Performance measure

Question 1

Performance Measurement

Answer 1

Measures what happened. Do both absolute (total return) and relative (active return vs. benchmark). You can also report risk and risk-adjusted returns here (e.g., volatility, Sharpe).

Question 2

Performance Attribution

Answer 2

Explains how the return (or risk) happened—e.g., allocation, selection, and interaction (equity); curve, spread, currency (FI); or factor contributions.

Question 3

Performance Evaluation/Appraisal

Answer 3

Judges’ skill vs. luck (is performance repeatable/skillful?). Uses statistics such as information ratio, t-statistics, persistence, and checks benchmark quality.

Question 4

Return Attribution

Answer 4

Analyzes the impact of active investment decisions on return

Question 5

Risk Attribution

Answer 5

Analyzes the risk consequences of those decisions

Question 6

Micro Attribution

Answer 6

Understanding the drivers of a manager’s returns and whether those drivers are consistent with the stated investment process.

Question 7

Macro Attribution

Answer 7

Measures the effect of the asset owners’ choice to deviate from the SAA.

Question 8

An Effective Attribution Process must consist of…

Answer 8

1. Account for all the portfolios’ return or risk exposure
2. Reflect the investment decision-making process
3. Quantify the active decision of the PM
4. Provide a complete understanding of the excess return/risk of the portfolio

Question 9

Return-based attribution

Answer 9

• Uses only total portfolio return
• Easiest to implement
• Least accurate
• Vulnerable to data manipulation
• Key assumptions/limitations:
• Model specification risk (wrong factors = misleading “exposures”).
• Exposures can be time-varying; a single regression over a long window averages them.
• “Alpha” is sensitive to factor set and constraints (e.g., whether weights must sum to 1, be non-negative).

Question 10

Holding-based attribution

Answer 10

• Uses actual holding from the beginning of the period
• All transactions are assumed to occurs at end of the day
• Accuracy improves when data has shorter time intervals.
• where performance came from vs a benchmark: allocation, selection (and interaction, depending on model).
• Key limitation: if turnover is meaningful, holdings-based can misattribute because it typically ignores intra-period trades (or approximates them).
• Timing assumption: the common simplifying assumption is all trades occur at the end of the period (or end of day) unless you have more granular holdings snapshots. That’s exactly why it breaks down with higher turnover.

Question 11

Transaction-based attribution

Answer 11

• Uses both the holdings and the transactions during the evaluation period
• Most accurate, but most difficult and time consuming to implement
  What it adds vs holdings-based: explicitly separates decision timing (when you traded) and often decomposes into effects like:
• Market/benchmark movement while you held the position
• Trading effect (implementation, timing, slippage)
• Sometimes cash flow / rebalancing effects
• Practical requirement: you need clean trade, price, and corporate action data, plus consistent mapping to benchmark/segments. Data quality is the usual failure point