1.1.1 - Capital Market Expectations Framework
What you'll learn
- Translate an investment objective into a precise request for market forecasts.
- Use the seven-step forecasting process to select evidence and document a view.
- Combine economic scenarios and reconcile forecasts across assets and horizons.
- Evaluate a forecast against the outcome it was designed to predict.
Estimated reading time: 17 minutes.
Why it matters
A pension committee that needs 6% and an analyst who expects 4% have raised a funding question. Raising the forecast to 6% would conceal it. In a Level III vignette, the same distinction determines whether to challenge a market assumption, revise the portfolio analysis, or bring an objective back to the committee.
Key terms
- Forecast origin
- The date as of which the forecaster forms the estimate. It fixes the information that could have influenced the decision.
- Investable proxy
- A specified index or portfolio used to represent an eligible exposure. Its holdings and return convention determine what the forecast describes.
- Conditional expectation
- The average outcome expected given particular information, such as a recession occurring during the forecast year.
- Risk premium
- The expected return increment above a specified reference return for bearing an exposure. Both legs need a common currency and horizon.
- Cross-sectional consistency
- Agreement among the economic assumptions underlying estimates for different assets over the same period.
- Intertemporal consistency
- Agreement among estimates for different periods once their horizons, units, and compounding conventions are reconciled.
- Forecast vintage
- The version of an estimate issued on a particular date, including the assumptions and data then available.
Core content
1.1.1.1 Specify what the investment decision requires
Capital market expectations describe the prospective distribution of outcomes for asset classes. Start with the decision the investment committee must make. A strategic allocation may require long-run total returns and a joint risk model for every eligible asset class. A near-term spending decision may instead turn on cash proceeds available on one date and the loss the investor could sustain before then. One institution can need both sets of estimates.
Forecast at the level at which a decision can change. A single investment-grade bond estimate may suffice if the only decision is the overall bond weight. If the committee can shift between government and corporate bonds, the single estimate hides the credit exposure being decided. Conversely, dividing a short history into many narrow categories can give unreliable differences enough apparent precision to drive large trades.
Table 1 translates four different mandate details into forecast specifications.
| Mandate detail | Estimate to obtain | Reason it matters |
|---|---|---|
| Maintain purchasing power over ten years | Ten-year compound real total return, or nominal total return with a compatible inflation assumption | Inflation changes the spending power of terminal wealth. |
| Meet a known payment in nine months | Nine-month cash flows, realizable values, and downside scenarios | An annualized long-run mean cannot establish cash availability on the payment date. |
| Evaluate foreign equities hedged to sterling | Sterling total return under the stated hedge policy | Local index performance omits the effect of implementing the hedge. |
| Set weights across several asset classes | A return vector and compatible volatility and correlation estimates | The risk contribution of an asset depends on its co-movement with other holdings. |
Attach conventions to each number. A total return includes investment income and price change. A price return omits income. An institutional gross market forecast can serve several clients, but a household objective stated after fees and taxes requires a client-specific translation before comparison. State the treatments actually assumed. Subtracting a fee from a return series that already includes it understates expected wealth.
Horizon and units are separate specifications. A ten-year forecast may be expressed as cumulative growth over ten years or as an equivalent annual rate. Neither describes the distribution of the coming year's return. Similarly, a euro investor's unhedged foreign return depends on both local asset performance and the change in the euro value of the foreign currency. A hedged return also reflects hedge terms and implementation. Simply renaming a local-currency estimate in euros leaves the exposure unresolved.
A Checkpoint
A sterling-based foundation invests for perpetuity. Its board asks whether it can pay a contractual euro grant in six months without selling private assets. Which estimate should the analyst develop first?
- Ten-year expected real sterling return on the endowment
- Euro cash available in six months under plausible liquidation and exchange-rate outcomes
- Long-run expected illiquidity premium on private assets
- Annualized sterling volatility of the strategic portfolio
B Give active views a common reference
A plan forecasts a 5% return for its broad equity benchmark. An active manager forecasts 8% for the portfolio and 7% for the same benchmark. On the manager's assumptions, the expected active return is 1 percentage point. The other 2 percentage points of the gap above the plan's forecast reflect a more optimistic market view. Separating those judgments helps the committee decide what evidence to request about security selection.
C Specify when the estimate will be refreshed
A committee can request a rolling ten-year forecast at each quarterly meeting. Ten years describes the prospective holding period. Quarterly describes how often the analyst revisits the assumptions and issues a new vintage. If the question asks how often to update the estimate, look for the decision calendar and the circumstances that would justify an earlier reassessment, such as a material change in starting yields.
