Bitcoin vs other assets: 15 you could buy, 54 you couldn't

From 17 September 2014 to 1 April 2026, $50 a week — 603 purchases, $30,150 — into each of 69 assets. Nvidia turned it into $1,963,971, or 65.2× what went in. Bitcoin finished 2nd at 44.5×. 64 of the 69 finished below 4×, and 3 failed to beat inflation.

Jon Phillips · 25 August 2026 · 6 min read

The honest way to start is with the result that does not help me: Nvidia beat Bitcoin.

Same money, same schedule, same eleven and a half years. $30,150 went into each of 69 different things at $50 a week, and Bitcoin came 2nd.

One number in that sentence needs qualifying before anything else does. Of the 69 things measured, only 15 were actually purchasable — the other 54 are 51 state house-price indices, the CPI, and two modeled savings products. They are in the ranking because the simulator offers them and because they are the honest yardsticks most people measure Bitcoin against. They are not a field of 54 rivals, and this post does not pretend they are. Bitcoin is 2nd of 69 and 2nd of the 15.

What was measured

One plan, run 69 times.

Everything is ranked on final value ÷ money paid in, not on total return. Those are different numbers for a schedule that buys over time, and the ratio is the one that stays comparable when two plans do not pay in exactly the same amount.

Three things to know before the table

I would rather put these above the result than in a footnote under it.

The window ends 1 April 2026, not today. State house-price indices are published quarterly. If Bitcoin were measured to its own last close and housing to a date five months earlier, Bitcoin would be credited with five months that housing never got — a comparison that looks rigorous and is not. So every asset shares one window, and the window ends where the slowest-reporting member of it ends.

A quarterly series buys at a stale price. A weekly plan buying a quarterly index pays the last published figure for up to three months. That is a real modelling artefact, it applies to all 51 housing rows and to CPI, and there is no way around it short of dropping them.

A house-price index is not a housing return. It excludes mortgage leverage, rental income, property taxes, maintenance and transaction costs — which between them are most of what actually happens to someone who buys a house. Treat those 51 rows as what the index did, not as what a landlord made.

The shape of it

64 of the 69 finished below 4×. Two finished above 40×. That gap is the whole story of the eleven years, and it is why the axis has to be logarithmic — on a linear one, sixty of these results are a single stripe against the left edge.

The dense block on the left is not one asset. It is 52 indices — 51 state housing series and CPI — sitting between 1.23× and 1.67×. Every state in the union fits inside a range narrower than the gap between Bitcoin and Nvidia.

What beat Bitcoin, and what that costs the argument

Nvidia. One thing, out of 69.

Bitcoin is 2nd out of 69 overall and 2nd among the 15 things you could actually have bought. That is a strong result. It is not the result the average Bitcoin comparison page reports, because the average Bitcoin comparison page picks the three assets it wants to be compared against.

The cut-off date is not what produces this. A reader is entitled to suspect that a window ending five months in the past was chosen to flatter something. So the same two assets, measured to 22 August 2026 — the latest date both still post a price for — give Bitcoin 48.8× and Nvidia 75.8×. The gap gets wider, not narrower. The shared window is the conservative choice here, and it is conservative in the direction that costs Bitcoin.

The inflation line

65 of the 69 beat CPI. Over 11.5 years the dollar's own purchasing power came to 1.23×, and almost everything cleared it.

The 3 that did not: Louisiana housing, the 3% CD and Bonds.

That is the sentence worth sitting with. A 1.19× outcome from the modeled 3% CD is not a loss in dollars — it finishes with more money than went in — but it finishes with less than the money was worth when it went in. The safest row in the table is one of the three that lost.

The modeled 7% college-savings plan lands at 1.51×, 35th of 69 — mid-table, ahead of most of the country's housing markets and behind every equity in the set. Both of these are constant-yield assumptions rather than measured markets, which is why they carry a modeled tag; a real 529 does not return exactly 7% a year and a real CD ladder does not return exactly 3%.

The full ranking

Nothing in the set finished below what was paid into it. The weakest, Bonds, still came to 1.11×. That is a fact about this specific 11.5-year window — one that contained no decade-long bear market in anything measured — and not a general property of these assets.

What is not in here, and why

3 things the simulator offers are missing from this ranking: Ethereum (no price before 9 November 2017), Solana (no price before 10 April 2020) and commercial real estate (its index stops at 1 January 2026).

Each was dropped for the same reason: it cannot cover the full window, and shortening everyone else's window to accommodate it would change 69 results to accommodate one. Ethereum and Solana in particular would look very different measured from their own first close, and they deserve that treatment rather than this one.

What this does not tell you

One schedule is not the distribution. Every figure here starts on 17 September 2014 and buys every week. Change the start date and the ranking moves — sometimes a lot. The companion piece on start dates measures that directly for Bitcoin, across every weekly start on record.

Past order is not future order. Nvidia won this window on a semiconductor cycle that had not happened when the window opened. Nothing here forecasts the next eleven years, and a ranking built on a period that ended in a rising market for nearly everything is not a neutral sample.

Fees, spreads, taxes and dividends are all absent. No transaction costs on 603 purchases, no capital-gains treatment, and equity figures are price-only — a dividend-paying index is understated here relative to its total return.

The universe is what this simulator carries, which is a choice: US equities, three metals, bonds, 51 US housing indices, CPI and one cryptocurrency with history back to 2014. Where each series comes from is written up in the knowledge base, and who is writing this is on the about page.

Run it on your own numbers

The whole table above is one schedule. If yours starts on a different date, or buys a different amount, or stops early, the answer is different. Open the simulator and use your own.