What "the market" actually is

Published 19 August 2026 · Updated 20 August 2026

Every time someone says a fund "beat the market" or "matches the market", they've quietly picked a basket. Change the basket, change the answer. This piece is about which baskets people mean, what else moves the answer once you've picked one — the currency, the window, the measure — and why, in aggregate, the average actively managed dollar must lose to the average passive dollar after costs. That last part isn't opinion. It's arithmetic that William Sharpe wrote down in three pages in 1991.¹

First, the thing itself: what is traded, and where

Most explanations of investing skip this part, which is odd, because everything after it is built on top of it. So before we get to what "the market" means in a performance claim, here is what is actually going on underneath.

A share is a slice of a company

A company can be divided into pieces, and those pieces are called shares. Hold one and you own a fraction of that business — not of its building or its coffee machines, but of the enterprise itself. In practice that gives you three things: a claim on the profits (paid out as a dividend when the company chooses to pay one, or reinvested in the business if it doesn't), usually a vote at the annual meeting, and a place in the queue if the company is ever sold or wound up. A long way back in the queue, as it happens: lenders get paid first.

Companies issue shares because they need money and would rather not borrow it. If you sell a tenth of your company to investors, you get cash you never have to repay, and in exchange those investors now own a tenth of everything you make from here on. That is the whole bargain. Everything else in this piece — funds, indexes, benchmarks, the arithmetic further down — is machinery built on top of that one idea.

An exchange is a matching service

Those second-hand trades happen on an exchange. Euronext runs the Amsterdam, Paris and Brussels markets among others; the New York Stock Exchange and Nasdaq run the largest American ones; there is the London Stock Exchange, Deutsche Börse in Frankfurt, and dozens more around the world. A company whose shares can be bought on one of them is listed there.

An exchange does something simpler than its marble lobbies suggest. Buyers say what they are willing to pay. Sellers say what they are willing to accept. The exchange keeps both lists — the highest offer to buy is the bid, the lowest offer to sell is the ask — and a trade happens the moment someone is willing to cross the gap between them: a buyer who accepts the seller's price, or a seller who accepts the buyer's. Then a new price is printed. That gap is called the spread, and crossing it is a small cost you pay for trading right now instead of waiting for someone to come to your price.

It used to be a room with people shouting in it. It isn't any more. The matching runs on computers in fractions of a second, and a share of trading never reaches a public exchange at all: large institutions can deal with each other directly. So "where is the market" has become a slightly awkward question — it is a process rather than a place. But it is a real process, and real ownership really does change hands, all day, in enormous quantity.

Market capitalisation: how "big" a company is, in market terms

If a company has issued a million shares and each one trades at €40, then the market is collectively valuing the whole company at €40 million. That number — price times the number of shares — is its market capitalisation, almost always shortened to market cap. (Those figures are made up, to show the sum.)

It is worth slowing down here, because market cap does more work later in this piece than any other term. It is not the company's profit, or its assets, or its sales. It is simply what the market is prepared to pay for the whole thing today, and it moves every time the share price does. A company can double in market cap without selling a single extra product, if enough buyers change their minds about it.

And even "all the shares" is narrower than it sounds. A good part of many companies sits locked in the hands of founders, families, or governments, and never comes up for sale. Index providers therefore count only the free float — the portion actually available to buy. MSCI's World index, for example, aims to cover "approximately 85% of the free float-adjusted market capitalization in each country" it includes.² So the thing being measured is already a selection: not every company, and not even every share of the companies it does include.

Two neighbours that share the same word

"The market" gets used for more than shares, and the differences matter enough to keep straight:

• Bonds are lending, not owning. Buy one and you have lent money to a company or a government on fixed terms — so much interest, for so many years, and then your money back. You own nothing and vote on nothing; you are a creditor, and creditors get paid before shareholders. Bond markets are enormous, they behave quite differently from share markets, and a claim that quietly mixes the two without saying so is doing something you should notice.
• Derivatives are contracts about prices. Options, futures and their many relatives take their value from something else — a share, an index, an interest rate, a barrel of oil. An option, for instance, is the right to buy or sell at an agreed price before an agreed date, without the obligation to do it. You can hold these without ever owning the underlying thing, which is both the point of them and the reason they can lose money faster than the thing they are derived from. They earn their own explainer and will get one. For now: they are contracts, not ownership, and when someone says "the market" they almost never mean these.