1.1.1.2 Build a forecast from evidence available at the decision date
The sequence below traces a forecast from the original request to the evidence used in the next round.
A longer sample can improve precision when the observations describe a stable process. Adding years from a different monetary regime, market structure, or index definition may instead make an estimate less representative. A shorter sample provides more relevant observations at the cost of greater sampling uncertainty. A defensible response to a regime change is to compare estimates across windows and explain which evidence still applies. Automatically discarding all older data can remove the only observations from a severe downturn.
Data availability matters even when the observation period is correct. A December economic statistic first released in February could not inform a January forecast. A subsequently revised December figure introduces a second discrepancy. Store the release date and vintage alongside the observation date. A model calibrated with revised data can appear better at recognizing turning points than the forecaster could have been in real time.
Table 2 shows how choosing a forecasting method determines which source to seek.
| Method | Inputs to seek | Question to resolve |
|---|---|---|
| Statistical estimation | Comparable observations of returns or predictors | Will the historical relationship plausibly persist over the forecast horizon? |
| Discounted cash flow | Current price and a defensible path for future cash flows | What return reconciles the assumed cash flows with today's price? |
| Risk-premium buildup | A reference return and increments for distinct priced exposures | Does the reference return already include a premium being added? |
| Equilibrium inference | Market weights, a covariance model, and assumptions about risk compensation | What expected returns would support the observed market allocation under the model? |
Models can disagree for identifiable reasons. A backward-looking equity mean can include a large historical valuation expansion. A cash-flow estimate using today's price may imply a lower return if repeating that expansion would require an implausible terminal valuation. Investigate the assumption driving the difference before combining the estimates. A survey of analysts or a panel's judgment can challenge a model, but apparent agreement adds little independent evidence when all respondents use the same starting forecast.
Match each piece of evidence to the inference it supports.
- Historical sample mean:
- Average of observed periodic returns in the selected window.
- Discounted cash flow estimate:
- Rate that equates projected receipts with today's price.
- Risk-premium buildup:
- Reference yield plus compensation for additional priced exposures.
- Equilibrium inference:
- Returns implied by market holdings and a model of risk compensation.
A Decide what additional observations can tell you
Five years of daily returns and five years of monthly returns cover the same economic history. Under independent, identically distributed additive returns, the standard error of the estimated annual mean is approximately annual volatility divided by the square root of the number of years observed. At 20% annual volatility, five years imply a standard error of about 8.9 percentage points under those assumptions. Switching to daily observations alone does not turn this into a precise expected-return estimate. Higher frequency can nevertheless help estimate volatility when prices are reliable.
Table 3 distinguishes three questions a committee might ask about a forecast.
| Committee question | Source of uncertainty | Relevant check |
|---|---|---|
| How widely could next year's realized return vary around its mean? | Dispersion of investment outcomes | Examine forecast volatility, downside scenarios, or a prediction interval. |
| Would the conclusion survive a different economic relationship? | Model uncertainty | Compare economically plausible models on the same forecast basis. |
| Would the conclusion survive a different estimate of a model input? | Input uncertainty | Vary the uncertain input while holding the model and other inputs fixed. |
A longer representative history can narrow a confidence interval for the mean even while individual annual returns remain volatile. A committee estimating the chance of a one-year funding shortfall needs the distribution of future outcomes. Using the standard error of the mean as annual portfolio volatility would make the shortfall analysis look increasingly safe merely because more historical data became available.
Suppose a dividend model implies a 6% equity return. Questioning whether dividends adequately represent distributions in a market dominated by buybacks challenges the model. Accepting its payout measure but asking how a lower sustainable growth input changes the result challenges an input. A sensitivity calculation answers the second question. To address the first, compare the estimate with one obtained from a model that captures total shareholder distributions.
1.1.1.3 Distinguish a scenario outcome from an overall expectation
A forecast can be conditional on today's information and still encompass several possible future states. For example, an analyst can use today's yield curve to assign a probability to contraction. The bond return projected if contraction occurs answers a narrower question than the return expected today across all possible states. A stress test can also examine a severe outcome without assigning it any probability. Such a test reveals vulnerability but supplies no probability weight for calculating a mean.
