The historical record, in one place

The first thing to understand is that a nominal high and a real one are different claims. Every price on this page is quoted in dollars of the day it was recorded. A dollar in 2011 bought considerably more than a dollar does now, so a 2011 peak and a 2025 peak are not directly comparable in purchasing power even though they sit in the same column. Converting the series into constant dollars would require a price index and a base year, and this site does not publish one; the honest thing is to say plainly that the figures here are nominal, and that any comparison across a decade or more carries that qualification with it. The same caution applies to returns: a nominal gain is not the same as a gain in what the money will buy.

The second thing is that year-end snapshots systematically understate what happened inside the year. A calendar year is an arbitrary container. Bitcoin's most violent moves have often occurred in the middle of one, and a series that samples only the last day will miss them entirely. The 2013 year is the clearest example in the record: the year-end figure is a fraction of the high the market reached in November of that year, and a reader who looked only at the year-end column would conclude that 2013 was a modest year. It was not. That is why the annual prices and returns page publishes open, high, low and close for every year, and why the highs and lows table above exists at all. The snapshot series is useful for comparing one year's closing level with another's; it is not a description of the year.

The third is that drawdowns and recoveries belong together, and neither belongs on its own. A decline is only half an episode. The other half is the climb back to the previous peak, which has historically taken longer than the fall itself. Reading a list of percentage declines without the recovery times makes Bitcoin look like an asset that falls hard and springs back; reading the recovery times without the depths makes it look like an asset that simply takes a long time to do anything. Both are distortions. The drawdowns page publishes the two measures side by side for exactly this reason, and the highs and lows above are the endpoints those episodes are measured between.

The fourth is a matter of evidence rather than interpretation. The earliest years of the record rest on thinner data than the modern ones. Before Bitcoin traded on venues with public, auditable order books, prices were quoted on forums and early exchanges with small volumes and wide spreads, and the figures that survive from that period are best understood as indicative rather than precise. This is not a reason to discard them — they are the only record there is — but it is a reason to read the source and vintage columns rather than treating every row as equally firm. Where a figure is contested or thinly evidenced, the note on the row says so.

Finally, it is worth being clear about what a historical record can and cannot support. It can tell you what happened, when, and how large the moves were. It cannot tell you what will happen next, and the sample is small: a decade and a half of data covering a handful of cycles, during which the market's structure changed materially more than once. Anyone who tells you the record implies a rule about the future is overreading it. The record is a description, and a good one; treat it as such and it will be useful for a long time.

Every figure on this page is drawn from the site's curated dataset and carries its own source and vintage. Nothing is estimated, interpolated or filled in, and no price, date or statistic appears here that is not in the underlying record. Where the record is incomplete, the gap is left visible rather than papered over.

The price series is built from Coinbase Exchange daily candles for the period the venue has covered, with Bitstamp and CoinDesk historical series used for the earliest years, before Coinbase listed the pair. The two sources are joined at the point where the Coinbase record begins, and the vintage column on each row states which stretch of the record that row belongs to. The highs and lows are point observations taken from that series: the highest and lowest levels the market printed at the moments the record identifies as cycle turning points, with the date each occurred.

The year-end capitalisation series is compiled differently, and the difference matters. For each calendar year, the closing price on the last day of the year is taken from the same price series, and the circulating supply for that date is taken from the protocol's issuance schedule — the block subsidy in force at the time, accumulated from the genesis block. Market capitalisation is the product of the two. The basis column states this convention on every row, and the series is labelled year-end throughout: it is a set of annual snapshots and must never be read as a daily or average measure. Intra-year extremes are published separately, in the annual prices and returns table and in the highs and lows above.

Two limitations are worth stating plainly. First, circulating supply is an estimate of coins that exist and are spendable, not a count of coins that are accessible; coins lost to forgotten keys are included in the figure, as they are in every published supply measure. Second, the earliest price observations rest on thinner data than the modern ones, as described above. Both limitations are properties of the underlying record rather than of this presentation, and the full provenance register on the data sources page sets them out in more detail.

A reader who wants to understand the shape of the record rather than its individual figures should start with the annual prices and returns, which show each year's range, and then read the drawdowns page, which shows what the market did between the peaks. A reader who wants to place the price moves against events should read the milestones and halvings pages alongside them. And a reader who wants to run numbers on any of it — what a purchase at a given historical price would be worth, what a steady accumulation programme would have produced, what a scenario would imply — will find the tools for that on ApexCalc.net, linked from the relevant pages here.