Rules Over Access: What the World's Biggest Fund Does Differently

The world’s largest fund manages 21,268 billion kroner. It holds zero private equity. Zero hedge funds. Its operating cost is 0.04%. And in 2025 it returned 15.1%.

Now look at Yale’s legendary endowment: three quarters of it in alternatives, with access to the global top 1% of managers. In its 2025 financial year it returned 11.1%, and 9.4% a year over the decade.

So the honest scoreboard is not “simple wins, complex loses.” Both work. The interesting question is what each of them costs you to run, and whether you could run it at all.

You cannot buy Yale’s access. It is not for sale to you at any price. But Norway’s edge is not access either. It is a set of rules, written down, and followed when it is unpleasant to follow them.

It is not about what they own. It is about how they govern.

This is the finale. We do not pick a stock. We build the machine.

What the biggest funds actually do

Before building anything, look at what the professionals actually hold.

Four giants, four completely different portfolios

All four are considered successful. They agree on almost nothing about what to own. Each bar is drawn to the same 100% scale.

Norway — Government Pension Fund Global

15.1% 2025 return

The largest single fund on earth. Owns roughly 1.5% of every listed company in the world.

• 71.3% Listed equities
• 26.5% Fixed income
• 1.7% Unlisted real estate
• 0.4% Renewable infrastructure

No private equity. No hedge funds. Almost the entire fund is bought on public markets — and its mandate does not even permit gold.

Fund size: 21,268bn kroner · as at 31 December 2025

CalPERS

36.1% in private markets

The largest public pension fund in the United States, paying real pensions to real people every month.

• 63.9% Public markets (equities + bonds)
• 19.3% Private equity
• 12.6% Real assets
• 4.2% Private debt

Deliberately moving deeper into private markets — the board raised the target from 33% to 40%. The public block is not split into equities and bonds in the monthly disclosure.

Fund size: ~$500bn · private-market weights as at May 2026

Yale — the endowment model

11.1% FY2025 return, net of fees

The portfolio that taught a generation of institutions to buy what the public markets cannot sell them.

• 59% Absolute return, buyouts, venture capital
• 25% Public equities, bonds and cash
• 16% Real estate and natural resources

The mirror image of Norway. Over the decade to June 2025 it returned 9.4% a year — ahead of the endowment median, and it requires access almost nobody has.

Fund size: $44.1bn · target allocation; return for the year to 30 June 2025

GIC — Singapore

3.8% 20-year return, after global inflation

One of the most secretive large investors in the world. It will tell you the shape of the portfolio, but not its size.

• 51% Equities (public and private)
• 26% Fixed income
• 23% Real assets

GIC publishes no annual return and no assets under management — only rolling 20-year figures, precisely so that nobody manages it to a single year.

Fund size: not disclosed · as at 31 March 2025

Norway holds almost nothing private. Yale holds almost nothing else. CalPERS is deliberately moving from the first shape toward the second. GIC will not even tell you what it is worth. There is no consensus allocation here — which is the point. What they share is not a portfolio. It is a process.

Read those four bars again, because the disagreement between them is the whole lesson.

Norway holds 71.3% listed equities and 26.5% fixed income, with 1.7% in unlisted property and 0.4% in renewable infrastructure. That is essentially a global index fund the size of a country. Yale is the mirror image: roughly 59% in absolute return, buyouts and venture capital, another 16% in real estate and natural resources, and only about a quarter in anything you could buy on a Tuesday morning. CalPERS sits between them and is moving deliberately toward Yale, having raised its private-markets target from 33% to 40%, with private holdings reaching 36.1% by May 2026. GIC runs 51% equities, 26% fixed income and 23% real assets, and refuses to publish either its size or its annual return.

Four of the most sophisticated investors on the planet, with the same objective and unlimited research budgets, have arrived at four irreconcilable answers. If there were a correct allocation, these people would have found it.

There is one more thing worth noticing, and it is the least glamorous number on the page. Norway’s 15.1% was 0.28 percentage points below its own benchmark index. The largest fund in the world, with every resource imaginable, slightly lost to the market last year, and reported it plainly in its annual report. It is not trying to be clever. That is the position.

Canada’s CPP Investments makes the same point from the other direction: C$714.4bn, a 9.3% net return in its 2025 financial year and 8.3% a year over ten. Solid, unspectacular, and achieved with a portfolio that looks nothing like Norway’s.

