How to benchmark your portfolio

Every portfolio tool shows you a benchmark, and most people never change it. The comparison then answers a question nobody asked, and it usually answers it flatteringly in one direction or brutally in the other.

A benchmark is not a scoreboard. It is a counterfactual: the return you would have got from an alternative you could genuinely have held instead. Choosing one badly does not make your portfolio worse. It makes the number beside it meaningless, which is worse than having no number, because a meaningless number still gets acted on.

The four ways a comparison misleads

1. The benchmark holds something different from you

This is the big one and it is almost universal. If you hold a globally diversified portfolio with a bond sleeve, and you compare it to a US large-cap equity index, the gap you are looking at is mostly your asset allocation, not your selection. You already decided to hold bonds and non-US equity. The comparison then charges you for that decision every year the US market leads, and credits you for it every year it lags, and in neither case tells you anything you did not already know when you set the allocation.

Some illustrative arithmetic. Say you hold 60% equity and 40% bonds, the equity sleeve does 12% and the bond sleeve does 2%:

Measure	Result
Your portfolio	8.0%, the weighted blend of your two sleeves
An all-equity index	12.0%, what you would see beside your figure
An allocation-matched blend	8.0%, the actual counterfactual

Those numbers are invented. The point they show is not: the same portfolio is "4 points behind" or "exactly on target" depending only on which comparison is displayed. Both figures are arithmetically correct. Only one of them is answering "did my choices work".

2. Price return compared against total return

A published index level usually excludes dividends. Your own portfolio return almost always includes them, because you received them.

Compare the two and you understate your own performance by roughly the yield of what you hold, every year, silently. Over a long horizon that is not a rounding error. The reverse mistake exists too: index providers publish total return versions of the same index, and comparing a price-only portfolio figure against one of those flatters the index.

The fix is not complicated, it is just easy to skip: find out which basis each side of the comparison uses, and make them the same. A figure labelled only "the index" does not tell you, and the difference between the two versions of one index is larger than most of the gaps people worry about.

3. The currency the comparison is measured in

If you report in Canadian dollars and the benchmark is a US index quoted in US dollars, the two series are denominated differently, and the exchange rate moves between them.

There is no universally right answer here, but there are two coherent ones:

• Convert the index into your reporting currency, so both sides experience the same currency movement you did. This measures your outcome.
• Compare in the index's own currency, so the comparison isolates the securities from the exchange rate. This measures the holdings.

What is not coherent is comparing an unconverted index level against a converted portfolio return, which is the default in more tools than you would expect. The time-weighted versus money-weighted distinction has the same character: the figures are all correct and the question is which one you are asking.

4. Comparing the wrong kind of return

An index has no deposits and no withdrawals. Your portfolio has both.

That means a money-weighted return, which deliberately reflects when your money arrived, is not comparable to an index at all. Comparing them measures your contribution schedule as much as your holdings. Use the time-weighted figure for any comparison against a benchmark, and keep money-weighted for the separate question of how your own money did. The time-weighted vs money-weighted calculator computes both from the same series, which is the quickest way to see how far apart they can sit on one portfolio.

Annualising compounds the problem, because an average return and a compound return are different numbers pulled from the same series. See CAGR versus average annual return for the gap between them and why the larger of the two is the one usually quoted. The CAGR calculator reports the compound rate between two balances and, once you declare your deposits, the rate your money actually earned.

Building a benchmark that matches what you hold

The honest version of a benchmark for a diversified portfolio is a blend: one index per sleeve, weighted to your targets, with a stated rebalancing rule.

Three things make it work, and each one is a place people quietly cheat:

• The weights are your targets, not your current holdings. Deriving weights from what you happen to hold today guarantees the benchmark tracks you, which removes the entire point of having one.
• The rebalancing rule is fixed in advance and applied mechanically. A benchmark rebalanced whenever it flatters you is not a benchmark.
• It is set once and left alone. Changing your benchmark after a bad year is the cleanest way to learn nothing from the year.