So what is "the market", then?

Put it together and you have something real but unhelpfully large: ownership stakes and loans in tens of thousands of companies and governments, changing hands continuously across dozens of venues, in more currencies than anyone tracks. Nobody holds all of it. No fund is measured against all of it. There is no ticker for it and no price for it.

Which means that the moment somebody claims to have beaten "the market", they cannot possibly mean that whole thing. They mean a selection from it. And which selection they picked turns out to change the answer more than almost anything the fund manager actually does.

The selection has a name: an index

An index is a rule-based list of companies, with a rule for how much of each to hold. The rules are written down, the list is published, and anyone who wants can build a portfolio that copies them. That is what makes an index usable as a yardstick: public, fixed in advance, and followable. But there are many such lists, and they don't agree.

Take two of the most-used definitions of "the world stock market", set out side-by-side in Figure 2:

Both are honest baskets. Both are published by the same index provider (MSCI). Both are what a fund manager might mean by "the world". But the MSCI ACWI holds nearly twice as many stocks in twice as many countries. If a manager brags about "beating world equities" using one of them, the interesting question isn't whether the outperformance is real — it's compared to which world?

Cap-weighted means the biggest listings dominate

Now that market cap is on the table, one more rule matters. Most widely-used indexes are cap-weighted: each company's slice of the basket equals its slice of the combined market cap. A company worth twice as much gets twice the weight — nobody decides that, it falls out of the arithmetic. Which sounds neutral, and has one large consequence for what "the market" actually looks like. Figure 3 makes it visible: it leans hard on whatever happens to be biggest right now.

When you buy a "global" index fund tracking MSCI World, roughly seven out of every ten euros end up in American listings — the factsheet's exact figure is 72.03%. That's not a marketing choice; it's what cap-weighted means when the American market is currently the biggest one. Which is fine, as long as you know it. If you thought you were spreading your risk across "the world" and the American market happens to have a rough decade, you'll find you were not quite as diversified as the label suggested.

Different baskets, different answers — in the same year

Because each index defines the basket differently, they don't agree on how "the market" did in any given year. And the disagreement isn't a fixed offset you could learn to correct for: in two consecutive years the three baskets swapped places, as Figure 4 shows.

All six numbers are correct. All of them describe "the market". They differ because they describe different baskets. That's the entire point: a US large-cap fund that returned 20% in 2024 could report that it "beat the world" — against the 17.49% of MSCI ACWI, that's a plus of about 2.5 points. Against its natural benchmark, the S&P 500 at 25.02%, the same fund fell short by about 5. Same trades, opposite story, depending only on which basket was named.

And notice what 2025 does to that trick. The S&P 500 went from the highest of the three to the lowest — below even MSCI World, which holds no emerging markets at all, so this is not simply a story about emerging markets having a good year. We're not going to tell you why it happened; no source on this page measures causes, and a tidy explanation invented after the fact is exactly the kind of thing this piece is about. What matters is the effect: a fund that chose the flattering comparison in 2024 would have to switch baskets to keep flattering itself in 2025. When you see a fund change the index it reports against, that is worth more attention than the return itself.

This is why a fund is measured against its benchmark — the specific index the fund itself declares it's trying to track or beat. When you read a fund factsheet, the benchmark is the sentence that turns the return into a claim; every other number depends on which sentence that is.

