What the Index DecidesEasy
An index is described as a thermometer and behaves as an allocator. Its rules decide what is bought, in what proportion, on which date — and the buying is done by people who never read the rules.
6 min read · 1 022 words · Updated
A measurement that became an instruction
- An index began as a summary. One number standing for a market, so that a conversation about "the market" had something behind it. What an index really is is that story in plain words.
- It is now an instruction, because money follows it. The moment a holding is required to look like an index, the index's rules are orders: buy this, in this proportion, by this date, at whatever price is available then.
- Which means every construction decision allocates capital. Not as a side effect — as its primary consequence. The rule that a company must be listed in a particular place, or must have a minimum free float, or must not have been listed for fewer than so many months, decides where money goes.
- And the rules were written to make a good measurement, not to make a good allocation. That mismatch is the whole subject of this page, and it is neither a scandal nor a defect: it is what happens when a description is used as a specification.
Four decisions, each with a consequence
- Inclusion — what counts as being in this market. Domicile, listing venue, sector classification, size threshold, minimum trading. Each is defensible; together they mean a fund tracking "the market" holds a specific list somebody chose, and excludes companies for reasons unrelated to their businesses.
- Weighting — how much of each. By market value, by an equal share, by a fundamental measure, capped at a ceiling. Market-value weighting is the default because it is the only scheme that requires no trading as prices move; that convenience, and not a claim about merit, is why it is everywhere.
- Free float — whose shares count. Only shares actually available to buy, so a company whose founder holds most of it enters at a fraction of its size. Reasonable, and it means index weight is not company size.
- Rebalancing — when the list changes. On an announced date, at a published price, in a known direction. The most predictable trade in any market is the one an index has already told everybody it must do.
Who is forced, and when
- A fund promising to track cannot exercise judgement about the price. Its job is to match the index, and matching means trading on the effective date whatever the price is doing. That is a forced buyer, publicly scheduled.
- Which is exactly the participant every asset-class page here separates out — the ones who have to trade, as against the ones who are choosing to. A price on a rebalancing date is set by the first group.
- And the other side knows. A published rule with a published date is an invitation, and the compensation for supplying that liquidity is a real cost borne by the holders of the fund — invisible, because it appears as tracking difference rather than as a fee. Costs and fees covers where it surfaces.
- The general lesson is the site's recurring one: a price is set by whoever must trade, and an index is a machine for creating people who must. The other side of the trade asks it the other way round.
What a benchmark quietly assumes
| The assumption | What it does to the holding |
|---|---|
| That size is the right weight | More money goes to whatever has already risen most, by construction rather than by choice. |
| That the listing venue defines the market | A domestic index can be dominated by revenues earned elsewhere, so a "home market" holding is a currency and geography bet nobody made. What a currency does is the other half. |
| That a sector label is a fact | Classification moves companies between sectors by decision, and every sector-weighted position moves with it. |
| That the past constituents are the same series | A long history of an index is a history of successive different lists, and the ones that left are usually the ones that failed. |
| That the number is comparable to your return | Most published levels ignore dividends. A price index and a total-return index of the same market are different series, and comparing a fund to the wrong one is arithmetic rather than performance — the question page is the plain version. |
The survivorship point, made carefully
- Companies that fail leave the index. Not through any concealment — the rules remove what no longer qualifies — but the resulting series is of a list that is continuously repaired.
- Which flatters nothing about a holder who tracked it, because they held the failures on the way down and the replacements afterwards. The index return is the honest return of following the rules.
- What it does flatter is inference. "This market returned X over fifty years" describes a rule for continuously replacing constituents, and a study of today's constituents over that period describes something that never existed. Reading a market number is where that error is general.
Reading an index before trusting one
- Find the methodology document, which is published, and read the inclusion rule and the rebalancing schedule. Two pages, and they say what the thing actually is.
- Check whether the level you are quoting includes income. This one difference explains more disagreements about performance than everything else combined.
- Count the top few weights. A broad index is frequently a concentrated position with a broad name, and the concentration is a fact about the rule rather than an opinion about the companies.
- Ask what the rule forces, and when. Not to trade around it — but because it explains price behaviour on specific dates that otherwise looks like news.
- And keep the two roles apart. As a measurement an index is a genuine public good; as an instruction it is a set of decisions somebody made for a different purpose. Both are true at once, and a reader who holds only the first has no account of half of what they see.
Information and education only. This describes how index construction works in general terms and names no specific index, provider or fund. It is not advice, not a recommendation of any approach, and nothing here takes account of your circumstances.
More in Analysis
- EasyWhat kills a dealFive ways a transaction fails to happen, counted across every deal on the site — and the one that decides most of…
- EasyWhat the Wrapper ChangesThe same exposure delivered five different ways
- EasyWhat Actually Drives a ReturnEvery return decomposes into four things, and only one of them is the story people tell
- EasyWhich Document GovernsThe document a reader is shown and the document that decides what happens are almost never the same one