Key takeaways
- Walmart is 0.736598% of the SPDR S&P 500 ETF Trust against Microsoft's 5.800066%, so it reads as 12.7% of Microsoft. On full market capitalisation it is 22.7%.
- Walton Enterprises held 44.11% of Walmart's shares in April 2026. The index counts the other 55.89%, which rounds to an investable weight factor of 0.56.
- Deutsche Telekom's 54.5% stake leaves T-Mobile US with an implied free float of 45.0%, the smallest of the five large constituents traced here.
- MSCI rounds free float up to the nearest 5% once it clears 15%, so a security with 57.0% of its shares available gets a factor of 0.60.
- Those five companies hold 3.12% of the fund while their full market caps come to 4.38% of the same denominator, a gap of 1.27 percentage points.
Free float adjustment counts the shares you can buy, not the shares that exist
You look up a company's market value, you look up its weight in the index that holds it, and the two don't agree. Walmart shows it plainly.
On 8 October 2026 the SPDR S&P 500 ETF Trust, which replicates the index share for share, gave Walmart a weight of 0.736598% and Microsoft 5.800066%. So Walmart reads as 12.7% of Microsoft. On full market capitalisation it's 22.7%. Nothing about the two businesses explains that gap. Free float adjustment does.
S&P Dow Jones Indices doesn't weight the S&P 500 by market capitalisation, and its own methodology says so. Each stock enters the index at its price times its total shares outstanding times a third term, and that third term is the free float adjustment factor. S&P defines it in one line: "IWF = (available float shares)/(total shares outstanding)", "where available float shares are defined as total shares outstanding less shares held by strategic holders." For Walmart the investable weight factor is 0.56.
MSCI puts the same point in one sentence: "The free float-adjusted market capitalization is used to calculate the weights of the securities in the indexes." Not the market capitalisation. The free float adjusted market cap, which is a smaller number for any company a family, a government or another listed company has a grip on.
What free float adjustment does to Walmart in the S&P 500
Walmart's 2026 proxy statement does most of the work. It reports 7,970,990,515 shares outstanding and names one holder above 5%: Walton Enterprises, LLC, with 3,516,197,849 shares, or 44.11% of the class.
S&P's rules put that block outside the float. Its April 2024 float adjustment methodology lists "Foundations or Family Trusts associated with the Company" and "Any individual person listed as a 5% or greater stake holder in a company as reported in regulatory filings" among the strategic shareholders whose shares come out. Subtract 44.11% and 55.89% is left. S&P reports factors "to the nearest one percentage point", so the number the index uses is 0.56. A rulebook this specific is the asset an index provider licenses, which is why the index licensing fees a tracker pays buy a methodology rather than a list of names.
None of that needs S&P to publish anything, because a fund tracking the index has to hold the adjusted share counts. Two lines from the same holdings file make the point.
| SPDR S&P 500 ETF Trust, 8 October 2026 | Walmart | Microsoft |
|---|---|---|
| Shares held by the fund | 54,414,004 | 90,642,802 |
| Shares outstanding, company filing | 7,958,079,155 | 7,425,545,491 |
| Fund's holding as a share of the company | 0.6838% | 1.2207% |
| Implied free float adjustment factor | 0.56 | 1.00 |
| Index weight | 0.736598% | 5.800066% |
The fund owns 1.2207% of every Microsoft share in issue and only 0.6838% of every Walmart share. Divide the second by the first and you get 0.5601. That's the Walton stake again, arriving by a different route.
Walmart's own annual report says something close in dollars. It puts the aggregate market value of its stock "held by non-affiliates" at $391.7bn in July 2025, and explains that for this purpose it treats its directors, executive officers and 5% holders as the affiliates. A company reporting its own float lands in the same place the index does.