So what do they actually have in common? Three things, and not one of them is an asset class:

1. A written governance framework — the allocation exists on paper before it exists in the market
2. Systematic rebalancing rules — mechanical, not discretionary
3. Ruthless cost discipline — Norway runs on 0.04%, roughly four kroner per ten thousand

The lesson is not to copy their allocation. It is to copy their process, because the process is the only part that is actually for sale.

The one number that travels

Of those three habits, two require a committee. One does not.

Norway runs the largest pool of capital on earth for 0.04% a year. You will not match that, because you are not buying at their scale. But the gap between what they pay and what a European retail investor typically pays is not a rounding error, and it is the single largest controllable variable in the whole exercise:

Who	All-in annual cost
Norway’s GPFG	0.04%
A broad global index ETF	~0.20%
A typical actively managed fund plus platform	~1.8%

You cannot control returns. You cannot control drawdowns. You can control this line exactly, permanently, and starting today, and unlike a market view it is guaranteed to work. Our fee calculator shows what the difference between those rows does to a portfolio over thirty years; the answer is large enough that most people assume it is a mistake.

That is the whole institutional edge, available at retail, for the price of reading a factsheet.

Why the old model is broken

If you know the 60/40 portfolio, 60% stocks and 40% bonds, you know the standard advice. It worked for 20 years. It is weaker now.

The whole point of 60/40 was negative correlation. When stocks fell, long-term bonds rose, so the bonds were a real safety net. That held through the years after the financial crisis, and then it flipped. For much of the 2010s, and again recently, stocks and bonds have often moved together, so the diversification quietly weakened even while 60/40 looked fine on the surface.

You can no longer assume stocks plus bonds gives you automatic protection. You need at least one asset that reliably behaves differently in a crisis.

So what actually diversifies?

Owning many things is not the same as owning uncorrelated things.

What actually diversifies

How closely each one moves with your shares

Gold vs S&P 500 ≈ 0

Genuine crisis diversifier

REITs vs S&P 500 ≈ 0.75

The same risk as your equities

High-yield bonds vs S&P 500 ≈ 0.80

The same risk as your equities

Bitcoin vs Nasdaq > 0.8

A leveraged technology position

Three of those four bars are nearly the same length. A portfolio of shares, REITs and high-yield bonds is not three positions. It is one position, bought three times, with three sets of fees attached.

The uncomfortable part is that these numbers are not constants. Correlations drift in calm markets and converge in violent ones, which means the measurement you rely on is at its least reliable on the day you need it. Plan for the bars being longer than this, not shorter.

Gold against the S&P 500 sits around 0, which is what a genuine crisis diversifier looks like. REITs come in around 0.75. High-yield bonds around 0.80. Bitcoin against the Nasdaq has run above 0.8 since the spot ETFs arrived.

If your “diversification” is stocks plus REITs plus high yield, you own the same risk three times. And Bitcoin is not digital gold — it has a role, but not as a diversifier.

The three-bucket framework

Before any allocation, one discipline. Three buckets:

• Floor — stability, capital preservation.
• Growth Engine — long-term compounding.
• Tactical — asymmetric bets and rebalancing ammunition.

Every asset must belong to exactly one bucket. If a holding does not clearly fit, that is your signal to question whether it belongs at all.

Which portfolio fits you

Four approaches. None is “the best.” Each fits a different situation.

The four shapes you can actually run

Same three buckets, four different weightings. None of these is the best one — each is the best one for a different person.

• Floor — stability and capital preservation
• Growth Engine — long-term compounding
• Tactical — asymmetric bets, rebalancing ammunition

The Lazy Portfolio

25 years and up

Under 30, decades ahead, and no wish to manage anything. Two or three ETFs and a standing order.

10% Floor · 85% Growth · 5% Tactical

Expect a fall of 40%+ at some point in a full cycle

One rule: do not sell in a crash.

The Conservative Builder

under 10 years

A shorter horizon, or an honest admission that a 30% fall would make you sell.

40% Floor · 55% Growth · 5% Tactical

Expect a fall of ~15% at some point in a full cycle

Smallest drawdowns and the best sleep — paid for in return and cash drag.

The Balanced Institutional

10–25 years

The closest an individual gets to the shape Norway or CalPERS runs.

