Author Details

Milan Horvath

Milan Horvath

Senior Associate, MSCI Research

Hamed Faquiryan

Hamed Faquiryan

Executive Director, MSCI Research

Andrew DeMond

Andrew DeMond

Executive Director, MSCI Research

Social Sharing

Extended Viewer

Matching Portfolios and Clients’ Expected Returns, with Factors

  • Today, many investment advisers implement strategic allocations based on broad asset-class assumptions that may ignore the characteristics of specific products.
  • Investment products within an asset class may have very different risk-adjusted returns, for example.
  • We develop an alternative approach that may enable more precise fund selection by estimating detailed expected returns for investable portfolios from broad strategic assumptions.

For investment advisers, matching the preferences of clients (often individuals) to an ever-growing set of investment products represents a fundamental portfolio-construction problem. Standard practice today is to build strategic asset-allocation models that combine broad capital-market assumptions (CMAs) with risk-return preferences. Clients and products are then matched according to a product’s asset class and the client’s risk-return preferences. Products within an asset class may have very different risk-adjusted-return profiles, however. Hence, matching allocations to products can be an imprecise process.


A new approach to matching clients and investments

We develop an alternative approach to this problem by combining CMAs with the MSCI Multi-Asset Class (MAC) Factor Model. This process is composed of two distinct steps. First, we estimate expected factor returns implied by a set of CMAs.1 Then we compute expected returns for any portfolio using its exposures and our CMA-implied expected factor returns.

As an example, the table below enumerates five-year hypothetical CMAs for a set of 14 fixed-income and equity indexes. This set of assumed expected returns is illustrative and intended to cover a sample of available products for a hypothetical investment adviser.


Assumed expected returns for a representative set of products


The choice of factor model (e.g., horizon, granularity and coverage) is the key ingredient of our approach and depends on the nature of the investment universe available to an adviser. For our example, we use the tier-two factors from the MSCI MAC Factor Model2 because it is aligned with the granularity of the assumed CMAs. The following exhibit shows that CMAs can be decomposed to factor returns.


The hidden factor structure of expected returns



The expected factor returns implied by the input expected index returns are illustrated in the exhibit below.


Implied factor returns

MSCI MAC Factor Model tier-two factor returns implied by the CMAs in the table above.


To make this process more concrete, the exhibit below plots the ratio of expected returns and risk forecasts for a set of anonymized funds, as well as their corresponding benchmarks. Funds A, B and C are pure equity funds. The bar chart above makes clear that strategic asset allocations that rely on asset-class categories would have obscured essential fund differences for both risk and expected return. For instance, our sample equity funds’ factor-implied total risk-adjusted expected returns differ by nearly 50%. Capturing this fund-level diversity may be equally vital for any given set of investor needs.


How the model maps index CMAs to funds’ expected returns

Ratios of factor-implied total expected returns and total risk forecasts for a representative set of anonymized equity mutual funds and their respective benchmarks as of Dec. 12, 2022.


These factor-implied estimates of risk and return may enable investors and their advisers to explicitly decide on trade-offs — not at the level of asset classes or even indexes, but for specific funds. After all, each fund comprises its own specific set of exposures to the drivers of risk and return. Factors may then act as a more precise tool to help advisers select products tailored for an individual investor’s preferences.



1In general, a set of CMAs may contain redundant information and an important technical detail of the process is reducing CMAs to the smallest possible set of factors with the greatest amount of explanatory power. For more details on this process, interested clients can send an email to for more technical documentation.

2For more details on the model and its features, please see: Peter Shepard, Andrew DeMond, Limin Xiao, Chenlu Zhou, and Jennifer Ahlport. “The MSCI Multi-Asset Class Factor Model.” MSCI Model Insight, January 2020.



Further Reading

Carrying on Through a Crisis, with Factors

The MSCI Multi-Asset Class Factor Model (client access only)

MSCI Fixed Income Factor Model (client access only)