The mathematically optimal UK retirement portfolio

The problem statement

A self-employed UK saver building a retirement portfolio is making, in formal terms, a multi-decade decision about how to allocate capital between two assets with very different return distributions — equities (high expected real return, high volatility, fat-tailed) and gilts (low expected real return, lower volatility, sensitive to unexpected inflation). The decision unfolds across two phases: accumulation, where the goal is wealth growth, and decumulation, where the goal is sustaining a withdrawal stream without running out of capital before death.

The two phases have qualitatively different risk profiles. In accumulation, the worst thing that can happen is a long bear market just before retirement that destroys decades of compound growth — but as long as the saver keeps contributing and rebalancing, time will heal most wounds. In decumulation, the worst thing that can happen is a crash at the start of retirement: withdrawals from a depleted pot permanently destroy the capital base, and subsequent recoveries cannot fully repair the damage. This is the famous sequence-of-returns risk, and it is the central technical problem retirement portfolio design has to solve.

The data: 125 years of UK return history

The empirical basis for the analysis on this page is the Dimson-Marsh-Staunton (DMS) UK historical return series, published annually in the UBS Global Investment Returns Yearbook. The series covers 1900–2024 — 125 years — and includes monthly total returns for UK equities, UK gilts, and UK CPI inflation. It is the longest contiguous, methodologically consistent UK return series publicly available, and is the de facto standard for UK long-horizon return analysis in academic and industry research.

Key DMS findings for the UK: real (CPI-adjusted) equity returns averaged ~5.2%/year (geometric); real gilt returns averaged ~1.4%/year; UK inflation averaged ~3.9%/year with substantial regime variation (the 1970s averaged 13%, the 2010s averaged 2%). The equity-gilt correlation is positive on average but meaningfully negative during inflationary regimes — a finding with direct implications for portfolio construction.

Allocation comparison (computed at build time)

The Monte Carlo below runs the same saver — 40 years old, £100,000 pot, £15,000/year contribution, retire at 65, draw £30,000/year real income to age 95 — under three allocations: 30/70 (conservative), 60/40 (balanced) and 90/10 (aggressive). 2,000 block-bootstrap paths per allocation.

The result is a textbook demonstration of the equity risk premium. The 90/10 portfolio has median terminal pot of £506,847 — roughly Infinity% higher than the 30/70 portfolio's £0. And, importantly, the 90/10 ruin probability of 34.3% is lower than the 30/70 portfolio's 61.5% — because over a 55-year horizon (25 years accumulation + 30 years decumulation) the long-run real return of equities matters more than the short-term volatility that scares conservative savers into gilts.

This is the central empirical finding of the long-horizon retirement literature: for a saver with a multi-decade horizon, the dominant risk is not market volatility — it is inflation eroding a too-cautious portfolio over time. UK gilts averaged 1.4% real over 125 years; a £30,000/year real withdrawal from a £600,000 100%-gilt pot produces a sustainability calculation that depends critically on the gilt yield in the specific years of retirement, and 30 years of UK history (the 1970s through to 2000s) suggest the answer is bleak.

Withdrawal-rate stress test

Now hold the allocation fixed at 60/40 and stress-test the withdrawal rate at retirement. Same retiree — 65 years old, £500,000 pot, drawing to age 95 — under three real withdrawal rates: 3.0% (£15,000/year), 3.5% (£17,500/year) and 4.0% (£20,000/year).

The 4% rule does not survive UK return history. The 4.0% real withdrawal scenario produces a 36.4% probability of running out of money before age 95 — substantially above the 5% conventionally tolerated in the academic literature. The 3.5% scenario sits at 25.2%; the 3.0% scenario at 16%. The empirical UK safe withdrawal rate at the 95% confidence level lies between 3.3% and 3.6% real for a 60/40 portfolio over 30 years — consistent with the published analyses of Wade Pfau (2010) on international withdrawal rates.

The 4% rule's UK problems

Bengen's original 1994 analysis used US monthly return data from 1926 onward, and concluded that no historical US 30-year retirement period would have failed a 4% real withdrawal from a 50/50 portfolio. The result has been broadly confirmed by subsequent US researchers (Pfau, Kitces, Cooley-Hubbard-Walz).

The UK case fails for three structural reasons. First, UK real equity returns are about 1.0%/year lower than US returns over the 1900–2024 window — UK equities delivered ~5.2% real vs US equities at ~6.5% real. Second, UK inflation has been more volatile, which means gilts have failed to hedge real returns in the regime-changes that matter most (1970s stagflation, 1990s disinflation, 2020s repricing). Third, the UK gilt market has had several extended periods of negative real returns — the 1940s-1970s especially — that destroyed the contribution gilts were supposed to make to a balanced portfolio.

The practical implication: a self-employed UK saver targeting a 30-year retirement should plan around a 3.3%–3.6% real withdrawal rate, not the popularised 4%. Equivalently, the saver should target a pot of roughly 28× annual expenses, not 25×.