For mutually exclusive states covering all possibilities, \(E[R]=\sum_{s=1}^{S}p_s E[R\mid s]\). Here \(S\) is the number of states, \(p_s\) is the probability of state \(s\), and \(E[R\mid s]\) is the expected holding-period return within that state. The probabilities sum to one. Before applying the formula, check whether two labels could describe the same outcome. For example, a recession and falling policy rates can occur together. They do not form two exclusive branches by themselves.
Use the branches in Fig 1 to choose the output warranted by the committee's question.
Key point: A cross-state mean requires both complete state coverage and defensible probabilities.
flowchart TD
A[Committee forecast request] --> B{Named state assumed?}
B -->|Yes| C[Report conditional outcome]
B -->|No| D{Exclusive states cover outcomes?}
D -->|No| E[Redefine scenario set]
D -->|Yes| F{State probabilities available?}
F -->|Yes| G[Compute weighted mean]
F -->|No| H[Report separate scenario outcomes]Worked case
A pension analyst estimates that for the next year a credit portfolio will return 8% in expansion and lose 4% in contraction. These are the only two states in the model. The probability of contraction is 25%. The portfolio earned 9% last year. The committee requests a central forecast and asks how sensitive it is to a less favorable outlook.
Expansion has the remaining 75% probability. The two contributions to the mean are \(0.75(8\%)=6\%\) and \(0.25(-4\%)=-1\%\), giving a one-year expected return of 5%. The forecast sits between the two state returns and closer to expansion, the more probable outcome. Last year's 9% is a realized observation and has no separate weight in this two-state model.
If contraction probability rises to 40% with state returns fixed, the mean becomes \(0.60(8\%)+0.40(-4\%)=3.2\%\). Moving 15 percentage points of probability from an 8% outcome to a negative 4% outcome reduces expected return by 1.8 percentage points. A 2% estimate would give the two states equal weights. Reporting only 8% would answer what the portfolio earns in expansion.
Fig 2 lets you vary both the chance of contraction and the loss within that state.
Key point: A worse contraction return increases the sensitivity of the mean to contraction probability.
{"x":{"label":"Contraction probability (%)","min":0,"max":100},"y":{"label":"Expected one-year return (%)","min":-12,"max":10},"curves":[{"label":"Expected credit return","expr":"8*(1-x/100)+c*x/100"}],"sliders":[{"name":"c","label":"Contraction return (%)","min":-12,"max":0,"step":1,"value":-4}],"readouts":[{"label":"Expected return at 25% contraction probability (%)","expr":"8*0.75+c*0.25","digits":2}],"try":"Predict how a deeper contraction loss affects the mean at 25% probability, then lower the contraction return from -4% to -8% to check.","say":"The line falls from the expansion return to the contraction return as contraction becomes more likely. A deeper loss steepens the line while the expansion endpoint stays fixed.","prompt":"a live graph of expected one-year return in percent against contraction probability in percent. The curve shows expected credit return. The line falls from the expansion return to the contraction return as contraction becomes more likely. A deeper loss steepens the line while the expansion endpoint stays fixed. On the page, a slider changes the contraction return in percent, and the curve redraws as it moves. It also displays expected return at 25% contraction probability in percent. Try this: Predict how a deeper contraction loss affects the mean at 25% probability, then lower the contraction return from -4% to -8% to check."}A Checkpoint
An equity team conditions its forecast on recession, while a credit team issues an average across recession and recovery. Both numbers are required for a joint recession stress test. Which adjustment directly resolves the mismatch?
- Multiply the equity estimate by the recession probability.
- Replace both estimates with their historical means.
- Use the credit estimate because it incorporates recession risk.
- Request the credit return conditional on the same recession assumptions.
B Identify what the state probabilities describe
Suppose contraction occurs in 20% of years in a representative long-run model. An analyst interprets today's credit conditions as raising the chance of contraction next year to 35%. Weighting state returns using 35% gives a central forecast conditional on today's evidence. Using the long-run state frequencies and corresponding long-run state means gives the model's unconditional mean. A request for the estimate across all future states still requires the analyst to establish which information set applies.
C Recover the risk hidden by a scenario mean
For the credit example, the overall mean is 5%. If 8% and negative 4% are the exact returns in the two states, variance comes entirely from the distance of each return from 5%. The standard deviation is \(\sqrt{0.75(0.08-0.05)^2+0.25(-0.04-0.05)^2}\approx5.20\%\). If the two figures are conditional means, returns can also vary within each state. Add the probability-weighted conditional variances to the between-state variance before taking the square root. With positive within-state dispersion, 5.20% understates the overall volatility.
