Customised benchmark

Performance only means something against a fair comparison. A value investor running a concentrated portfolio should not be judged against an index full of growth stocks, and a factor-based portfolio may not resemble any published index at all.

When market indices fail the tests of a good benchmark, the workbook's answer is a customised benchmark. A portfolio manager following a particular strategy and style builds an investment universe reflecting that style to focus its research, and then picks the most attractive securities from it. That universe itself can serve as the benchmark.

The benefit is a truly like-for-like comparison. The cost is exactly what it sounds like — building and maintaining it is expensive.

When it is needed (Chapter 10, section 10.4.2). When market-based indices do not meet the criteria of a good benchmark given the manager's strategies and style — the workbook's example is factor-based investing.

The five criteria of a good benchmark (section 10.4.1).

1. Constituents and their weights are clearly defined.
2. It is investable — a passive exposure to it is possible.
3. It is consistent with the portfolio's investment approach (blue-chip portfolio → blue-chip benchmark; value style → value-oriented benchmark).
4. It has the same risk-return profile as the portfolio.
5. Its performance is measurable.

Advantage and disadvantage. A customised benchmark meets the requirements of a valid benchmark. But the cost of construction and maintenance is much higher than the fee paid for using market-based indices.

The regulatory layer for PMS (Chapter 9, section 9.3.8). For performance reporting, each investment approach is tagged to one strategy, APMI prescribes a maximum of three benchmarks per strategy, and the portfolio manager selects one of those. So a customised benchmark is an evaluation tool; the benchmark shown in regulated performance disclosures must come from APMI's list.

What benchmarks feed. Chapter 10's attribution analysis splits excess return over the benchmark into allocation, selection and interaction effects — in its example, a portfolio return of 8.70% against a benchmark of 7.80%, an excess of 0.9%.

Illustrative manager and figures, except where noted.

A portfolio manager runs a quality-factor equity approach: only companies with strong profitability, low debt and stable earnings growth (Chapter 9's definition of the quality factor).

The problem. NIFTY 50 holds banks with high leverage and cyclical commodity companies. In a year when cyclicals rally, the approach lags NIFTY 50 by 4 percentage points — not because stock selection was poor, but because the index contains businesses the strategy excludes by design.

The customised benchmark. The manager defines its screened universe of about 120 quality stocks, weighted by clearly stated rules, and computes the universe's return. That year the universe returns 11%; the portfolio returns 12.5%.

The second comparison answers the question an investor who chose a quality approach actually cares about. But for regulated performance disclosure, the manager must still show the approach's TWRR against the one APMI-prescribed benchmark it selected for the strategy.

Chapter 10 (Performance Measurement and Evaluation of Portfolio Managers, 10% weightage), sections 10.4.1 and 10.4.2, with the regulatory benchmark rules in Chapter 9, section 9.3.8. Expect questions on the criteria of a good benchmark, on when a customised benchmark is appropriate, and on its main disadvantage (cost).

• The disadvantage is cost, not validity. The workbook says customised benchmarks meet the requirements of valid benchmarks.
• "Investable" is one of the five criteria — a passive exposure must be possible.
• The benchmark must match the portfolio's style, not simply the broad market.
• Regulatory disclosure uses APMI's list — a maximum of three benchmarks per strategy, one selected per investment approach.
• Changing a PMS benchmark tag requires offering a no-exit-load exit and drops the prior track record from reporting.
• Benchmarking and peer group analysis are the two parts of performance evaluation in Chapter 10 — a customised benchmark is not a peer group.