"Beat the market" is arithmetic before it's a claim

Once you have a clean definition of "the market" and a fund that measures itself against it, an old, awkward result kicks in. William Sharpe — the same Sharpe whose name is on the Sharpe ratio, which is roughly a measure of return per unit of risk — wrote it down in 1991 in three pages in the Financial Analysts Journal. His two assertions:¹

The proof is embarrassingly small. Split every dollar in the market into two piles: one held passively (owning every stock in market proportions) and one held actively (everyone else — the pickers, the traders, the hedge funds). The market return is just the value-weighted average of all those dollars. The passive pile, by construction, gets the market return before costs. Since the two piles together are the market, whatever the passive pile got, the active pile must have got too — on average, before costs. That's assertion (1).

Assertion (2) is even shorter. Both piles have costs, but active costs are higher (research, trading, salaries). Equal returns minus higher costs equals lower net returns. That's the whole proof, in one picture in Figure 5.

Three things this doesn't say, and it's important to be strict about them:

• It doesn't say your active manager will lose. It says the average active dollar loses. Some active managers, by definition, will beat the average — and the ones that do so beat it at the expense of the others. The arithmetic is about the pool, not any one fund.
• It doesn't say "active is bad". It says active is expensive. Anything that raises the total cost of participating (higher fees, more trading, more taxes on turnover) subtracts directly from the return, and there is no free lunch in the gross number to make up for it. For any active fund, higher costs subtract more directly from the arithmetic than lower costs do; the direction is fixed.
• And it doesn't promise that every study will agree with it. Sharpe is blunt about what to conclude when one doesn't: "Empirical analyses that appear to refute this principle are guilty of improper measurement."¹ He then lists how the mismeasuring happens, and the list is worth carrying because it tells you what to check in any comparison, including the ones further down this page. The "passive" side may not be truly passive — index funds that sample rather than hold everything, or that charge enough to cost as much as an active fund. The "active" side may be only part of the picture, because most studies count professional managers and leave out private investors, who are active too. And the funds may hold things their index doesn't: his example is equity funds that keep some cash, measured against an all-equity index, where "the funds are generally beaten badly by the index in up markets, but sometimes exceed index performance in down markets".¹ Studies that quietly drop the funds that shut down during the period have the same problem in reverse — survivorship bias, which he says "will tend to produce results that are better than those obtained by the average actively managed dollar". And the one he calls "possibly most important in practice": counting each manager once, when the arithmetic is about dollars and each manager's return should be weighted by the money they run. That last one has a section of its own further down, because it changes the answer. Note which way all of this cuts. It is not a softening of his claim — it's a warning that a study can flatter active management as easily as it can punish it, and that you should ask which one you're looking at.

What that arithmetic looks like when it plays out

Sharpe's result is about averages, so you'd expect real data to be noisy over one year and steadier over ten. That's what shows up in the twice-yearly SPIVA scorecards — S&P Dow Jones Indices' running measurement of how many active funds beat an index, by category and by horizon.⁵ Worth knowing exactly which index, given everything above: SPIVA does not use whichever benchmark each fund declares for itself. It assigns one per category, on the grounds that fund returns "are often compared to popular benchmarks such as the S&P 500, regardless of size or style classification".⁵ So a small-cap fund is measured against a small-cap index whether or not its own marketing prefers a friendlier one. Figure 6 shows the shape of it.

One boundary before the numbers, because it matters and is easy to lose: everything that follows is about shares. The same scorecards measure bond funds too, in their own categories against their own benchmarks, and that picture is not this one. If you take a single thing from this section, take it about equity funds and equity indexes — not about investing in general.

Over a single year, the noise still shows. In 2025 it showed in the unflattering direction: 78.78% of large-cap US funds missed the S&P 500, which S&P Dow Jones Indices calls "worse than the 65% rate observed in 2024 and the fourth-worst year for active large-cap managers over the 25-year history of our SPIVA Scorecards".⁵ But read the yearly series before you conclude anything from that: in 2014 the same measure sat at 87%, in 2007 at 45%.⁵ A single year can say almost anything.

The long windows are where the noise runs out. Over fifteen years, 89.93% of active US large-cap funds fell short of the S&P 500; over twenty years, 92.89%. Zoom out to all US equity funds across every market-cap bucket and 95.01% underperformed the S&P Composite 1500 over twenty years.⁵ Put in the terms that matter to someone choosing a fund: pick one at random and hold it for two decades, and roughly nineteen times in twenty you would have done better with the index it was measured against.

