Choosing a benchmark for your own portfolio

You now have a rate. A rate alone is not an assessment, because there is no such thing as a good return in the abstract — only a return that was better or worse than what was available for the same risk.

The Mutual funds subject covers choosing the right benchmark for judging a fund. This chapter is the harder and less-discussed problem: a benchmark for your whole portfolio, which no factsheet provides.

Why a fund's benchmark will not do

A fund benchmark compares one category against its own index. Your portfolio holds several categories at once, in weights you chose.

If you hold 60% equity and 40% debt and compare yourself to an equity index, you will lose in every rising market and win in every falling one — and neither result tells you anything, because you were never trying to be an equity fund. The comparison measures your asset allocation, which you already know, rather than your execution of it.

The four requirements

1. It must match your asset mix. The benchmark's weights should be your weights. Otherwise you are measuring allocation rather than skill.

2. It must be a total return index. A price index counts only price changes; a total return index adds the dividends or coupons the constituents paid. You received those payouts, so a benchmark that excludes them is being asked to run with weights on.

SEBI made this mandatory for funds precisely because the gap is material — over a decade with a 1.5% dividend yield it is roughly the size of an active fund's entire fee. Clause 6.9.1(a) requires scheme returns to be shown against the benchmark Total Return Index. Hold yourself to the same standard, because comparing your portfolio to a price index flatters you by the dividend yield every single year.

3. It must be investable. The benchmark should be something you could actually have bought instead — an index fund that exists, at its real cost. A theoretical index with no tracking error and no expense ratio is not an alternative you were ever offered.

4. It must be chosen in advance. This is the one people break. Choosing the benchmark after seeing your result means you will pick the one you beat, every time, without noticing you are doing it. Write it down before the period starts.

Building a blended benchmark

For a multi-asset portfolio, weight the component indices by your own target allocation:

$R_{\text{benchmark}} = \sum_i w_i R_i$

where $w_i$ is your target weight in asset class i and $R_i$ the total return of a representative investable index for it.

Use target weights, not actual drifted weights. If your plan is 60/40 and drift took you to 68/32 during a rally, benchmarking against 68/32 credits you for the drift — but the drift was not a decision, it was the absence of one. Benchmarking against your target measures whether rebalancing was worth doing, which is the question.

Rebalance the benchmark on the same schedule you rebalance. Otherwise you are comparing a rebalanced portfolio with an un-rebalanced benchmark and attributing the difference to skill.

What beating it proves, and does not

Over one year: almost nothing. Single-year differences of a point or two are within the range that asset-class timing, a slightly different index, or plain luck produces. A year is too short to contain evidence about process.

Over five or ten years, consistently: something, but less than it feels. Persistent outperformance can come from genuine skill, from taking more risk than the benchmark — which chapter 6 adjusts for — or from a style that happened to suit the period.

What it never proves is that the next period will repeat. Recency and reinforcement learning in Behavioural finance is precisely about the error of concluding otherwise.

The most useful comparison is often the humblest one: your portfolio against a single cheap diversified index fund at the same risk level. If you cannot beat that over a long period net of your costs and your time, the index fund is the better answer — and discovering this is a success of measurement, not a failure of investing.

Working the problem

60% Indian equity, 25% debt, 15% gold; portfolio returned 11%.

An honest benchmark, weighted as held, each component a total return index and each investable:

Weight	Component
60%	A broad Indian equity total return index, as tracked by an index fund that exists
25%	A debt index matching the actual duration and credit quality held
15%	A domestic gold price series, including the cost of the vehicle used

Benchmark return = 0.60 × equity TRI + 0.25 × debt index + 0.15 × gold.

Two details that decide whether the comparison is honest. The debt component must match the duration held — benchmarking a short-duration portfolio against a long-duration index makes you look brilliant when rates rise and foolish when they fall, for reasons unrelated to skill. And the gold component must be net of how you actually held it; the Gold and commodities chapters cover why making charges and vehicle costs differ enough to matter.

Then subtract your costs — expense ratios, brokerage, and any advisory fee — because the benchmark alternative has its own costs and you should compare like with like.

If you beat it by two points, what would I conclude? Over a single year, essentially nothing about skill. Specifically:

Two points is well inside noise for a multi-asset portfolio over one year. Small differences in index choice alone can produce it: a different equity index, a debt index of slightly different duration, or a gold series measured at a different point all move the benchmark by a point or more.

Check first whether it is an allocation artefact. If your actual weights drifted from 60/25/15 and your best-performing asset drifted upward, part of the two points is drift rather than selection — which is why the benchmark uses target weights.

Check whether you took more risk. If your equity sleeve was concentrated in small caps while the benchmark used a broad index, you did not beat the benchmark; you ran a different portfolio that happened to win. Chapter 6 is the adjustment that makes this visible.

What I would actually do: record it, change nothing, and look again after five years with the same benchmark written down in advance. The discipline of fixing the comparison beforehand is worth more than any single year's result, because it is the only thing that stops the benchmark from being chosen to flatter.

The point

A return needs a comparison to become an assessment, and your benchmark is not a fund's benchmark — it must match your own asset weights, use total return indices rather than price indices, consist of things you could actually have bought at their real cost, and be chosen before you see the result. Blend component indices at your target weights rather than your drifted ones, so the comparison measures rebalancing rather than crediting you for its absence. Beating it by a point or two over one year is within noise from index choice alone; the value lies in fixing the comparison in advance and looking over a decade.