1.1.1.4 Reconcile the forecasts before using them together
Cross-sectional consistency asks whether the forecasts for several assets can coexist at a given horizon. If two teams assume different inflation rates for the same domestic economy and year, a ranking of nominal returns can reflect their inconsistent baselines. Align the shared assumption and rerun the estimates. The effects need not have the same sign. An unexpected inflation shock can reduce the purchasing power of fixed bond payments while affecting companies differently according to their pricing power and costs.
An unusual ranking is a reason to investigate. A short-horizon equity forecast can fall below the cash return if the analyst expects valuation compression. A forecast of higher default losses can coexist with a higher expected credit return if starting spreads have widened by enough. Ask which component offsets the adverse effect and whether the magnitude works. Enforcing a fixed ordering would erase economically meaningful views.
Table 4 gives a forecast submission for a one-year US dollar nominal allocation model. The investment policy permits unhedged foreign equities.
| Asset class | Submitted return | Return basis | Horizon |
|---|---|---|---|
| US Treasury bills | 3.0% | US dollar nominal total return | One year |
| US corporate bonds | 2.5% | US dollar real total return | One year |
| Foreign equities | 6.0% | Local-currency nominal total return | One year |
| US equities | 6.4% | US dollar nominal compound annual return | Ten years |
Under a fixed inflation assumption, nominal return \(r_n\), real return \(r_r\), and inflation \(i\) for the same period satisfy \(1+r_n=(1+r_r)(1+i)\). All three enter as decimals. At 2% inflation, a 2.5% real return corresponds to \(1.025(1.02)-1=4.55\%\) nominal. Subtracting inflation goes in the wrong direction when starting from a positive real return and converting to nominal at positive inflation. With uncertain inflation, convert within jointly specified scenarios if exact expected purchasing power matters. Dividing two expected growth factors generally does not produce the expected value of their ratio.
Intertemporal consistency allows a weak first year and a healthy long-run outlook. It requires a credible bridge between them. In a specified wealth path, compound growth factors across successive periods. If a five-year path compounds at 4% annually but the first year is flat, years two through five must collectively deliver the entire five-year gain. With a constant annual rate \(g\) over those four remaining years, \((1+g)^4=1.04^5\), giving \(g\approx5.03\%\). The recovery assumption is now explicit and can be assessed.
Check the risk estimates alongside the returns. For initial portfolio weights \(w_i\) and compatible one-period expected returns \(\mu_i\), the portfolio expectation is \(\mu_p=\sum_i w_i\mu_i\). Portfolio variance is \(\sigma_p^2=\sum_i\sum_j w_iw_j\operatorname{Cov}(R_i,R_j)\). The covariance of returns on assets \(i\) and \(j\) equals their correlation times their two standard deviations. Use the same horizon and return units across the matrix. For a given pair of positively weighted risky assets, a higher correlation raises portfolio variance when both volatilities remain fixed.
Every correlation lies between negative one and positive one, and a complete covariance matrix must be symmetric and positive semidefinite. The latter condition means every set of weights gives a nonnegative variance. Three assets cannot each be perfectly negatively correlated with both of the others. If two assets each move exactly opposite the first, they move together. Checking each pair against its permitted range would miss this error.
Worked case
A foundation requires a 3% one-year real return after fees. Its policy portfolio begins with 60% in equities and 40% in bonds. Compatible nominal total return forecasts are 7% and 3%, respectively. The agreed scenario assumes 2% inflation. A 0.5% fee is charged on ending assets. The foundation has a ten-year strategic review scheduled next quarter. Does the current policy mix meet the one-year hurdle in expectation under these assumptions?
Start with the weighted gross nominal estimate, \(0.60(0.07)+0.40(0.03)=0.054\). The portfolio's 5.4% lies between its component returns. A charge on ending assets leaves \(1.054(0.995)=1.04873\) per initial dollar. Divide by the inflation factor to express ending wealth in initial purchasing power: \(1.04873/1.02-1\approx2.82\%\). The estimated gap is approximately 0.18 percentage points, or 18 basis points, below the real after-fee requirement.
A committee memo can state the decision in one sentence: under the agreed assumptions the current mix falls modestly short, so assess the objective and feasible policy alternatives using the existing risk constraints. Comparing 5.4% directly with 3% would reverse the conclusion. The scheduled strategic review changes neither the time period of this question nor the reason for the shortfall.