Three index providers, three definitions of free float
The principle is shared. The arithmetic isn't, and a reader comparing a US tracker with a global one is comparing two different sets of rules.
| S&P Dow Jones Indices | MSCI | FTSE Russell | |
|---|---|---|---|
| What the factor is called | Investable weight factor | Foreign inclusion factor | Investability weight |
| What counts as strategic | Officers and directors as a group, plus holders at 5% or more | Any disclosed strategic holding, "regardless of the size of the shareholding" | 0% for individuals, corporations, foundations, treasury and private equity; 10% for sovereign wealth funds; 30% for a single portfolio holding |
| Precision | Nearest 1 percentage point, and anything at 0.96 or above becomes 1.00 | Rounded up to the closest 5% above 15% free float, to the closest 1% below it | Four decimal places |
| Eligibility floor | Not set in the float adjustment methodology | A factor of 0.15 | 10% free float for the FTSE UK Index Series |
| When factors move | Annual review, the Monday after the third Friday of September | At index reviews | At the quarterly review |
Those differences bite. MSCI's rule is that "for securities with free float greater than 15%, the FIF is equal to the estimated free float, rounded up to the closest 5%". Its own worked example takes a security with 57.0% free float and assigns it a factor of 0.60. The index then tells a tracker to buy 3 percentage points more of that company than MSCI's own estimate says is for sale.
FTSE Russell goes the other way and keeps every digit: "Free float is calculated using publicly available information and is rounded to four decimal places and reported as a percentage. For example, a free float of 4.99999% is rounded to 5.0000%."
The same shareholder, three different answers
Take a sovereign wealth fund whose stake sits between the two thresholds, above S&P's 5% line and below FTSE Russell's. FTSE Russell restricts sovereign wealth funds only at 10% or more, so those shares stay in the float. S&P lists "Sovereign Wealth Funds" among its strategic holders, so the same stake comes out.
Now reverse it. A pension fund above FTSE Russell's 30% line is restricted there, because a single portfolio holding that large "will be regarded as strategic". S&P puts "Pension Funds (including Government Pension and Retirement Funds)" firmly in the public float. One register, two weights.
Individual holders split the same way. FTSE Russell restricts "All individual holders disclosed within corporate filings", regardless of amount, and says so explicitly for disclosed officer and director holdings. S&P reaches individuals only when they are "listed as a 5% or greater stake holder", though officers and directors do come out as a group once another strategic block clears that line.
MSCI's factor also prices in what foreign investors are allowed to own
MSCI's version does a second job, which is why it's a foreign inclusion factor rather than a free float factor. Where no statutory limit applies, the factor is just the estimated free float. Where one does, MSCI takes the "Lower of" the free float and the foreign ownership limit less any strategic stakes already held from abroad.
It also tracks how much of a permitted quota is already used, which it calls foreign room, and its own example is arithmetic anyone can follow: a limit of 40%, foreign holdings of 20%, so foreign room of 50%. MSCI monitors foreign room quarterly in 10 emerging markets, Indonesia among them, and says securities with low foreign room "may be in some instances not eligible for index inclusion or subject to a weight reduction".
This is the part of free float adjustment a UK investor in a global tracker meets without noticing. A company can be large, liquid and listed, and still enter at a fraction of its size because its home market caps what an overseas fund may hold. Below a factor of 0.15, MSCI says a security is "in general" not eligible for the investable universe at all.
Four more S&P 500 companies where free float adjustment moves the weight
The same test runs on any constituent with a block on its register. Each disclosed stake below comes from the company's own proxy statement. Each implied factor is the fund's holding divided by shares outstanding, scaled against the constituents the index treats as fully floated.
| Company | Disclosed strategic holder | Stake | Free float left | Implied factor |
|---|---|---|---|---|
| T-Mobile US | Deutsche Telekom AG, 592,065,875 shares | 54.5% | 45.5% | 0.45 |
| Walmart | Walton Enterprises, LLC, 3,516,197,849 shares | 44.11% | 55.89% | 0.56 |
| Oracle | Officers and directors as a group, 1,166,800,963 shares | 40.9% | 59.1% | 0.60 |
| Eli Lilly | Lilly Endowment at 9.8% and PNC Financial Services at 5.4% | 15.2% | 84.7% | 0.85 |
| Coca-Cola | Berkshire Hathaway, 400,000,000 shares | 9.29% | 89.8% | 0.90 |
The scaling needs a check of its own. For a company with no strategic block at all, the fund's holding divided by shares outstanding should be the same number across every name, because the only thing left varying is the index divisor. Apple, Microsoft, JPMorgan Chase, Procter & Gamble, Costco and Home Depot all land between 0.9997 and 1.0004 on that test. The chart above plots the five restricted names against Microsoft at 1.00.
