## **Learning Outcomes**

This article explains how to benchmark investment managers using peer groups and universes, including:

- Defining peer groups and performance universes and clarifying their role in manager selection, monitoring, and performance evaluation.
- Distinguishing appropriate from inappropriate peers by mandate, style, risk profile, constraints, and investable universe.
- Describing practical approaches to constructing peer groups from databases, consultant universes, and style classifications while maintaining consistency over time.
- Evaluating the strengths and weaknesses of peer-group and universe benchmarking relative to custom, investable benchmarks.
- Interpreting manager performance using percentile ranks, quartiles, and median-manager comparisons across different measurement periods.
- Identifying key statistical and data biases—such as survivorship bias, backfill bias, heterogeneity, and reclassification effects—that can distort perceived relative performance.
- Assessing how universe construction choices influence apparent manager skill and the reliability of performance comparisons.
- Applying these concepts to typical CFA Level 3 exam questions that test judgment about benchmark quality, risk-adjusted interpretation, and the appropriate role of universe results in an overall evaluation framework.

## **CFA Level 3 Syllabus**

For the CFA Level 3 exam, you are expected to understand the key issues surrounding manager selection and peer group benchmarking, with a focus on the following syllabus points:

- Describing methods for constructing peer groups and universes for manager evaluation
- Explaining the advantages and limitations of peer group/universe benchmarking
- Interpreting a manager’s performance in the context of peer universes
- Identifying biases and common pitfalls in universe-based performance analysis

## **Test Your Knowledge**

_Attempt these questions before reading this article. If you find some difficult or cannot remember the answers, remember to look more closely at that area during your revision._

1. What are two major risks or biases associated with using universe or peer group benchmarks for manager performance evaluation?
2. Which considerations are important when constructing a peer group against which to evaluate a manager?
3. True or False: Benchmarking to a universe always offers a more objective comparison than benchmarking to a custom index.
4. If a manager lies at the 80th percentile in a peer universe, what does this indicate?

## **Introduction**

Peer group benchmarking and performance universes are widely used to compare investment managers' results. While appealing in their simplicity, these methods are susceptible to several biases that can undermine their reliability for performance evaluation. For the CFA exam, you must know how peer universes are constructed, how to interpret manager percentiles, and—most importantly—the key pitfalls of universe-based approaches. This article covers the key principles and required exam knowledge points in a succinct, assessment-driven way.

> **Key Term: peer group (peer universe)**
> A collection of similar investment managers or funds grouped together for the purpose of comparing investment performance on a relative basis.
>
> **Key Term: percentile rank**
> A statistical measure indicating the position of a manager’s performance result relative to the rest of a peer group or universe.

## Why Use Peer Groups and Universes

Peer group and universe comparisons are a common benchmarking approach in institutional and private investment management. They provide a method to judge whether a manager’s historical performance was above or below the “pack” of comparable strategies. By displaying percentile rankings (e.g., top quartile or 45th percentile), they also create a sense of competitive positioning.

Peer group benchmarking is typically used alongside (but never as a substitute for) a properly designed investable market or custom benchmark.

## Constructing Peer Groups

A peer group should include investment portfolios or funds with a similar mandate, objective, style, and constraints as the evaluated manager. Peer groups are commonly created by:

- Selecting funds from the same style or strategy (e.g., US large-cap growth equity)
- Using third-party databases of mutual funds, pension funds, or institutional accounts
- Filtering by investment process, risk tolerance, or geography

Ideally, peer groups should reflect comparable investable universes and risk exposures. The group should be consistent over time—retrospective “reshuffling” introduces lookback bias.

> **Key Term: survivorship bias**
> The tendency for poorly performing funds or managers to disappear from databases, resulting in upwardly biased universe performance statistics.

## Universes vs. Custom Benchmarks

Universe (peer group) benchmarking allows performance to be interpreted as a relative position (“top quartile”) rather than in absolute return terms. However, universes are not necessarily investable, and no investor can replicate the median or average peer group fund. For an investable, rules-based comparison, a custom or published index is preferred.

### **Worked Example 1.1**

**An equity manager’s annual return is 8.5%, placing them in the 10th percentile of a style universe, while the median return in the group is 6%. What does this mean?**

> **Answer:**
> The manager outperformed 90% of their peers in that investment style. However, such a percentile comparison does not guarantee future outperformance and may be impacted by biases in universe construction.

## Limitations and Risks of Universe Benchmarking

Peer group benchmarking suffers from several key limitations:

- Lack of investability: The median or average portfolio in a peer group is not a real fund. No investor can allocate to the “median manager.”
- Survivorship and backfill bias: Dead or closed funds drop out, which inflates historical returns. Some databases add new managers’ historical data only after strong early performance (backfill bias).
- Heterogeneity: Peer groups may include a wide range of constituent investment styles, risk levels, or constraints, making “apples to apples” comparison difficult.
- Subjectivity: Peer group construction typically requires subjective choices regarding inclusion criteria or style definitions.
- Changing Constituents: The list of peers may change over time, which can result in historical returns being adjusted and percentiles restated retroactively.

> **Key Term: backfill bias**
> The upward bias in historical universe returns caused when only successful managers add their prior returns to peer databases when joining.
>
> **Key Term: median manager**
> The portfolio whose return falls exactly in the middle of a peer group when ranked by performance.

### **Exam Warning**

> Do not assume a manager outperforming a universe is necessarily adding alpha. Many universes have a positive bias from survivorship and backfill, and relative performance is not specific to a defined, investable portfolio.

## Interpreting Percentile Rankings

Percentile ranking shows where a manager’s performance fits within their group:

- A low percentile (e.g., 5th percentile) means the manager is among the top performers.
- A high percentile (e.g., 95th) means near the bottom.

It is important to know that a percentile ranking is only as meaningful as the quality and relevance of the peer group. Differences in style, risk, and time horizon can distort the comparison.

### **Worked Example 1.2**

**A manager is in the 40th percentile in one peer universe but the 65th in another, for the same return period. Why the difference?**

> **Answer:**
> The respective groups may use different inclusion criteria (e.g., style definition, net vs. gross returns) or have different survivorship and backfill biases. This highlights the lack of standardization in universe benchmarking.

### **Revision Tip**

> Always cross-check universe-based peer group results with absolute returns versus a custom, investable benchmark.

## Summary

Benchmarking against peer groups and universes is common in manager performance evaluation, but it is not without serious weaknesses. Universes suffer from survivorship and backfill bias, lack investability, and can be highly sensitive to inclusion rules. While percentile rankings may appear simple, they are only as meaningful as the peer group’s quality and relevance. A custom investable benchmark remains the gold standard for portfolio evaluation, with peer group results best used as secondary context.

## **Key Point Checklist**

_This article has covered the following key knowledge points:_

- Describe peer groups/universes and their use in manager selection and monitoring
- Identify how to construct relevant peer groups for benchmarking
- Recognize limitations—survivorship bias, backfill bias, subjectivity, and lack of investability
- Interpret manager percentile rankings in the context of universe-based benchmarking
- Emphasize the primacy of investable custom benchmarks over peer universe comparisons

## **Key Terms and Concepts**

- peer group (peer universe)
- percentile rank
- survivorship bias
- backfill bias
- median manager

Manager selection and monitoring - Benchmarking peer groups and universes