Skip to content

Level III Core · Capital Market Expectations, Part 2: Forecasting Asset Class Returns

Capital Market Expectations Framework and Forecasting Challenges

Updated 8 October 2026 · Fact-checked

Capital market expectations (CME) are your forecasts of risk, return and correlation for asset classes, used to set the strategic asset allocation. You build them in a set sequence: specify the expectations needed and horizon, research history, choose a model, find sources, interpret conditions, document, then monitor. Then you test for data limits, biases and model error.

Understand Capital Market Expectations Framework and Challenges

Capital market expectations are a manager's views on how asset classes will perform: expected returns, volatilities and correlations. They are inputs. If they are wrong, a technically perfect asset allocation will still be wrong for the client.

The CFA framework is a process with seven steps. (1) Specify the final set of expectations needed, including the asset classes and the time horizon. The client's investment objectives and IPS inform what is needed. (2) Research the historical record. (3) Specify the method or model, with its assumptions and limitations. (4) Determine the best sources of information. (5) Interpret the current investment environment using the selected models and data. (6) Provide the set of expectations and document the results. (7) Monitor performance and use it to refine the process. Step 1 comes first because it decides which forecasts the rest of the process must produce.

The process is exposed to many errors. You will be asked to spot them from a short vignette. As a study aid, not a CFA classification, you can group them into four families: data limitations, biases and judgment errors, model and parameter uncertainty, and ex post risk and structural change.

Data limitations include: limits in economic data (it is published with lags, and later revised); data measurement errors and biases, such as transcription errors, survivorship bias and appraisal (smoothed) data that understates volatility and correlation; and limitations of historical estimates, such as time-period bias (a result depends on the sample window) and a short or unrepresentative sample.

Biases and judgment errors include: data-mining bias (searching until a pattern appears). Analyst errors include failing to account for conditioning information and misinterpreting correlations. Psychological traps add anchoring, status quo, confirming evidence, overconfidence, availability and prudence (favouring conservative, defensible forecasts).

Model and parameter uncertainty covers model uncertainty (the wrong structure) and parameter uncertainty (the right structure with noisy inputs).

Ex post risk and structural change covers two linked points. Relationships that held in the past may stop holding when the structure or regime changes. For example, correlations tend to rise in crises, so diversification can fail when it is needed. Also, ex post risk from a calm period can be a biased measure of ex ante risk because the bad outcome did not occur in that sample.

The exam rarely asks you to recite. It gives you a case and asks you to name the problem and say how to fix it.

Key rules to remember

CME process (order)
Specify expectations and horizon → Research historical record → Specify method, assumptions and limits → Determine best information sources → Interpret current environment → Provide and document expectations → Monitor and refine
Learn the order and what each step asks. Step 1 is to specify the expectations needed, including the horizon; the client's objectives and IPS inform it.
Error families
Data limitations | Biases and judgment errors | Model and parameter uncertainty | Ex post risk and structural change
This grouping is a study aid, not a CFA classification. Place every pitfall in one family before naming it, but use the curriculum term in your answer.
Smoothed data effect
Appraisal-based data → reported σ lower and correlation with other assets lower than true values
Real estate and private equity are typical cases. Diversification looks better than it is.
Ex post vs ex ante
Ex post = realised history; ex ante = forward-looking estimate
A calm sample period can produce a low ex post risk that is not a good ex ante guide.

How to solve Capital Market Expectations Framework and Challenges questions

Use this method for any vignette about building or criticising a set of capital market expectations.

  1. 1Read the command word: identify, explain, justify, recommend. Answer only what it asks.
  2. 2Find the stage of the CME process the analyst is at, and the horizon and asset classes from the IPS.
  3. 3Underline each questionable action in the vignette: the data used, the sample, the model, the way the number was derived.
  4. 4Name the specific pitfall using exact terms, such as time-period bias, survivorship bias, smoothed data, data mining, anchoring, or model uncertainty.
  5. 5State the consequence in one clause: for example, risk understated or return overstated.
  6. 6Give the fix: longer or different sample, out-of-sample test, adjust (unsmooth) data, use several models, stress correlations, or document assumptions.
  7. 7Check that you have given as many items as the question asked, in the order requested.

Quickest way: Name, effect, fix in one line

When to use it: Use for essay items worth a few points, when time is short.