A Convert a foreign return using the exchange-rate quote
For an unhedged holding with income reinvested locally, let \(S_0\) and \(S_1\) be starting and ending units of base currency per unit of foreign currency. For a specified path, \(1+R_{\text{base}}=(1+R_{\text{local}})S_1/S_0\). Suppose the 6% local equity return in Table 4 occurs while the foreign currency loses 5% against the dollar. Each initial dollar becomes \(1.06(0.95)=1.007\) dollars, a 0.70% gain. With the reciprocal exchange-rate quote, invert the exchange-rate ratio. For an expected return across uncertain joint outcomes, convert each path and then weight it by probability. Multiplying expected gross returns omits any covariance between the asset and currency returns.
1.1.1.5 Evaluate the forecast that was actually issued
Keep the dated estimate, benchmark identity, horizon, method, and input vintage. At the end of the horizon, obtain the realization for the same exposure. Comparing a gross local-currency forecast with a client's net base-currency outcome confounds forecast error with fees and currency movements. Comparing a strategic forecast with one quarter of returns gives a timely observation but no completed observation of the strategic horizon.
Define signed error as \(e_t=R_t-F_t\), where \(R_t\) is the realized return and \(F_t\) is the corresponding forecast. Use percentage points for the difference of two percentage returns. A positive error under this convention means the forecast was too low. Across \(n\) comparable observations, mean error is \(\sum_t e_t/n\), and mean absolute error is \(\sum_t |e_t|/n\). The first captures direction in the observed record. The second measures the average size of a miss even when signs cancel.
Table 5 preserves each forecast alongside its matched realization so both error measures can be calculated.
| Forecast year | Forecast at start of year | Realized return |
|---|---|---|
| Year 1 | 4% | 7% |
| Year 2 | 5% | 1% |
| Year 3 | 3% | 4% |
Judge the process over repeated observations and against a reasonable alternative available at the time, such as an unchanged long-run estimate for the same index. An anticipated low-probability contraction can occur even when the probability assessment was sound. Equally, a portfolio can earn exactly its forecast return because an unexpectedly large valuation gain offsets an earnings shortfall. Check the components before crediting the method.
Overlapping forecasts provide less independent evidence than the count of forecast dates suggests. Two ten-year forecasts issued one year apart share nine years of outcomes. When errors persist, investigate whether the source data changed, the economic relationship weakened, or the issue arose during implementation. Archive the original estimate before issuing a revision. A revised estimate is a new forecast vintage with a new information set.
See also. Forecasting Data and Biases develops the evidence checks. Economic Forecasting Methods compares tools for forming a view. Economic Growth and Structural Shocks, Business Cycles and Asset Returns, Inflation and Portfolio Exposures, Monetary and Fiscal Policy, and Yield Curves and International Linkages develop the common economic assumptions. Fixed-Income Return Expectations applies those assumptions to bond portfolios.
A Score the feature that drove the decision
An analyst forecasts 7% for equities and 4% for bonds. The corresponding realized returns are 2% and negative 1%. Both forecasts overshoot by 5 percentage points, yet the analyst correctly ranks the assets and predicts their 3-percentage-point spread. The ranking was informative for a choice between equity and bond exposure. The level errors still matter to a pension plan projecting the resources available to pay benefits. A review should report whichever dimensions informed the original decision.
B Recall check
Retrieve each distinction before reading its answer.
- Which two framework steps lie between historical research and interpretation of current conditions?
- Specify methods and information requirements, then select the best information sources.
- What must be true of the states in a probability-weighted mean?
- They are mutually exclusive, cover all possible outcomes, and have probabilities summing to one.
- Which condition ensures a covariance matrix gives a feasible variance for every portfolio?
- Positive semidefiniteness ensures all portfolio variances are nonnegative.
- Which forecast version belongs in an evaluation of a past decision?
- The estimate issued at the decision date, using the information then available.
- How much realized history do consecutive annual vintages of a ten-year forecast share?
- Nine years when their starting dates are one year apart.
