The same money, put in different places, over a hundred years.
Pick any of the six asset classes below. Watch a single sum travel through each — the real path, with its real crashes, drawn only across the years each record can honestly cover. Not a ranking, not a forecast.
1Choose assets to compare
Each is a broad asset-class index — the whole market, not any single stock, house, or fund.
The six lines' years and price paths come from historical records — every crash sits in its true year — but their return bases differ: U.S. equities and U.S. housing are source-data total returns; Canadian equities are an estimated total return (price + fixed dividend yield); Canadian housing is a model total return (prices + assumed net rent, ~4.5% mid-century → ~1.6% today); China A-shares and gold are price-only, with no income at all. Each line begins at the first year its data can be trusted, so start points differ — the multiples are not a strict asset ranking.
2Real or nominal?
Over a century, inflation dominates. Inflation-adjusted ("real") is the default — it answers "how much more could this money actually buy?" The nominal toggle re-inflates each series by a single long-run average inflation rate per country, not year-by-year CPI — nominal values in distant decades are indicative only.
3How should the lines start?
"Each its own history" shows every asset across its full record — but because they begin in different years, a line already high in, say, 1900 had simply been running longer, not "winning." "Same starting line" answers the fairer question: if you invested the same sum in each on the same day, where would they stand today? It restarts every selected line at ×1 from their latest common year — so the timeline shortens to whichever line is youngest.
One sum, across the long run
real · inflation stripped out · log scale
U.S. equities
Canadian equities
Gold
U.S. equities×26.9kover 154 yrs · 6.8%/yr
Canadian equities×9.9over 37 yrs · 6.4%/yr
Gold×4.7over 125 yrs · 1.2%/yr
Each line is a sum starting at 1.0; years and price paths come from historical records (crashes sit in their true years), but return bases differ by line (total return / estimated / model / price-only — see notes below). The dashed line marks a model series (Canadian housing, with an assumed net rent of ~4.5%→1.6%). The nominal toggle re-inflates by a single long-run average inflation rate per country, not year-by-year CPI. Each database is used as of its latest public release; later years are extended by the author from public market data. The y-axis is logarithmic — without it, hundred-fold gaps flatten the lower lines. Each line is drawn only across its data span.
What these lines are saying
The multiples on the chart change with your current basis (real/nominal) and starting-line settings, and the lines use different return bases — what follows describes each line's origin and shape only, not a ranking of assets. U.S. equities: annual historical total return from 1871 (Shiller data, dividends reinvested). You can see 1929–32 halve the line, the grind of the 1970s, the dot-com and 2008 craters — and long-horizon compounding accumulating between the holes. Gold: a price-only series from 1900, earning no rent or dividend. Violent swings — the 1980 spike, then a deep drawdown lasting nearly two decades — around a line with limited long-run rise; on an inflation-adjusted basis it behaves more like stored purchasing power than compounding capital. Canadian equities: an estimated total return from 1988 — the historical price record plus dividends folded in at a long-run average rate (not actual year-by-year dividends). You can see 2008 give back about a third, and the 2015 and 2022 pullbacks. Closest to home for a Canadian investor.
Reading these lines honestly: these are six long-run historical series on different bases, here to show how compounding, inflation, and the choice of starting point change what we see — not a strict ranking of assets. Years and price paths trace historical records, so every crash sits in its true year; but income treatment differs by necessity: U.S. equities and housing include dividends/rents from source data; Canadian equities add dividends at a long-run average rate; Canadian housing is a model series (prices plus assumed net rent, shown dashed); China A-shares exclude dividends; gold earns nothing. Different assets begin in different years, so under "each its own history" a higher line is not "the winner" — it may simply have had more time. Use "same starting line" for a same-year comparison. Past paths are history, not a map of the future.
Curiosity about the long run is where a durable plan usually starts.
Historical data is for illustration only, at the asset-class/index level; the lines use different return bases and are not a strict asset ranking. Series bases: U.S. equities from Robert Shiller's dataset (real total return, dividends reinvested, from 1871); gold from a public annual gold-price series deflated by U.S. CPI (price only, from 1900); China A-shares from published year-end SSE Composite index levels, deflated by China's CPI (real, price index, from 2000); Canadian equities from S&P/TSX annual price returns plus a long-run average dividend yield, deflated by Canadian CPI (from 1988); U.S. housing from the Jordà-Schularick-Taylor Macrohistory Database (housing total return incl. rents, deflated to real, from 1891); Canadian housing is a proxy built from the JST Canadian nominal house-price index plus a time-varying net rental yield (about 4.5% mid-century declining to ~1.6% recently, reflecting the collapse in Vancouver/Toronto rental yields), deflated to real (from 1921). It is for educational purposes only, does not represent future results, and is not personalized investment advice or a promise of returns.
Data attribution. Housing series are derived from the Jordà-Schularick-Taylor Macrohistory Database, used under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License, database consulted July 2026; each database is used as of its latest public release, with later years extended by the author to 2025 from public market data. Rate-of-return data: Òscar Jordà, Katharina Knoll, Dmitry Kuvshinov, Moritz Schularick, and Alan M. Taylor. 2019. "The Rate of Return on Everything, 1870–2015." Quarterly Journal of Economics, 134(3), 1225–1298. Database: Òscar Jordà, Moritz Schularick, and Alan M. Taylor. 2017. "Macrofinancial History and the New Business Cycle Facts." In NBER Macroeconomics Annual 2016, volume 31, ed. Martin Eichenbaum and Jonathan A. Parker. Chicago: University of Chicago Press.