  1. Write the pitfall name first, in the exam's wording.
  2. Add the effect: higher or lower return, risk or correlation.
  3. Add one fix.
  4. Stop. Extra text earns nothing, and only the requested number of responses is graded.

Common mistakes in Capital Market Expectations Framework and Challenges

  • Confusing data-mining bias with time-period bias

    Both involve a sample producing a misleading result.

    Fix: Data mining means many tests until something fits. Time-period bias means the result depends on the window chosen, and a different window would change it.

  • Treating smoothed appraisal data as accurate risk

    The numbers look stable and are published by a reputable source.

    Fix: Say the volatility and correlation are understated, and that the data should be unsmoothed before it is used in optimisation.

  • Mixing up model uncertainty and parameter uncertainty

    Both are described as forecasting error.

    Fix: Model uncertainty is the wrong framework. Parameter uncertainty is a right framework with imprecise inputs.

  • Starting the process at data collection

    Candidates assume numbers come first.

    Fix: The first step is to specify the expectations needed, including the asset classes and horizon. The client's objectives and IPS inform this step.

  • Naming a problem without a consequence or a fix

    Candidates think naming the term is enough.

    Fix: Use name, effect, fix when the command word is explain or recommend.

  • Assuming a stable correlation from history

    Past averages feel reliable.

    Fix: Remember that correlations can change with regime and rise in stress, so stress-test the portfolio.

Worked examples

Example 1

An analyst forecasts the return for a private real estate allocation using ten years of appraisal-based index returns. The index shows an annual standard deviation of 5% and a low correlation with equities. Identify the data problem and explain how it affects the allocation. Recommend one fix.

Show the solution
  1. Identify the data: appraisal-based index, so values are smoothed.
  2. Smoothing spreads price changes over time, so reported volatility is understated.
  3. It also lowers the measured correlation with equities, so diversification looks better than it is.
  4. An optimiser would therefore over-allocate to real estate.
  5. Fix: unsmooth the returns to estimate true risk and correlation, or use a transaction-based or listed-based proxy.

Answer: The problem is smoothed appraisal data. It understates volatility and correlation, overstating the diversification benefit and leading to over-allocation. Unsmooth the data or use a transaction-based series.

Example 2

A strategist tests 40 economic indicators against equity returns over 2010-2019 and finds that one has strong predictive power. She uses it to set the next five-year forecast. Identify two biases or pitfalls and give a fix for each.

Show the solution
  1. She tested many variables and kept the one that fit: data-mining bias.
  2. The result relies on a single decade that may not be typical: time-period bias.
  3. Fix for data mining: require an economic rationale and test out of sample.
  4. Fix for time-period bias: use a longer or different sample and check whether the relationship holds across regimes.

Answer: Data-mining bias (fix: economic rationale and out-of-sample testing) and time-period bias (fix: longer, varied sample, checked across regimes).

Exam tips

  • Learn the pitfall names exactly. Graders look for the standard term.
  • For every pitfall, be ready to say the direction of the error: risk understated, return overstated, or correlation too low.
  • Use the vignette's facts. A generic definition without the case detail often earns less.
  • Give only as many responses as asked. Extra ones are not evaluated.
  • Link the fix to the client's IPS horizon where the question allows.

Capital Market Expectations Framework and Challenges in other exams

The same ground in other exams, if you are preparing for more than one or want another angle on it.

Capital Market Expectations Framework and Challenges: frequently asked questions

What are the steps in developing capital market expectations?

Specify the final expectations needed, including the horizon, research the historical record, and specify the method with its assumptions and limitations. Then determine the best information sources, interpret the current environment, provide and document the expectations, and monitor performance to refine the process. Learn the order, because the exam can test it.

What are the main limitations of economic data in forecasting?

Data are published with lags and later revised. Histories may be short, may contain survivorship bias, and may be transformed or smoothed. Definitions can change, so series are not always comparable over time.

What is the difference between ex post and ex ante risk?

Ex post risk is measured from realised history. Ex ante risk is the forward-looking estimate. A calm historical period can give a low ex post figure that understates future risk.

How do I tell data-mining bias from time-period bias?

Data mining comes from running many tests and keeping the one that works. Time-period bias comes from a result that holds only in the chosen sample window. Check which of the two the vignette describes.