Does any of this survive the trip to Europe?

Fair question, and one worth asking of every number on this page: all of the above is measured on US funds, against US indexes, in dollars. S&P Dow Jones Indices publishes regional editions too, and the European one lets you check whether the arithmetic travels. Figure 7 shows the two euro-denominated categories that matter most to a reader here — Global Equity, which is where a world index fund lives and which the scorecard calls the category with the largest number of available funds, and the pan-European Europe Equity.

Over ten years, 98.44% of euro-denominated Global Equity funds trailed the S&P World, and 97.02% of Europe Equity funds trailed the S&P Europe 350.⁶ Different continent, different currency, different benchmark — and the same shape of answer. One caveat S&P attaches itself, and we'd rather you heard it from us: the euro version of the S&P World index only launched in May 2020, so roughly the first four years of that ten-year comparison run against a back-calculated index rather than one that existed at the time.⁶

What we are not going to do is tell you whether Europe is "worse" than America at this, and the reason is worth more than the answer would have been. The obvious move is to hold one index steady and compare the funds chasing it: both scorecards have a category measured against the S&P 500, so put them side by side. We drafted exactly that, and then went and read how each scorecard defines its category. They don't define it the same way. The American edition's large-cap bucket holds large-cap funds only, and gives mid- and small-cap funds their own benchmarks. The European edition's "U.S. Equity" bucket pools large-cap, mid-cap, flex-cap and small-cap funds together and measures all of them against the S&P 500.⁶

That difference does real work, and it works in one direction. Over the same ten years the S&P 500 returned 14.82% a year, the S&P MidCap 400 10.72% and the S&P SmallCap 600 9.81%.⁵ A mid-cap fund judged against the S&P 500 is starting behind for reasons that have nothing to do with the manager. So some unknown part of any Europe-versus-America gap would be the composition of the buckets rather than the skill in them — which is exactly the pitfall SPIVA describes in its own methodology, and exactly the pitfall this piece is about. We could have published the comparison with a footnote. It reads better without one, and it would have been the same mistake we spend the rest of these pages warning you about.

What Figure 7 does support is narrower and still worth having. Within Europe, on its own terms, the same lesson lands: one year says little, ten years says a lot. Both euro-denominated series climb steadily across the four windows — unlike the American one, which dips twice on its way up. Same destination, different road. It's the horizon that does the work, not any particular window inside it.

The money you count it in

Back at Figure 4 we said the same year gives different answers depending on the basket. It also gives different answers depending on the currency you count in. The European scorecard reports the S&P 500 in euros and in sterling; the American one reports it in dollars. Same index, same windows:

Look at the one-year column first, because it is the one that would have made the better headline: the same index, the same twelve months, and a euro investor saw less than a quarter of what a dollar investor saw. Now look at the rest of the row, because this piece would be a hypocrite if it stopped there. Over five years the euro investor came out ahead — 15.37% against 14.42% — and over ten years it is the sterling investor who leads. Currency moves the answer, sometimes a great deal, but it does not move it consistently in one direction, and a single year tells you as little here as it did anywhere else on this page. We are not going to tell you why any of it happened; none of these sources measures causes.

What survives is the question, not a number. "The S&P 500 returned almost 18% last year" can be true in a headline and not true of your account. If you hold a US tracker in euros and don't hedge the currency, the dollar figure isn't what landed. So: in whose money?

One trap in the other direction, though, because it hides inside the scorecard's own labels. "Europe Equity" appears twice in that table: once for funds priced in euros, once for funds priced in pounds. The euro row gives 81.83% over one year and 97.02% over ten; the sterling row gives 77.78% and 91.84%.⁶ Tempting to read that gap as another currency effect. Don't. These are not the same funds counted in different money — they are different funds, and not by a little: 974 euro-priced funds against 99 sterling-priced ones at the start of the one-year window.⁶ A tenfold difference in sample. Same category name, same benchmark, two populations — the lesson of this whole piece arriving one level down: check what's actually in the basket before you compare two numbers that share a label.

