Bitcoin price history, year by year: 2010 to 2025 from one sourced dataset

Where these numbers come from, and what they are not

Every price in this article comes from one file: the site’s published dataset of year-end and halving-window observations, version 1.1.0, which takes its prices from Coin Metrics, a data firm whose community data is published as a downloadable spreadsheet. The metric is Coin Metrics’ PriceUSD, one observation per day in Coordinated Universal Time, and the annual figure is the value recorded for 31 December. Prices are nominal dollars, not adjusted for inflation.

Coin Metrics builds its reference price by selecting exchange markets that pass volume and quality thresholds, discarding prices more than 3 percent from the median, taking a volume-weighted median within each minute, and averaging over a 61-minute window. That is a careful answer to “what was the price?” for the years when many exchanges traded around the clock. For 2010 and 2011, when a handful of thin markets set the price, the dataset itself cautions that early figures are less comparable with later ones.

One more thing this table is not: a forecast. The dataset’s own caveats say that four halvings are not a controlled experiment, that its fitted trend line is descriptive rather than predictive, and that nothing here is advice. Past prices tell you what happened. They tell you nothing certain about what happens next.

A source limitation matters when reusing the older version 1.1.0 file: its fields labeled price_minus_365d use the same calendar date in the previous year. For these leap-year halvings that is 366 days earlier. The annual values and following-365-day returns used in this article were independently reproduced from the pinned daily series; the older pre-window labels should not be reused as exact 365-day measurements.

2010 to 2013: from eight cents to hundreds of dollars

The first observation in the Coin Metrics series is 18 July 2010, at about $0.086. Two months earlier a programmer had paid 10,000 BTC for two pizzas, a story our Pizza Day article tells. The year ended at $0.30. By the end of 2011 the price was $4.71, and by the end of 2012, a month after the first halving on 28 November 2012, when the price was $12.33, it was $13.55.

Then came 2013. The year closed at $729.56, and the exchange handling much of the trading was Mt. Gox, where customers held balances in the company’s database rather than coins of their own. In February 2014 it suspended withdrawals, and on 28 February 2014 it applied for civil rehabilitation in Japan, saying that almost all of the roughly 850,000 BTC it thought it held had disappeared. A rescan of an old wallet on 7 March 2014 turned up 199,999.99 BTC, cutting the missing total to about 650,000. Our Mt. Gox dossier follows the creditor process, which was still paying out more than a decade later.

2014 to 2017: a long winter, then the first mania

From its 2013 high the price fell 84.52 percent to its low and took 1,177 days to recover, one of four completed drawdowns of more than 76 percent in the record. Year-ends tell the story in slow motion: $320.66 at the end of 2014, $429.68 at the end of 2015, $968.97 at the end of 2016. The second halving passed on 9 July 2016 with the price at $651.94.

2017 was the year Bitcoin became a household word. The year closed at $13,921.48, roughly fourteen times the previous year-end. The market cycles dossier lays each of these cycles over the others and is careful about what the overlap does and does not show.

2018 to 2021: the second winter and the pandemic-era peak

The fall from the 2017 peak was 83.78 percent, and recovery took 1,080 days. 2018 closed at $3,687.20 and 2019 at $7,167.40. The third halving arrived on 11 May 2020 with the price at $8,591.65, in the first months of the pandemic.

2020 ended at $29,022.67 and 2021 at $46,355.12, the highest year-end value to that point. The dataset records the return over the 365 days after the 2020 halving at about 559 percent, the third such rise in a row, and the dossier explains why three in a row is still not a law.

2022 to 2025: collapse, recovery and the ETF era

From the 2021 peak the price fell 76.67 percent, the shallowest of the four big drawdowns, and recovered in 847 days. 2022 closed at $16,524.22 and 2023 at $42,217.16.

On 10 January 2024 the Securities and Exchange Commission approved eleven exchange rule changes allowing spot bitcoin exchange-traded funds, products that hold bitcoin and trade on stock exchanges, and the first began trading on 11 January 2024. The order said the commission had approved the listing of the products, not endorsed bitcoin. The fourth halving followed on 20 April 2024 at $64,907.99, and the year closed at $93,389.73. 2025 closed at $87,516.98, below the previous year-end, with five of the largest trusts holding about 1.214 million BTC between them on 31 December 2025. Our ETF dossier tracks the holdings fund by fund.

What sixteen year-ends do and do not tell you

Laid end to end, the numbers show a price that rose by orders of magnitude, punctuated by four falls of more than three-quarters. The dataset also records what happened in the year after each halving: rises of about 8,069 percent, 284 percent, 559 percent and 31 percent. It is tempting to read that as a pattern. The dataset’s caveats say why you should not: four episodes are not a controlled experiment, and each coincided with changes in demand, liquidity, leverage, regulation and the wider economy that a halving date cannot separate out.

The same file fits a straight line through the price on logarithmic axes and finds that it describes the past closely, with an R-squared of about 0.96 in log space. The dossier labels this a contested descriptive fit, not a protocol rule or a forecast, and shows that the fitted coefficients change materially when the starting year is moved. The daily residuals are strongly serially correlated, which means the fit’s apparent tightness says little about its use out of sample.

So the honest reading is the boring one. These are the recorded prices, sourced and reproducible, and anyone can download the file and check them. What the market will do next is not in the file, and no arrangement of the past can put it there. This site publishes history, not advice.