How an Equity Index Is Built

Someone tells you the market rose 1.2% today. Rose by whose measure? There are thousands of listed companies, they moved in every direction, and no natural law says how to blend them into one number. Somebody had to choose — which companies count, and how loudly each one gets to speak. An index is that choice written down. It is a rule, not a measurement.

Getting comfortable with this matters more than it sounds. An index is the benchmark a fund is judged against, the thing a tracker fund must replicate, and the definition of "the market" in every risk model. Change the rule and you change all three.

Who gets in

Every index starts with an eligibility screen: a country or exchange, a minimum size, a minimum liquidity, a free-float requirement, sometimes a sector or a profitability test. A committee or a mechanical rule then picks the members and sets a schedule for reviewing them, typically quarterly or annually.

This is not a formality. Because trillions of dollars track major indices, being added forces every tracker to buy — the inclusion effect — so the eligibility screen is itself a market-moving instrument.

How loudly each member speaks

Then comes the weighting rule, where the real disagreement lives.

Price-weighted. Each member's weight is its share price divided by the sum of all prices. Simple, historic, and slightly absurd: a company's influence depends on how many pieces it happens to have sliced itself into. A 2-for-1 split halves its weight without changing anything real.

Capitalisation-weighted. Weight by market value — price times shares outstanding. This is what most major indices use, because it represents the aggregate portfolio actually held by all investors and needs almost no trading to maintain.

Float-adjusted capitalisation-weighted. The refinement everyone now uses in practice. Only shares genuinely available to the public count, excluding government stakes, founder blocks and cross-holdings. If half a company is locked up, the index shouldn't ask trackers to buy shares that are not for sale.

Equal-weighted. Every member gets the same weight, which tilts the index toward smaller companies and requires constant rebalancing as prices drift apart.

Worked example: one day, three answers

The four companies above:

	price	shares	market cap
Alpha	300	20m	6.0bn
Beta	40	400m	16.0bn
Gamma	120	100m	12.0bn
Delta	20	300m	6.0bn

Prices sum to 480, so price weights are $300/480 = 62.5\%$, then 8.3%, 25.0% and 4.2%. Total capitalisation is 40bn, giving cap weights of 15%, 40%, 30% and 15%. Equal weights are 25% each.

Now suppose Alpha rises 10%, Beta falls 5%, Gamma is flat and Delta rises 20%.

• Cap-weighted: $0.15(10) + 0.40(-5) + 0.30(0) + 0.15(20) = 2.5\%$.
• Equal-weighted: $(10 - 5 + 0 + 20)/4 = 6.25\%$.
• Price-weighted: $0.625(10) + 0.083(-5) + 0.25(0) + 0.042(20) = 6.67\%$.

Same four stocks, same day, and "the market" was up 2.5%, 6.25% or 6.67% depending entirely on the rule. Big Beta's fall dominates the cap-weighted number and barely registers in the other two.

The divisor: keeping the line continuous

Index levels must not jump when something happens that changes prices but not value. The fix is a divisor:

$$\text{index level} = \frac{\text{sum of prices (or caps)}}{\text{divisor}}.$$

In words: the numerator carries real market moves, and the divisor is a fudge factor reset whenever a mechanical change would otherwise create a fake one.

Take the price-weighted index. Prices sum to 480 and the divisor starts at 4, so the level is 120. Gamma now does a 2-for-1 split: its price halves to 60 and prices sum to 420. Nothing about anyone's wealth changed, so the index must still read 120. Set the new divisor to $420 / 120 = 3.5$ and it does. The same mechanism absorbs additions, deletions, spin-offs and rights issues.

Common pitfalls

• Comparing price return to total return. Most headline indices exclude dividends. Over decades that gap is enormous, and quoting the wrong one flatters or damns a strategy unfairly.
• Assuming the index is the market. It is a sampled, screened, committee-shaped subset with rules that change.
• Ignoring concentration limits. Many indices cap any single member's weight, so a mega-cap's true influence is smaller than its market share implies.
• Backtesting against today's members. Reconstruct the historical membership or survivorship bias will hand you returns nobody could have earned.