And how you measure it

One honest limitation while we're here: the scorecard has no Netherlands category. We checked the full report, its Morningstar category mapping and its glossary, and there is no Dutch fund category and no Dutch benchmark; the Netherlands is folded into Europe Equity and Eurozone Equity. If someone shows you a Dutch fund and calls it "market-beating", there is no national scorecard to hold it against — which is itself a reason to ask which index the fund is being measured on.

How this changes the questions you ask

Once "the market" is a chosen basket and "beating it" is expensive by construction, a few claims start to fall apart on the first honest question. Not because the fund is dishonest — usually it's more that the sentence is doing more work than the number supports:

• "Beat the market." Which market? Named in a single sentence? Was the fund's declared benchmark that same one, or a friendlier one? Over what window? After fees, or gross?
• "Global fund." Which country weights right now? Track MSCI World and about seven of every ten euros go to US listings; track MSCI ACWI and it's closer to six. Either way the fund's fate rides mostly on how America does. That's a choice; make it knowingly.
• "Consistent alpha." Alpha, which is roughly the extra return a fund delivers above what a simple risk-adjusted benchmark would predict, doesn't survive being redefined. Over which horizon? Against what benchmark? Net of the fund's actual expense ratio, or before?
• "Track record." Twenty years of a single fund is one sample, not twenty. Sharpe's arithmetic says that in the pool the fund fishes from, most rods come home empty over long horizons — and even one impressive rod says less about skill than about the shape of a bell curve.

Sources

1. Sharpe, William F., "The Arithmetic of Active Management," Financial Analysts Journal 47(1), January/February 1991, pp. 7-9. Author-hosted reprint with permission: web.stanford.edu/~wfsharpe/art/active/active.htm. Verbatim quotations in this piece are from that reprint.
2. MSCI Inc., MSCI World Index (USD) — Index Factsheet, JUL 31, 2026. Publisher URL: msci.com/documents/10199/255599/msci-world-index.pdf. Number of constituents, country weights and top-10 concentration used above are taken from this PDF, which we archived to our local sources folder. Its annual return table is on a gross basis and we do not use it: the return figures in this piece all come from the ACWI factsheet, which reports net. See the fine print.
3. MSCI Inc., MSCI ACWI Index (USD) — Index Factsheet, JUL 31, 2026. Publisher URL: msci.com/documents/10199/255599/msci-acwi-net.pdf. Number of constituents, country coverage, US weight and annual net returns used above are taken from this PDF, archived locally.
4. S&P Dow Jones Indices, SPIVA® U.S. Scorecard Year-End 2024. Publisher URL (may be region-blocked to some crawlers): spglobal.com/spdji/en/documents/spiva/spiva-us-year-end-2024.pdf. We retrieved the file via the Internet Archive at web.archive.org (snapshot December 2025) and archived it locally. Used here for the 2024 figures only: the S&P 500's 2024 return (Report 3) and the 2024 small-cap one-year rate (Report 1a).
5. S&P Dow Jones Indices, SPIVA® U.S. Scorecard Year-End 2025. Publisher URL: spglobal.com/spdji/en/documents/spiva/spiva-us-year-end-2025.pdf. This is the current edition and the source of every US horizon figure in Figure 6 and the paragraphs around it. By report: the headline quote and the yearly series from the Summary (p. 1) and Exhibit 10 (p. 11); the underperformance percentages from Report 1a (p. 12); index returns from Report 3 (p. 17); the asset-weighted fund returns in the "counting money" box from Report 4 (p. 19); the S&P 500's return in euros and sterling from the European edition below. ⚠️ The publisher blocks automated access to this file, so the link above may return an error in a browser extension or reader; we hold an archived copy and quote page numbers so any figure can be checked against it.
6. S&P Dow Jones Indices, SPIVA® Europe Scorecard Year-End 2025. Publisher URL: spglobal.com/spdji/en/documents/spiva/spiva-europe-year-end-2025.pdf. The current European edition, with data for periods ending 31 December 2025. By report: Figure 7 and the euro/sterling category percentages from Report 1a (p. 10); the risk-adjusted percentages in the Danish box from Report 1b (p. 11); the fund counts of 974 and 99 from Report 2 (p. 13); index returns including the S&P 500 in euros and sterling from Report 3a (pp. 17-18); the asset-weighted euro figures behind the €10,000 comparison from Report 3b (p. 19); the category mapping that pools US large-, mid-, flex- and small-cap funds (Exhibit 10, p. 36); the Novo Nordisk quotation and the "largest number of available funds" statement from "2025 Highlights" (p. 2). ⚠️ Same access caveat as above; we hold an archived copy.