30% Floor · 60% Growth · 10% Tactical

Expect a fall of ~25% at some point in a full cycle

Needs genuine rebalancing discipline to earn its keep.

The Growth Maximizer

25 years and up

A long horizon, stable income from elsewhere, and a demonstrated stomach for pain.

10% Floor · 75% Growth · 15% Tactical

Expect a fall of 35%+ at some point in a full cycle

Highest long-term compounding. Requires iron discipline, not optimism.

Read the red numbers before the green ones. The drawdown is the price of the allocation, and it is the part people agree to in theory and refuse in practice. Pick the row whose fall you could sit through without selling — that is the one that will actually compound.

The Lazy Portfolio. Under 30, decades ahead, no desire to manage anything. Two or three core ETFs, meaning global equity, maybe a bond ETF and a little gold. Monthly savings plan. Don’t touch it. Not optimal, no crash buffer, but it beats most active portfolios and it beats doing nothing. One rule: do not sell in a crash.

The Conservative Builder. Shorter horizon, or you cannot stomach a 30% drawdown. A large Floor of government bonds, gold and cash, a majority in core equity, a sliver tactical. Smallest drawdowns and the best sleep, at the cost of return and some cash drag.

The Balanced Institutional. The closest an individual gets to Norway or CalPERS, for a 10–25 year horizon: roughly 30% Floor, 60% Growth, 10% Tactical. For scale, J.P. Morgan’s 2026 long-term assumptions put a USD 60/40 at 6.4% a year over the next 10–15 years, rising to 6.9% for a version holding 30% in alternatives. Note what that second number implies. The entire reward for adding illiquid, expensive, access-gated assets is half a percentage point.

The Growth Maximizer. 25+ year horizon, stable income elsewhere, stomach for serious pain. Small Floor, heavy Growth, larger Tactical. Highest long-term compounding, and drawdowns of 35%+ are very likely over a full cycle. Requires iron discipline.

Notice what the four bars have in common: the Growth Engine is the majority in every single one. Even the Conservative Builder holds more growth than floor. The difference between the most cautious and the most aggressive row is not really the engine. It is how much ballast sits underneath it, and therefore how far the thing falls before it recovers.

Three questions decide it:

1. Time horizon. Under 10 years → Conservative. 10–25 → Balanced. Over 25 → Growth or Lazy.
2. The sleep test. Maximum drawdown you can hold without selling? 15% → Conservative. 25% → Balanced. 35%+ → Growth.
3. Income stability. Stable income outside the portfolio → more risk affordable. Living off it → bigger Floor.

(We built a free risk self-assessment that walks through exactly this — seven questions, nothing leaves your browser.)

The right portfolio is the one you can stick to. The best allocation on a spreadsheet is worthless if you panic-sell at −30%.

The seven-point audit

Run this on your own portfolio tonight.

1. Cost. All-in below 0.50%, including spreads, tracking difference and tax drag, not just the headline TER.
2. Correlation. A genuine diversifier, or many correlated things? Stocks plus REITs plus high yield at 0.75–0.80 is not diversification.
3. Bucket assignment. Every position fits one. If not, it probably does not belong.
4. Floor test. If equities halve tomorrow, do your Floor assets still cover two to three years of expenses? If not, your Floor is too thin.
5. Rebalancing rules. Written. Threshold-based beats calendar-based, and a 5% deviation is a solid trigger. Not “when I feel like it.”
6. Concentration. No single position above 5% except your core ETF. Bitcoin at 1–2% maximum; at 4%, BlackRock’s own research puts it at roughly 14% of total portfolio risk.
7. An Investment Policy Statement. One page.

The one document that matters most

The Investment Policy Statement is your governance, and your defence against yourself.

One page: goals, time horizon, target allocation, rebalancing trigger, and, above all, what you will not do. Write it literally: “I will not sell during a drawdown of less than X%.”

Since I have now told you twice to write one without ever showing you what one looks like, here is mine, or near enough.

The one page

An Investment Policy Statement, filled in

Not a template to complete. Somebody else's, so you have something to disagree with. The figures are placeholders and yours will be different; the shape is the part worth copying.

Why this money exists

Retirement from roughly 2049, plus a house deposit I may want in the next six years. The house money is not in this portfolio.

Time horizon

Over 20 years for everything in here. Anything I might need within five years lives in cash and is not counted as part of this.

