What this means for a holder
Over the five years ending 28 Jul 2026, $100 in bitcoin became $159.63, the same $100 in the S&P 500 with dividends reinvested became $181.41, and in gold $223.88. Those three numbers are what almost every comparison chart on the internet stops at, and on their own they are close to useless: they describe one pair of dates chosen by whoever drew the chart.
The columns that matter for deciding anything are the last three. Bitcoin’s worst fall inside this window was −77%, against −24% for the index. It spent 62% of all trading days more than a fifth below its own previous high; the S&P spent 3%. That difference is not a footnote to the return — it is the return, seen from the inside. Money that had to be withdrawn during one of those stretches never collected the number in the second column.
The rolling table gives the more durable version of the question. Across every start date in our common history, bitcoin finished ahead of the S&P total return index in 71% of one-year windows and 93% of three-year ones, ether in 72% of three-year ones. Both of those are high enough to be worth stating and small-sampled enough to be worth doubting: they rest on roughly 4 independent three-year periods, all of them inside a single era in which this asset class went from nothing to a trillion-dollar market. That is not something that can happen a second time in the same way.
The way we would read the page: crypto has been the higher-return and far higher-volatility corner of a portfolio, and the honest planning input is both columns together. Neither “it beat the market” nor “it is too volatile” survives contact with the whole table.
What this page cannot tell you
- It is not a portfolio. Four separate single-asset lines are not an allocation. Holding 5% bitcoin next to 95% index funds behaves nothing like either line here, and this page does not model it.
- The window is doing most of the work. Switch between the periods above and the ranking changes. Any conclusion that survives only one choice of window is not a conclusion.
- Gross of everything. No fund fees, exchange fees, spreads, storage, dividend withholding or capital gains tax. All four lines are flattered, crypto usually the most.
- Survivorship. Bitcoin and ether are on this page because they are still here. The same chart drawn in 2017 would have carried names that are now worthless, and this calculation cannot include them.
- Nominal dollars. Nothing is adjusted for inflation, so every line overstates what it bought you in real terms — the benchmarks by a smaller amount than the headline numbers suggest.
- One currency, one tax residence, no rebalancing. Everything is in US dollars, bought once and never touched.
Read the full guide
Bitcoin vs Ethereum for the Long Run: What Holders Should Track
This page puts crypto next to traditional assets. That one puts the two majors next to each other, and explains what the price series alone does not capture.
Frequently asked questions
Is this the S&P 500 with dividends?
Yes. We use the S&P 500 total return index, which reinvests dividends. The plain price index that most charts show leaves out 1.6–1.9% a year on our own numbers, and over the last ten years that gap is worth about a fifth of the index’s total gain: $100 became $404 with dividends and $342 without. Comparing bitcoin against the price index would quietly stack the deck in bitcoin’s favour.
Which gold price is this?
The LBMA afternoon auction in US dollars per troy ounce — the benchmark the physical market itself settles against. No storage cost, insurance or dealer spread is deducted, so real bullion returns are somewhat lower than the line shown here.
Did bitcoin beat the stock market?
Over the five years to 28 Jul 2026, bitcoin returned +60% against +81% for the S&P 500 total return index — 1.1× less than the index. That is one window. The rolling table on this page runs the same question from every start date we hold, because a single pair of dates can be chosen to say almost anything.
Why do the numbers change when I switch the period?
Because they should. Every one of these figures depends entirely on where the window starts and stops, and crypto’s cycle is long enough that a five-year window and a ten-year window can disagree about which asset won. That sensitivity is the honest finding — treat any single “bitcoin returned X%” headline, including ours, as one draw out of many.
Are fees, taxes and inflation included?
None of them. These are raw closing prices in US dollars: no fund expense ratio, no exchange fee, no spread, no capital gains tax, no inflation adjustment. A real index fund costs 0.03–0.4% a year, real bullion costs more, and real crypto costs whatever your exchange charges. Everything here is gross of all of it.
Do trading days line up between crypto and stocks?
Crypto trades every day, the stock market and the London auction do not. Every series on this page is read on the S&P’s trading days only, taking the last known close for each on that date. Comparing on all 365 days would stretch Friday’s stock price across the weekend and understate its volatility.
How this page is calculated
- Crypto. Daily closes in UTC from CoinGecko with Binance as a fallback — the same series behind the Hodlometer Index and the DCA calculator. Bitcoin from 18 Aug 2011, ether from 9 Mar 2016.
- Stocks. The S&P 500 Total Return index (^SP500TR), which reinvests dividends, via the Yahoo Finance chart API. The price-only index understates stock returns by 1.6% a year over the last five and 1.9% over the last ten, measured on the two series side by side.
- Gold. The LBMA afternoon auction price in US dollars per troy ounce, published by the London Bullion Market Association. Spot metal, not a fund and not futures.
- Common calendar. Every series is sampled on S&P trading days, taking each one’s last known close on or before that date. Crypto weekends are therefore excluded from the comparison rather than being matched against a stale stock price.
- Growth of $100 is each series divided by its own first value in the window and multiplied by 100. A series that begins later than the window (ether before March 2016) starts at 100 on its own first day and is labelled as such.
- Annualised is the compound rate that turns the first value into the last over the exact elapsed time — not an average of yearly returns.
- Rolling windows take every trading day as a start date, hold for the horizon in calendar days, and compare the two total returns. Overlapping windows are counted once each and the independent-period count is printed next to them.
- No forecast is derived from any of this. The page reports what four price series did and stops.