How we checked. Verified 19 and 20 August 2026 against the six sources above, all opened and read by us; each number in the piece traces to a specific page/line in one of them, and the source list says which report each set of figures comes from. Every source carrying figures was read out twice by two readers working independently and unable to see each other's work — the original three sources for the first version of this piece, and for this revision the two new scorecards, the European one, and a second pass over the MSCI factsheets. Every one of those readings agreed on every value. What changed on 20 August, and why. The version published on the 19th leaned on the SPIVA U.S. Year-End 2024 scorecard, because the 2025 edition — already out by then — was not reachable from our end that evening. We said so on the page rather than pretending 2024 was current, and then went and got it, along with the European scorecard. Rebuilding on those two took several passes through our own review, and the corrections are worth printing, because they're the kind a reader can't otherwise see. The newly written European section at first repeated the very mistake we had just fixed, calling a Year-End 2024 edition "the most recent" when a 2025 one existed. A later draft claimed the risk-adjusted measure moved every other European category against active management; it doesn't — it moves some each way, and the counts now come with the horizon they belong to. Another compared European against American percentages that were measured against different benchmarks, which is the exact error this piece is about. We first tried to repair it by holding one index steady and comparing the funds chasing it — and then found that the two scorecards fill that category differently, so the repair had the same flaw as the original. The comparison is gone; what's left is the explanation of why it can't be made, which is above. What has not changed through any of it: the argument, and none of the conclusions. A few source-internal subtleties are worth flagging. Gross versus net. The two MSCI factsheets are on different bases — the MSCI World factsheet reports gross returns (dividends before withholding tax), the MSCI ACWI factsheet reports net (after). The first version of Figure 4 put one of each side by side, which flattered the gap between them by roughly half a point. Both MSCI figures now come from the ACWI factsheet, so the comparison is net against net; the S&P 500 numbers are index total returns, which carry no foreign withholding for a US-listed basket. Rounding. The "top 10 stocks are roughly a quarter" figure is exact for the MSCI World basket (26.41%) and different for MSCI ACWI (24.21%); we round to "roughly" only in the figure subtitle. Where SPIVA's own summary rounds to whole numbers and its tables don't — 79% against 78.78%, for instance — we quote the summary when we quote and use the table when we count. Three as-of dates in one piece. Every underperformance figure — Figures 6 and 7, American and European alike — is measured over windows ending 31 December 2025, from the current edition of each scorecard. Figure 4 is different by design: it shows two calendar years side by side, so its 2024 bars end 31 December 2024 and the S&P 500's 2024 return comes from the Year-End 2024 edition, as does the 29.69% small-cap figure in the box above. The index snapshots in Figures 2 and 3 (constituent counts, country weights, concentration) are a later photograph again: as MSCI published them on 31 July 2026. That last one is deliberate rather than a mismatch — the point of those two figures is what the baskets look like now, not what they looked like when the fund data closed. SPIVA publishes editions for other regions as well, which we haven't cited here. Note that there is no separate UK scorecard: the British fund categories sit inside the Europe edition, in the same sterling section the 77.78% and 91.84% above come from. Nothing on this page is investment advice; it's an explainer of how "the market" is measured. Think we got something wrong? Tell us at info@proofofreturns.com — we correct in public.