Coca-Cola is the tidy one. Berkshire Hathaway is itself a listed company, and S&P excludes "Shares held by another Publicly Traded Company", so Berkshire's 400,000,000 shares leave the count. Coca-Cola's directors and officers added 38,886,151 shares in March 2026, about 0.90% of the 4,302,549,243 shares the company reported outstanding later that year. Together those two blocks come close to the 10% the implied factor says is missing.
Eli Lilly shows the judgement in the system. The Lilly Endowment's 9.8% is a foundation associated with the company, which S&P treats as strategic without argument. The PNC Financial Services Group's 5.4% is a bank, and banks are usually float, but it clears the 5% line and the implied factor of 0.85 only closes if those shares come out too. Where a provider lands on a holder like that is a decision, not a calculation.
The adjustment cuts five weights and lifts every other one
Index weights sum to 100%, so anything taken off one line is spread across the rest. Those five companies carry 3.12% of the fund between them. Measured against the same float-adjusted denominator, their full market caps come to 4.38%. The 1.27 percentage points in between don't disappear.
They go to the lines whose registers are clean, which means an index holder owns slightly more Microsoft and Apple than market value alone would hand them. That compounds the concentration already visible in the MSCI World top 10 holdings, and it is one of several rules that fix a weight before any judgement about value enters. The caps that limit sector concentration in some index families are another.
The case against free float adjustment
The strongest objection is that none of this is measured. It's estimated from disclosure. MSCI says its "estimation of free float is based solely on publicly available shareholder information". FTSE Russell concedes that a float sometimes "cannot be accurately determined due to unclear or insufficient information within the standard public filings". A 5% reporting line means a register can carry several strategic stakes just under it and still read as fully floated.
The second objection is timing. S&P reviews factors once a year and takes its data from "after the market close five weeks prior to the third Friday in March, June, September, and December". Oracle shows what that costs. Its implied factor of 0.60 matches the 40.9% held by officers and directors in its September 2025 proxy, not the 38.4% in the September 2026 one, which was filed after the review had already run.
The third objection is conceptual, and critics of index investing press it hard. Free float adjustment deliberately concentrates tracker demand on the part of a register that is already changing hands. It makes an index a better description of what you can buy and a worse description of the economy. A company half owned by its founders is still a whole company.
Against that sits the obvious defence. Without the adjustment an index would instruct every tracker to buy shares that aren't for sale, and the instruction would be unfillable at exactly the top of the register. FTSE Russell is explicit that replicability is the point: adjustment "ensures an accurate representation of the available investable opportunity set is included within the index calculations and ensures replicability".
What this analysis cannot tell you
Start with the limitations of the method. No provider publishes its factors free of charge, so every implied factor here is reverse-engineered from one fund's holdings on one day. The factors themselves only move at a review, so the arithmetic reproduces on any day's file, but the sample is five companies out of roughly 500, picked because their registers are concentrated.
The reverse-engineering breaks in two specific ways. A company that split its shares between the index's reference date and the holdings date reads as having a factor above 1.00, which is impossible. A company with more than one listed share class has each class adjusted separately, while a single filed share count covers all of them, so the division means nothing. Alphabet, Meta Platforms and Berkshire Hathaway are all left out here for that reason.
This is also one provider, one index and one country. Disclosure thresholds differ, and so do the rules written around them: S&P notes that in Japan "individual shareholders and treasury shares with strategic interests of less than 5% are also considered in the calculation". And none of this says whether a float-adjusted index earns more than an unadjusted one. That is a different question, closer to the ground covered by tracking difference vs tracking error.
What would change the conclusion
If the Walton family sold down below the strategic threshold, Walmart's factor would go from 0.56 to 1.00 and its index weight would rise about 78.5%, with nothing happening to the business. That is the measure of how much of a weight is a disclosure artefact rather than a valuation.
If reporting thresholds fell, measured floats would fall with them, because more strategic stakes would become visible. A move to a 1% line would cut the float of family-controlled companies again without a single share changing hands.
And if the three providers converged on one definition, the gap between a global tracker's weights and a US tracker's would narrow for exactly the companies where it is widest now: the ones with a sovereign wealth fund, a founder or a listed parent on the register. The date to watch is the September review, the one moment in the year when S&P's factors move on new ownership data rather than on a corporate action.