A blend takes some setup and it is the only comparison that can actually tell you whether your selection added anything, because it holds your allocation decision constant and varies only what you picked inside it.

Three yardsticks that answer better questions

A market index is not the only counterfactual available, and for a retail investor it is often not the most useful one.

Yardstick	The question it answers
The allocation you chose	Did my selection and timing beat simply holding my own targets?
A single fund you'd buy	Was all of this better than the one-ticker alternative?
The risk-free rate	Was I paid anything for taking risk at all?

The second is the most uncomfortable and the most clarifying, particularly for a portfolio of individual positions: the realistic alternative to picking holdings is usually one asset-allocation ETF, not the S&P 500, and that is the comparison that says whether the effort paid. The third matters in the years people forget it: a portfolio that returned less than a government bond yield took risk for nothing, and no index comparison surfaces that.

What Luum compares against

Stated plainly, because a benchmark feature that is vague about its basis is part of the problem this page describes.

• A fixed set of broad indexes and index ETFs, covering US large-cap, Canadian, US small-cap, global developed and total-international exposure. These are price returns, so they exclude dividends, the mismatch in point 2 above, named rather than hidden. Each one is converted into your portfolio's own base currency before comparison, so both sides carry the same currency movement.
• The default is the S&P 500, for every portfolio, Canadian or US. It is a default, not a recommendation, and for most Canadian portfolios it is an allocation comparison rather than a selection one.
• "vs the target you set", the allocation counterfactual from the table above. It holds your own target weights, receives your own contributions and withdrawals on the dates they actually happened, and rebalances monthly. It charges no tax and no trading costs, so it is illustrative rather than attainable. It appears only when the portfolio has an allocation target, because without one there is nothing to compare against.
• The risk-free rate, as a compounded Government of Canada 5-year yield. It is deliberately not filed with the indexes: it is a government bond yield rather than something you bought.
• Up to three comparisons you choose yourself, each one instrument's price return, available when that instrument has enough price history to cover the period you are looking at.

How portfolio returns are calculated sets out the method behind each figure.

What to keep track of

• Which basis each side uses, price or total return, and in which currency. Without both, a gap is not interpretable.
• One benchmark, chosen before the period, rather than the best-looking one chosen after it.
• The comparison against your own targets, which is the only one that isolates selection from allocation.
• Your combined figure across accounts, since a benchmark applied to one account of several compares against a fraction of your portfolio. That is the same reason per-account views mislead.

This article is educational. It does not recommend a benchmark, an index or an allocation, and choosing between them depends on circumstances this page knows nothing about.

Common questions

What benchmark should I compare my portfolio to?

The one that holds what you decided to hold. For a diversified portfolio that means a blend, one index per sleeve, weighted to your targets, rather than a single equity index, because a single index comparison measures your allocation decision rather than your selection. If you want one comparison instead of several, the most informative is usually against the allocation you set for yourself.

Why is my portfolio always behind the S&P 500?

Frequently because it is not trying to be the S&P 500. A portfolio holding bonds, Canadian equity and international equity will trail a US large-cap index whenever US large-cap leads, by construction and not by error. The comparison worth making is against a blend matching your own targets. If your portfolio is in fact US large-cap equity and still trails, that is a selection question and a real one.

Should the benchmark include dividends?

It should match whatever your own figure includes, and your own figure almost certainly includes dividends because you received them. Most quoted index levels are price returns and exclude them, so the common error is comparing your dividend-inclusive return against a dividend-exclusive index and concluding you underperformed by roughly the yield. Index providers publish total-return versions for exactly this reason.

Is it wrong to change my benchmark?

Changing it because your holdings or targets genuinely changed is correct. Changing it after a period you did not like is how a comparison stops being able to tell you anything, since a yardstick chosen with the result already known cannot inform the judgement it is supposed to support. Pick before, not after, and record which one you picked.