The most-asked numbers in index investing — what does the MSCI World return per year? what about the S&P 500 over ten years? thirty? — drive thousands of searches this site tracks, and the answers circulating online are mostly true, misleading, or both at once. The honest version needs three tables and one warning, and this guide provides all four: the long-run averages, the year-by-year reality behind them, the valuation context (the P/E question), and the arithmetic of actually using historical returns without being fooled by them.
Every number below should be checked against the live index factsheets linked in the sources — this guide teaches the shape of the data and the reading method; the indices publish the current figures monthly. Verified framing, August 2026.
The long-run averages: what the record actually says
The headline numbers, in their honest ranges: the MSCI World has delivered roughly 8–9% annualized in USD across the multi-decade record (dividends reinvested — the only honest measure), landing near 7–8% in EUR terms depending on the window’s currency swings. The S&P 500’s long-run number runs higher — roughly 10% nominal annualized across the near-century record, with the recent decade far above it (the American platform era) and famous decades far below (2000–2009’s negative “lost decade”). Inflation-adjusted, both indices’ real returns land in the 5–7% band — the number retirement math should actually use. The critical fine print: these are geometric averages across windows containing 50% crashes and decade-long droughts — the average was earned by holders who sat through years that looked nothing like the average, which is the entire lesson of the next table.
This analysis sits within our all-world ETF comparison hub.
Comparing funds is step one — the Masterclass turns it into a complete plan — fund choice, broker setup, YOUR country’s taxes and the first purchase, in 55 text lessons.
See the full curriculum — €147, 30-day guarantee →The year table: what “average” hides
The distribution behind the averages, from the indices’ published annual records: single years have ranged from roughly +30%+ (the great years: 2019, 2021, 2023-class) to −40%+ (2008), with the “average” 8–10% year being among the rarest outcomes — markets overshoot in both directions and almost never land on their mean. The clustering matters more than the range: strong years follow strong years more often than intuition expects, crashes cluster into regimes (2000–02, 2008–09, 2022), and recoveries have historically been violent (2009: +30%; 2020’s rebound inside the crash year; 2023 after 2022). The practical readings: any specific year forecast is astrology; any decade expectation built on the long-run bands is engineering; and the sequence — not the average — decides outcomes for anyone adding or withdrawing money on a schedule, which is everyone. The tables’ real product is calibration: an investor who knows the distribution stops treating −15% years as emergencies or +25% years as genius.
The P/E question: what valuations add
The other famous search — the S&P’s P/E ratio — deserves its honest frame: the index’s multiple (live on the S&P DJI site) has ranged across the modern era from single digits (1980) to bubble-era extremes, and its predictive content is real but slow — starting valuations correlate meaningfully with decade-scale returns (high P/Es preceded the weak decades, low ones the strong) and barely at all with next-year outcomes. The current era’s persistently elevated multiples are the market pricing the platform economy’s margins and growth — either correctly (the earnings keep arriving) or expensively (the mean-reversion case) — and the honest use of the number is expectation-setting, not timing: elevated starting valuations argue for planning on the bands’ lower half, never for waiting in cash. The world shelf’s diversification and the valuation-tilt instruments are the actionable versions of the concern; the P/E chart itself is context.
Using the numbers: the four honest rules
One — plan on real, not nominal: the 5–7% real band drives retirement math; nominal averages flatter every projection. Two — expect the distribution, not the mean: write the −40% year into the plan before it writes itself in. Three — respect the sequence: savings plans harvest volatility (the calculator guide’s territory); withdrawal plans fear it (the stability layer’s justification). Four — never extrapolate the recent decade: the American platform era’s returns sit far above the long-run bands — planning on their continuation is the era’s most popular arithmetic error, and planning on their reversal is its second. The bands, the distribution, the sequence — that’s the whole toolkit the famous numbers actually support.
Reader scenarios
“So what return do I actually put in my retirement spreadsheet?” 5–6% real for world equity is the defensible planning band — conservative end for elevated-valuation starts; the difference between 5 and 7 compounds enormously, which is why the spreadsheet should show both.
“MSCI World or S&P 500 — the higher historical number wins?” The S&P’s edge is the American era’s residue — the forward question is concentration versus diversification, not whose backtest reads higher; the comparison guides frame it properly.
“Where do I find the actual year-by-year tables?” The index factsheets (linked in sources) publish annual returns; the backtesting tools reconstruct any window in any currency — ten minutes with either replaces every forum table, stale screenshots included.
“Does the 30-year average guarantee anything for my 30 years?” Nothing — it bounds the plausible; your window will write its own line in the table. The plan’s job is surviving every line the record contains, which automated diversified investing — and only that — has historically done.
The guide’s closing instruction is its simplest: bookmark the two factsheets, run the real-return bands in your planning spreadsheet, and let every future “what does the market return” conversation start from the distribution instead of the mean. The numbers themselves will keep moving — new years joining the tables, new eras bending the averages — but the reading method is permanent, and the investor who owns it is inoculated against the two errors that consume everyone else: planning on the recent decade, and panicking in the years that make the averages possible.