Target allocation

Floor 30%: short-dated government bonds and gold. Growth 60%: one global equity index fund. Tactical 10%: at most 2% in any single position.

What I will pay

All-in under 0.50% a year, counting fund fees, spreads, tracking difference and platform charges. I check this once a year, in January.

When I rebalance

When any bucket drifts more than 5 percentage points from its target. Not on a date, not on a feeling. I check quarterly and act only if the trigger is hit.

What I will not do

I will not sell any part of the Growth bucket during a drawdown of less than 40%. I will not add a new asset class without writing down which bucket it belongs to first. I will not act on anything I read on a day the market has moved more than 3%.

Six headings. It fits on one side of paper and takes about twenty minutes, most of which is spent on the last one. That is the correct distribution of effort: the first five describe a portfolio, and the sixth is the only part that will ever be tested.

The last section is the only one that will ever be tested, which is why it is the longest. Everything above it describes a portfolio, and describing a portfolio is easy. Committing in writing, in calm weather, to a specific thing you will refuse to do in a bad one is the entire exercise.

Why it matters more than any fund choice: the biggest destroyer of returns is behaviour, not products.

How large is it? Large enough to matter, and smaller than the figure usually quoted. Morningstar’s Mind the Gap 2025 is the careful measurement: over the ten years to December 2024, the average dollar in US funds earned 7.0% a year while the funds themselves returned 8.2%, a gap of 1.2 percentage points a year given up to nothing but timing.

Where you put the money changes that gap more than which fund you pick. Allocation funds, the all-in-one kind, showed a gap of just 0.1 points. Sector equity funds: 1.5 points. The structure that gives you the fewest decisions to make costs you the least.

Be sceptical of the bigger numbers. DALBAR’s annual study is the one usually cited, and its results swing violently: it put the equity-investor gap at 8.48 points in 2024 and 0.72 points in 2025. A measure that moves that far in one year is not measuring a stable trait, and researchers have argued in print that the method overstates the gap. It also cannot be recomputed, so I rebuilt it from public SEC filings: across 7,073 US funds the base case for 2020 to 2025 is 0.36 points a year, with a range of −0.06 to +0.62 that hangs on a single share-class assumption (the paper, with data and code). We use the 1.2-point figure here because it is the conservative one, and even the smaller number is still enough to matter.

Read that against the cost table above. 1.2 points a year is roughly a fifth of what a 60/40 portfolio is expected to earn at all. You can save 1.5 points by choosing a cheaper fund and still hand it back by selling in March and buying in September. We built a behaviour gap simulator that runs the two versions of you side by side: same market, same starting capital, one of them flinches.

The IPS is written in calm weather precisely so it can be obeyed in a storm. (It is the same discipline that separates a real exit from hope when it is time to sell.)

What the endowment model actually costs

One final lesson, and it is a humbling one, though it is not the lesson most people take.

Yale beat the median university endowment by an estimated 1.4 points a year over the decade to June 2025. That is real, and it is the number Yale reports.

Now hold it next to a second number. The researcher Richard Ennis has found that large university endowments, as a group, have lagged a simple index portfolio by roughly 1–2 percentage points a year over the same sort of period, while paying enormous fees to do it. Some large endowments spend, in a single year, sums approaching what they collect in undergraduate tuition, on investment management.

Both can be true at once, and that is the uncomfortable part:

Yale is winning a race that the whole field is losing.

Being the best of the endowments is not the same as beating the index, and the benchmark you choose decides which story you get to tell. This is why the audit above starts with cost rather than return. Cost is the one number that cannot be reframed by picking a friendlier comparison.

And David Swensen, the man who invented the endowment model, told individual investors plainly: do not copy me, use low-cost index funds.

Their edge cannot be copied. But their principles can: disciplined allocation, systematic rebalancing, written governance. Those cost nothing.

The whole series in one sentence

Eight asset classes, one framework. And the conclusion is not about what to buy:

The question is no longer what to own. It is whether you have the discipline to follow the process.

Take the risk assessment. Write your one-page IPS. Run the seven-point audit tonight.

The giant funds do not win because they can buy things you can’t. They win because they wrote the rules down, and then followed them. That part is available to anyone.

Educational content only — not investment advice, and not a personal recommendation. Institutional figures are point-in-time and drawn from the funds’ own reports. Speak to a qualified, licensed professional before acting.