Indexes · Large cap
S&P 500, Selection and Float Weighting
Not the 500 largest American companies: a committee selection from companies that pass an earnings gate, weighted by the shares actually available to trade. Both of those depart from what the name suggests.
Selection, not ranking
The index is built in two stages. Published eligibility criteria cover US domicile, an eligible listing, liquidity relative to market value, a minimum public float, and profitability, positive earnings in the most recent quarter and across the trailing four quarters. A committee then selects from the eligible companies.
Two things follow that the name obscures. A company can be large enough and still not be a constituent, because eligibility is not membership and the committee decides. And membership is not reproducible from the published rules: two people applying them carefully will produce a list, and it will not be exactly this one. Anyone testing a strategy on index membership is testing a committee’s output as well as a market.
The earnings gate
Positive earnings in the latest quarter and over the trailing four quarters is a requirement for entry. It is worth stating plainly because it is often assumed to be absent from a headline index, and because it is the same rule that makes the S&P 600 behave unlike the Russell 2000.
As with the small-cap index, the gate applies at entry. A constituent that later makes losses is not automatically removed, so the index is a set of companies that were profitable when they joined. That is a weaker claim than "an index of profitable companies", and it is the accurate one.
What weighting does to the number
Influence follows float-adjusted market value, so the largest constituents dominate. That single property produces the behaviour the whole breadth section exists to detect: an index can rise while the majority of its own members fall, carried by a handful of heavyweights, and nothing in the index number reveals it.
Two comparisons make the gap visible and both are free. The index against its own equal-weighted version answers the question in returns: when the weighted version outruns the equal-weighted one over weeks, the largest members are doing the work. The index against an advance/decline count answers it in participation. They frequently agree, and when they do not, the disagreement is the information.
Rebalancing, and the dates to avoid
| Event | Effect |
|---|---|
| Quarterly share and float update | March, June, September, December. Every tracking fund adjusts at the same reference prices, concentrating obliged trading into the close. |
| Addition or deletion | As events require. Moves the individual stock hard — funds must buy or sell it — and the index barely at all. |
| Quarterly options expiry | Coincides with the rebalance dates, adding a second source of mechanical volume to the same sessions. |
For anything on this site that reads volume, those are dates to exclude rather than interpret. The volume is real and it expresses no view: funds trade because a rule tells them to. A cumulative volume line or a breadth reading computed across a rebalance is recording index maintenance.
Divisors and why the level is not a price
One mechanical detail explains a question the index invites. The published level is not a sum of prices or of market values. It is the total float-adjusted market value divided by a divisor, a number adjusted whenever the index composition changes so that the level does not jump.
Without it, replacing one constituent with a larger one would move the index by an amount that had nothing to do with the market. The divisor absorbs those changes, which is why the level is continuous across additions, deletions and share-count updates, and why the level itself is a scale with no natural units. Only its changes carry meaning, exactly as with the cumulative measures in the indicator library.
Deletions are the part nobody watches
Additions get the attention, a company entering the index is news, and the buying by tracking funds is visible. The deletions on the other side of each change are the same size in trading terms and pass almost unremarked.
They also matter for anyone using index membership in a study. A current constituent list contains no trace of the companies that were removed for shrinking, failing, or being acquired, so a historical test run on today's membership measures the survivors and reports a flattering result. This is the survivorship problem in its most common form, and it is why complete historical membership, with the date of every addition and deletion, is a commercial product while the current list is free.
Why the misdescription matters
"The 500 largest US companies" is the version almost everyone carries, and each part of it is wrong in a way that changes conclusions. Not the largest, a committee selects. Not simply 500 companies by size; an earnings gate applies. Not weighted by company size, weighted by the shares available to trade.
None of that makes the index worse; it makes it a specific rule with specific properties, which is exactly how this reference treats every measure. The number is only as meaningful as the methodology behind it, and quoting the index without knowing the methodology is the same error as quoting an oscillator level without its period.
Frequently asked questions
Is the S&P 500 the 500 largest US companies?
No, and it never has been. It is a selection made by a committee from companies that meet published eligibility criteria, and those criteria include profitability, a minimum public float, liquidity and US domicile. Companies large enough by market value are routinely absent, and the committee has discretion over both additions and timing. "The 500 largest" is the most common misdescription of the index.
What are the eligibility criteria?
US domicile, a listing on an eligible exchange, sufficient liquidity measured against market value, a minimum public float, and positive earnings, the most recent quarter and the sum of the trailing four quarters. Meeting them makes a company eligible rather than a constituent; the index committee decides who is actually added, and it can leave an eligible company out.
What does float adjustment do?
It counts only the shares available to trade. Stock held by founders, governments, or in cross-holdings by other companies is excluded from the weight, so a business whose shares are largely locked up carries less index weight than its headline market value implies. Without the adjustment the index would be pushed around by shares nobody can buy, a real distortion before the major providers moved to float weighting in the 2000s.
When is it rebalanced?
Share counts and float are updated on a quarterly schedule, in March, June, September and December, while additions and deletions happen as events require, a constituent taken private or merged is replaced promptly rather than at the next quarter. Those quarterly dates, and the quarterly options expiry that coincides with them, concentrate obliged fund trading into single sessions and are dates to exclude from any volume reading rather than interpret.
Why does the index sometimes rise while most stocks fall?
Because it is capitalisation-weighted: influence follows tradable market value, so a handful of the largest constituents can carry the number while the majority of the list declines. The index cannot show you that, which is the entire reason the breadth measures on this site exist. Comparing the index against its own equal-weighted version, or against an advance/decline count, is how the gap becomes visible.
How concentrated is it?
Considerably more than the name implies, and the degree changes over time, which is why no figure is quoted here, a concentration number printed on a static page is out of date within months and then gets cited. The structural point is stable: under capitalisation weighting the largest few constituents carry a share of the index far out of proportion to their number, so "500 companies" describes the count and not the exposure.
How does it differ from the Russell 1000?
Both cover American large caps and they are built in opposite ways. The Russell 1000 is ranks 1 to 1,000 of a mechanical ranking, with no earnings test and no committee. The S&P 500 is a smaller, curated list with a profitability gate at the door and discretion over timing. The two hold different companies and behave similarly most of the time, which makes the differences easy to forget and worth knowing.
Where do I get the current constituents?
From S&P Dow Jones Indices, and in practice from the daily holdings disclosure of any ETF tracking the index. This page does not reproduce a constituent table: membership changes as events require, so a static list is wrong continuously rather than annually, and a current list used on historical data introduces survivorship bias, since every company deleted for failing is missing from it.