Key takeaways
- Sweeping UK weight from 0% to 100% against MSCI World, portfolio volatility stayed inside half a point, between 9.55% and 10.03%, for every mix from 10% to 70% UK.
- Return per unit of volatility fell from 1.26 all-global to 0.72 all-UK, because MSCI World compounded at 13.04% a year in sterling from 2012 to 2025 against 8.13% for the UK.
- Across 32 countries and 50 years of Dimson, Marsh and Staunton data, global investing produced higher Sharpe ratios than domestic in the vast majority of markets; the US was the standout exception.
- Vanguard's April 2021 analysis found volatility was reduced most at 35% to 55% international in every market it examined; this sweep's minimum landed at 60% global instead.
- The return gap flips sign: MSCI World beat the UK by 11.33 points in sterling in 2024, then the UK beat MSCI World by 13.05 points in 2025.
Between 10% and 70% UK, the risk needle barely moved
What's the right split between UK and global equities? It's one of the first questions a UK investor types into a search box, and the confident answers on offer run all the way from 0% to 100%.
Here's what a sweep of the actual data shows. It takes the 14 calendar-year returns of the MSCI United Kingdom and MSCI World indices from 2012 to 2025, both in sterling on MSCI's net total return basis, from MSCI's July 2026 factsheet. It blends them at every UK weight from 0% to 100%, rebalanced once a year. Annualised volatility stays between 9.55% and 10.03% for every mix from 10% to 70% UK. Across that whole span, the risk side of a global equity allocation moved by less than half a percentage point.
Return is a different story. MSCI World compounded at 13.04% a year over the window; the UK index managed 8.13%. That 4.91 point gap, not risk, is what separated the outcomes, and it's the one input nobody hands you in advance. In 2024 the gap ran 11.33 points in the world index's favour. In 2025 it ran 13.05 points the other way.
The sweep: 11 portfolios from all-global to all-UK
The arithmetic is simple enough to reproduce in a spreadsheet. Each portfolio holds the MSCI United Kingdom index at a fixed weight, with the rest in MSCI World, compounding year by year with an annual rebalance. The table reports the geometric annualised return, the standard deviation of the 14 annual returns, and the ratio of the two. That ratio is a rough cousin of the Sharpe ratio without the cash-rate adjustment; it ranks these portfolios the same way.
| UK weight | Annualised return | Volatility | Return per unit of risk |
|---|---|---|---|
| 0% | 13.04% | 10.35% | 1.26 |
| 20% | 12.13% | 9.75% | 1.24 |
| 40% | 11.19% | 9.55% | 1.17 |
| 60% | 10.21% | 9.77% | 1.05 |
| 80% | 9.19% | 10.38% | 0.89 |
| 100% | 8.13% | 11.31% | 0.72 |
The middle column is the shape the chart plots: a wide, shallow valley. Volatility bottoms at 9.55% around 40% UK, and it's 10.01% at 10% UK and 10.03% at 70% UK. The first column has no valley at all, just a slope. Every 20 points of extra UK weight cost roughly a point of annualised return in this window, which is why the ratio in the final column falls from 1.26 to 0.72.
Why risk went flat: the two legs disagreed at the right moments
The flatness isn't luck; it's correlation. Over these 14 years the two return series moved together with a correlation of 0.57, low enough that each leg kept absorbing some of the other's bad years. In 2020 the UK index fell 13.23% in sterling while MSCI World rose 12.32%. Two years later the roles swapped: in 2022 the UK returned 7.15% while the world index dropped 7.83%. A portfolio holding both got a smoother ride than either leg alone. That's why the 40% UK blend undercut the all-global portfolio on volatility, even though the UK leg on its own was the more volatile of the two by this measure, at 11.31% against 10.35%.
MSCI's own risk figures, computed from monthly rather than annual returns, tell the same story from a different angle. Over the 10 years to July 2026 the factsheet puts UK volatility at 11.94% and MSCI World at 11.92%, near enough identical. What differed was the reward for bearing it: a 10-year Sharpe ratio of 0.62 for the UK against 0.90 for the world index.
50 years of data put global ahead on risk-adjusted return, with one giant exception
A 14-year window is a single sample, so it's worth asking what a much longer one says. The Dimson, Marsh and Staunton database behind the UBS Global Investment Returns Yearbook 2025 compares domestic against global investment for the 32 countries in its world index from 1974 to 2024. The authors' summary is blunt: over the last 50 years, global investment led to higher Sharpe ratios than domestic investment in the vast majority of countries.
The exception was the biggest market on earth. One of the few countries where the rule failed was the US, where, in the Yearbook's words, investors "would have been better off remaining in US stocks". The authors call that a cautionary tale about uncertainty, not a strategy. For a UK reader the relevant half of the finding is the first half: the UK sat with the majority, where going global raised risk-adjusted return.
The same source carries a warning about what global now means. At end-2024, US market concentration stood at its highest level for 92 years, and the ten largest companies accounted for around a quarter of global equity value. A fully global equity portfolio today is, to a meaningful degree, a position in one country and a handful of firms. MSCI's July 2026 factsheets make it concrete: the US is 72.03% of MSCI World and the UK 3.61%. Concentration is no refuge at home either, only worse: the ten largest UK stocks were 53.32% of the UK index, HSBC alone 11.26%, against 27.40% for the top ten in the developed world outside the UK. From 2000 to 2024 that world portfolio still delivered a 3.5% annualised real return and a 4.3% equity premium over bills. But the concentration changes what diversification the global index actually buys you, and how far your own portfolio already leans that way is the subject of our guide to home bias and market weights.
