portion.
The spending/portfolio shortcut is quick and produces a plausible-looking number, and the tax and fee layers feel like refinements rather than the point of the question.
What to do instead
Build the requirement in a fixed sequence: (1) real after-tax spending need / investable assets gives the real after-tax return required; (2) add inflation to get the nominal after-tax requirement; (3) gross up by dividing by (1 - tax rate) for the pre-tax nominal requirement; (4) add investment management fees on top; (5) check liquidity and time-horizon constraints. Then choose the allocation whose pre-tax expected return clears the fully loaded number, and consider asset location - placing the most tax-inefficient assets in the tax-deferred account - to reduce the required gross-up.