“One-sentence summary for the spreadsheet?” Plan on 5–6% real, expect years that look nothing like it, and let automation collect the average the distribution hides.
The tables educate; the automation collects; the decades decide — in that order, always.
Averages inform; distributions train; sequences decide.
Quick-start checklist
- Bookmark the live factsheets; retire the stale screenshots.
- Plan on 5–6% real; stress-test at less.
- Write the −40% year into the plan explicitly.
- Match the sequence tools to the phase: plans harvest, withdrawals defend.
- Reread this page whenever a decade’s returns start feeling normal.
Verdict
The famous averages are true, useful and dangerous in equal measure: 8–10% nominal long-run, 5–7% real, earned through a distribution whose individual years almost never resemble the mean and whose sequence decides every real investor’s outcome. Used as calibration — bands for planning, distributions for expectations, valuations for humility — they’re the foundation of every sound plan on this site; used as promises, they’re the seed of every disappointment. The tables don’t predict; they educate — and the educated holder, automating through whatever years arrive, is the only reliable beneficiary the record has ever produced.
The sequence lesson, worked: two investors, one average
The guide’s most important concept deserves numbers. Two investors each average 7% real across thirty years. Investor A (accumulating: saving monthly throughout) meets the bad decade first — buying cheap for ten years, then compounding the recovery: the sequence flatters them; volatility was their friend. Investor B (withdrawing: retired at the start) meets the same bad decade first — selling assets into depressed prices to fund living costs: the portfolio may never recover despite the identical average, the failure mode every retirement study centers. Same market, same average, opposite outcomes — sequence risk in one paragraph. The toolkit follows directly: accumulators need automation and nothing else (the sequence works for them); withdrawers need the defenses — the stability layer covering years of spending (selling cash, not equities, in the bad stretch), flexible withdrawal rules (trimming drawn amounts in down years), and the glide constructions (the min-vol review’s territory). The famous averages contain both investors’ fates without distinguishing them — which is why this guide keeps insisting the distribution and the sequence, not the mean, are what the tables actually teach.
One more scenario
“Where does the 4% withdrawal rule fit these numbers?” It’s the sequence lesson productized — a rate calibrated to survive the record’s worst sequences, not its averages; the real-return bands above are its inputs, and its own literature (updated for current valuations) is the withdrawal phase’s required reading.
The currency layer: EUR investors and the famous numbers
The tables’ quiet complication for this site’s readers: the indices’ famous averages are USD-natives, and euro investors’ realized returns add the currency’s own history. The mechanics: an unhedged euro holder’s return equals the index’s USD return plus the dollar’s move against the euro — the 2000s’ weak dollar subtracted from strong local returns; the 2010s-20s’ strong dollar added to them; and across the multi-decade record the currency’s contribution roughly washed, at the price of window-to-window noise measured in whole percentage points annually. The planning translations: long-horizon euro accumulators can use the USD bands with a modest haircut for conservatism (the wash assumption, humbly held); defined-horizon euro investors — the five-year crowd — face genuine currency risk the bands don’t capture (the hedged share classes’ actual audience); and the historical comparisons flooding German investing forums (MSCI World ‘Rendite’ tables in euros) legitimately differ from the USD canon — both true, denominated differently. The rule that survives every version: currency noise washes for the patient and matters for the scheduled — one more variable the automation absorbs and the timing crowd inherits.
Frequently asked questions
What is the MSCI World’s average return over 30 years?
Roughly 8% annualized in USD with dividends reinvested across the modern 30-year windows — the live factsheet’s trailing figures update the precise number monthly; EUR results shift with the currency’s own history.
What did the S&P 500 average over the last 10 years?
The recent decade ran well above the long-run 10% — the platform era’s exceptional stretch; the factsheet’s trailing-10 figure is the current answer, and the guide’s fourth rule is its warning label.
Is there a table of MSCI World returns by year?
The index factsheets and public backtesting tools publish the full annual series — every year since inception, any currency; the sources above link both.
What is a good P/E for the S&P 500?
The historical median sits well below recent readings — but “good” depends on rates and margin structure; the honest use is decade-expectation calibration, never entry timing.
Do these averages include dividends?
All figures here are total return — dividends reinvested; price-only numbers understate equity returns by the yield and produce most forum confusion.
What return should a 100% world-equity portfolio expect?
The bands: 7–9% nominal, 5–7% real, across long horizons, with the full distribution en route — and the automation to survive it as the plan’s real engine.
What was the worst 10-year period for the S&P 500?
The 2000s printed a negative nominal decade — the record’s standing warning against decade extrapolation in either direction.
Do the averages differ much in EUR versus USD?
By the currency’s own path — euro investors’ realized returns add the dollar cycle’s swings; long horizons wash much of it, specific windows don’t.
Cited sources: MSCI World index factsheets, S&P DJI index data, and Curvo historical backtesting data. Verify current fund figures before investing. Educational content, not financial advice.
European UCITS/ETF investor with 11+ years building a global dividend and index portfolio, and founder of Financial Expert Class. I write every guide here from first-hand experience investing across EU brokers and tax regimes — not theory. More about me →