The best case for holding more at home than market weight suggests
The counter-argument deserves its numbers too, because there are real ones.
Income and cost sit on the home side of the ledger. The UK index yielded 3.05% at July 2026 against 1.53% for MSCI World, and sterling dividends skip the layer of withholding and conversion charges that overseas investing costs add to a global book. Those drags are small in any one year and permanent in all of them.
Currency is the bigger swing factor. A sterling investor holding MSCI World owns a mostly foreign-currency asset, and in 2022 that mattered enormously: the index lost 17.73% in dollars but only 7.83% in sterling, because the pound weakened at the right moment. The two figures sit on slightly different dividend bases, which changes tenths of a point, not the story. The cushion works in reverse when sterling strengthens, which is the heart of the currency hedging decision that sits alongside this one.
Then there's the crash objection. Critics of international diversification observe that it fails precisely when it's wanted, because markets become more correlated during downturns. Asness, Israelov and Liew, in the paper AQR publishes on exactly this charge, accept the short-run observation and reject the long-run conclusion: over longer horizons, economic performance drives returns rather than short-lived panics, and the diversification does its work. The 2020 and 2022 rows above are small-scale examples of that mechanism operating.
And the US result cuts both ways. Staying home worked for exactly one country in the 50-year sample, and it happened to be the largest, most exceptional market of the period. Treating that as evidence for a heavy UK weight amounts to a bet that the UK is the next US; the 1974 to 2024 record puts the base rate against it.
Why the optimum moves, and why global equity allocation is a range, not a point
Vanguard's April 2021 paper ran this style of analysis for 5 developed markets and found volatility was reduced most with an international allocation of between 35% and 55%, wherever the investor lived. This sweep's minimum landed at 60% global, or 40% UK. Both are right about their own samples, and that's the point: when a curve is this shallow, the exact location of its minimum is mostly noise. When the window moves a few years, the bottom slides; the width of the valley is what survives. Vanguard's projected 30-year correlation between UK and international equities was 0.65 as of September 2020, against the 0.57 realised in this sample, and small differences like that are enough to move the minimum around.
Set that flat range against what UK investors actually hold and the gap is striking. Vanguard's paper put UK investors at roughly 4.9 times their market's weight, at a time when the UK was 3.5% of global market capitalisation. The index covering everything listed outside the UK, MSCI ACWI ex UK, held 2,569 stocks at end-2024, roughly 85% of the global opportunity set beyond this island. Nothing in the sweep says market weight is the answer. It says something narrower: the risk-based case for any particular UK equity weight between about 10% and 70% is weak, because those portfolios behaved almost identically. What's left to choose on is return expectations, cost, currency and temperament, which is why a global equity allocation behaves as a range rather than a number.
The global market itself refuses to hold still, which is another reason point estimates age badly. The US was 58.3% of world equity value in September 2020 and as low as 29% in the 1980s. Back in 1900, railroads were 63% of the US market and almost 50% of the UK's. Any fixed split back-tested a generation ago was tested against a market that no longer exists.
What 14 annual observations cannot tell you
The sweep is honest arithmetic on real index returns, and it's still a small sample: 14 annual observations from one window, in one currency, spanning a period in which the US ran one of the strongest stretches in its history. A different 14 years reorders the return column completely. The 1974 to 2024 evidence is the better guide to the long run, and even that is one path through history, not a forecast.
The instruments are simplified too. Both indices cover large and mid caps only, the returns are an index's rather than any fund's, with no fees in them, and MSCI World itself still holds UK stocks, 3.61% of it in July 2026, so the 0% UK portfolio isn't quite 0%. The return-to-volatility ratio ignores the cash rate a true Sharpe ratio subtracts; on the factsheet's cash-adjusted 10-year Sharpe ratios the ordering, 0.90 world against 0.62 UK, comes out the same. Annual data also hides everything inside a year, including the drawdowns that persuade people to abandon an allocation before the averages can rescue it.
Platform fees, FX conversion spreads and an investor's own tax position sit outside the index maths entirely, and for a small portfolio they can outweigh most of what's in the table.
What would change the conclusion
If the correlation between the UK and the world index climbed from 0.57 toward 1, the valley would flatten into a line and risk would stop distinguishing any mix at all; only costs and currency would remain. A decade of divergent, deglobalising markets pushing that correlation lower would do the opposite and deepen the case for holding both.
If the return gap keeps changing sign, as it did between 2024 and 2025, the return column compresses and the table converges; a run of years like 2025, when the UK beat the world index by 13.05 points, would rewrite the ranking with the home leg on top. And if the quarter of global value now sitting in the ten largest companies unwinds, global itself becomes a broader, different bet than the one measured here.
The numbers worth watching aren't the weights. They're the correlation, the return gap and the concentration of the index being diversified into. LedgerTouch shows the UK weight your holdings actually sum to across accounts; the sweep above runs the same way on whatever that number turns out to be.