Common mistakes
Table 6 distinguishes errors that can survive an otherwise correct calculation.
| Tempting choice | Why it looks plausible | Detail that settles the issue |
|---|---|---|
| Raise expected returns to the spending hurdle | The hurdle is an approved policy number. | Policy approval establishes a requirement. Market evidence determines the forecast. |
| Use a ten-year annualized return for next year | Both numbers are expressed per year. | The ten-year estimate summarizes growth across the entire horizon. |
| Reject a low equity forecast solely because cash yields more | Equities usually command long-run risk compensation. | A documented near-term valuation decline can reverse the expected ranking. |
| Validate a model using the final revised economic series | The revised series measures the economy more accurately. | The revision release date follows the decision date. |
| Declare an accurate process from zero mean error | The average realized return equals the average forecast. | Opposite signed misses cancel; absolute errors reveal their size. |
Approach
When a committee receives conflicting forecasts, trace the discrepancy to the first point at which the estimates differ.
Self-check quiz
Question 1
Using Table 4, which submission can enter the requested model without adjustment?
- US Treasury bills
- US corporate bonds
- Foreign equities
- US equities
Question 2
A portfolio must compound at 5% annually over two years on the committee's specified central wealth path. Year one's assumed return is 0%. Which year-two return completes the path?
- 5.00%
- 10.00%
- 10.25%
- 2.50%
Question 3
Using Table 5 and actual minus forecast, which pair gives mean error and mean absolute error, respectively, in percentage points?
- 0.00 and 0.00
- 0.00 and 2.67
- 2.67 and 0.00
- 0.00 and 4.00
Question 4
An analyst selects an equity model that infers return from current price and projected dividends. Historical research and specification of the model's input requirements are complete. What is the next framework step?
- Interpret current dividend growth relative to the business cycle.
- Publish the return implied by the latest available price.
- Compare the model's new forecast with its future realization.
- Select price and dividend data sources appropriate to the index.
Question 5
Two analysts forecast next year's domestic nominal bond and equity returns. Both use the same inflation assumption. The committee finds their forecasts inconsistent because the bond forecast assumes a recession while the equity forecast assumes an expansion. What best resolves the issue for a central allocation forecast?
- Estimate both assets in each common state and aggregate using the same state probabilities.
- Average the bond and equity forecasts and assign the result to both assets.
- Retain both because agreement on inflation establishes macroeconomic consistency.
- Increase the equity estimate until the expected equity premium matches its historical mean.
Answer key
Question 1 - Answer: A. Treasury bills have the required one-year US dollar nominal total return. B needs an inflation conversion, C needs the unhedged dollar return, and D summarizes a ten-year growth path.
Question 2 - Answer: C. The target ending growth factor is \(1.05^2=1.1025\). With capital unchanged in year one, year two must deliver 10.25%. A repeats the annual target, B adds the two annual rates without compounding, and D divides the annual rate by the number of years.
Question 3 - Answer: B. The signed errors are positive 3, negative 4, and positive 1 percentage points. Their mean is zero and their average magnitude is \(8/3\approx2.67\) percentage points. A allows cancellation in both measures, C reverses the measures, and D reports the largest absolute miss.
Question 4 - Answer: D. Once the method and required inputs are specified, the analyst selects suitable data sources. A interprets inputs before source selection, B issues a conclusion before analysis, and C belongs to monitoring after the forecast period.
Question 5 - Answer: A. A common set of states and weights produces comparable central estimates while preserving different asset sensitivities. B erases economically meaningful return differences, C overlooks the incompatible growth assumptions, and D imposes a historical premium without reconciling those assumptions.
Rapid review
- Required return → compare it with the market-supported estimate on the same basis.
- Forecast request → specify exposure, proxy, horizon, currency, hedge, and return convention.
- Information needs → define the model's inputs before choosing their sources.
- Named economic state → use the return conditional on that state.
- Central scenario estimate → weight mutually exclusive, exhaustive states by probability.
- Same horizon, different assets → reconcile assumptions and joint risk estimates.
- Different horizons, same wealth path → link successive gross returns by compounding.
- Historical forecast evaluation → preserve the issued vintage and match the realization.
- Offsetting forecast errors → inspect absolute misses alongside signed bias.
- Manager's expected active return → subtract the manager's comparable benchmark forecast from the portfolio forecast.
- Shortfall probability → use the distribution of future returns, including uncertainty within scenarios.
- Long-run state frequencies → distinguish them from probabilities updated for today's conditions.
- Unhedged foreign holding → translate ending local wealth at the ending exchange rate.
- Correct asset ranking → check return levels separately before judging a funding projection.
Keep going with the full CFA Level III Exam Prep Course
You just studied the way members do, in the same lesson interface. Full access unlocks every lesson, the question bank and flashcards for a full year.
One payment per program