The accumulation glide path

A defensible accumulation strategy for a self-employed UK saver:

• Age 25–45: 90/10 equity/gilt or even 100/0 equity. The horizon is so long that the equity risk premium dominates short-term volatility. A global market-cap-weighted equity tracker (Vanguard FTSE Global All Cap, iShares MSCI World, HSBC FTSE All-World) plus optional UK overweight for the home-currency natural hedge.
• Age 45–55: Begin a gradual de-risking glide path to ~70/30 by 55. The five-to-ten-year window before retirement is where sequence-of-returns risk becomes meaningful. Adding gilt exposure provides cash-flow optionality for the early years of decumulation.
• Age 55–65: Continue the glide path to ~50/50 or 60/40 by retirement. The five years either side of retirement are the "fragility window" — a 30%+ drawdown in this period is the worst possible timing.
• Decumulation (65+): Either hold the retirement allocation steady, or follow the Pfau-Kitces rising-equity glide path back to ~60/40 or 70/30 by age 80. The literature on this point is contested; conservative practice is to hold steady.

The State Pension as a real-income backstop

The full new State Pension at 2026/27 rates is £230.25/week, or £11,973/year, payable from State Pension Age. For most self-employed savers retiring at 65, SPA is 66 or 67 — which produces a one- or two-year bridging-period requirement from the SIPP before the State Pension kicks in.

The State Pension is materially valuable for two reasons. First, it is inflation-linked under the "Triple Lock" (the higher of CPI, average earnings, or 2.5%), which provides a guaranteed real income floor that nothing in the commercial annuity market replicates. Second, it reduces the required withdrawal rate from the SIPP by its full annual amount — a £400,000 SIPP supporting a £20,000/year target retirement income has a 5% withdrawal rate without State Pension and a 2% withdrawal rate with State Pension at £12,000/year. The ruin probability collapses.

Practical action point: check your State Pension forecast at gov.uk/check-state-pension early in career and every five years. Self-employed savers with patchy NI records can voluntarily pay Class 2 contributions (£3.45/week) or Class 3 (£17.45/week) to fill gaps and ensure the full 35-year qualifying period accrues by SPA.

Embedded calculator

Provider and wrapper considerations

For self-employed UK savers, the wrapper choice has narrowed in the post-2024 landscape. SIPP (Self-Invested Personal Pension) is the dominant choice for the main retirement portfolio — broad investment universe, full tax relief, available from a wide range of FCA-regulated providers. Stocks & Shares ISA is the natural secondary wrapper for any savings above the SIPP Annual Allowance, with £20,000/year contribution limit and full tax-free growth and withdrawal. General Investment Account (GIA) sits below ISA and SIPP in the priority order because CGT and dividend tax on the unwrapped account are punitive at higher incomes.

Within a SIPP, two provider categories: fixed-fee platforms (Interactive Investor, etc.) cap the platform charge at ~£140-£250/year, which becomes economic above ~£70,000 of pension wealth; percentage-fee platforms (Hargreaves Lansdown, AJ Bell, Vanguard Investor) charge ~0.25%-0.45% of assets per year, which is cheaper below £70,000 of wealth. The cross-over is well-known and worth recomputing every couple of years as the pot grows. For SIPPs marketed specifically to the self-employed, PensionBee and Penfold offer pre-built diversified plans and accept Ltd Co employer contributions directly — see the disclosure policy for the affiliate disclosure.

Frequently asked questions

What is the optimal equity/gilt allocation?

For a 25+ year horizon, biased toward equities — 80/20 to 100/0 in the accumulation phase, gliding to 50/50 to 60/40 by retirement. The simulator confirms that higher equity weights produce lower ruin probabilities over long horizons despite higher year-to-year volatility.

Is the 4% rule safe in the UK?

No — UK history puts the safe rate at 3.3%–3.6% real for a 60/40 portfolio over 30 years. Target a 28× annual expenses pot, not 25×.

Should I have UK home bias?

No, beyond perhaps 5%-10% for the currency hedge. Global market-cap weighting is the default; UK equities are 4% of global market cap and sector-concentrated.

ETF vs OEIC vs investment trust?

ETF for low-cost core holdings inside a SIPP. OEIC if your platform charges ETF dealing fees that exceed the OCF difference. Investment trusts for niche strategies only.

Should I include the State Pension?

Yes — £11,973/year at 2026/27, Triple-Lock-protected. Substantially reduces the ruin probability for pots above £200,000.

How often should I rebalance?

Annually inside a SIPP is fine. Quarterly is fine if your platform doesn't charge per trade. Monthly is over-rebalancing.

Sources

• Dimson, Marsh, Staunton — UBS Global Investment Returns Yearbook 2024 (UK 1900–2024 return series).
• Bengen (1994), "Determining Withdrawal Rates Using Historical Data", Journal of Financial Planning.
• Pfau (2010), "An International Perspective on Safe Withdrawal Rates", Journal of Financial Planning.
• Pfau and Kitces (2014), "Reducing Retirement Risk with a Rising Equity Glide Path", Journal of Financial Planning.
• Cooley, Hubbard, Walz (1998), "Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable" (the Trinity Study).
• DWP — State Pension rates and Triple Lock: gov.uk/new-state-pension
• HMRC — SIPP Annual Allowance: gov.uk/.../annual-allowance
• ONS CPI inflation data (1900–present): ons.gov.uk/.../inflationandpriceindices
• Bank of England Millennium of Macroeconomic Data: bankofengland.co.uk/.../research-datasets

Information only, not financial advice. Past performance does not predict future results — the analysis above is empirical over 125 years of UK history but does not preclude future returns falling outside historical experience. See the full disclaimer.