# Bitcoin Market Cycles and the Power-Law Debate: Rallies, Drawdowns and Evidence

*Bitcoin’s price history contains extraordinary appreciation and repeated losses exceeding three quarters of a prior peak. The record is measurable; the popular stories attached to it require more restraint.*

- Research question: What does the historical price record establish about Bitcoin’s ascents and drawdowns, and what can halving comparisons or a power-law fit legitimately explain?
- Canonical: https://degreesofsatoshi.com/history/bitcoin-market-cycles-and-power-law/
- Published 2026-08-09 · last reviewed 2026-08-09 · covers 2010-07-18/2025-12-31
- Author: Degrees of Satoshi editorial project (https://degreesofsatoshi.com/authors/degrees-of-satoshi-editorial-project/)
- Citation files: https://degreesofsatoshi.com/history/bitcoin-market-cycles-and-power-law/citation.bib · https://degreesofsatoshi.com/history/bitcoin-market-cycles-and-power-law/citation.ris · https://degreesofsatoshi.com/history/bitcoin-market-cycles-and-power-law/citation.csl.json
- Evidence data: https://degreesofsatoshi.com/history/bitcoin-market-cycles-and-power-law/evidence.json

## Key finding

Four completed daily-close drawdowns exceeded 76%, while the full-period log-log regression fits its own historical sample closely. Neither repeated recovery nor a high in-sample R-squared turns a halving narrative or power curve into a causal law or dependable forecast. ([Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json); [GetBlockSubsidy implementation](https://github.com/bitcoin/bitcoin/blob/128456b62d5e38abea031f97f823d5b28aef9357/src/validation.cpp#L1844-L1855); [Bitcoin Price Prediction: Peer-Reviewed Evidence and Social Media Discourse](https://arxiv.org/abs/2606.00071))

## Verified record

| Observation | Peak or sample | Trough or result | Exact measurement | Source |
| --- | --- | --- | --- | --- |
| Daily price series | 18 July 2010 | 31 December 2025 | 5,646 positive UTC daily PriceUSD observations; nominal U.S. dollars | [Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json) |
| 2011 drawdown | 8 June 2011 · $29.029921 | 18 November 2011 · $2.105066 | −92.7486%; prior peak first recovered on 19 February 2013 | [Coin Metrics community BTC data · pinned revision](https://github.com/coinmetrics-io/data/blob/f1a36afb962731c387bb03982758ab0103063da5/csv/btc.csv) |
| 2013–2017 drawdown | 4 December 2013 · $1,134.932231 | 14 January 2015 · $175.637641 | −84.5244%; prior peak first recovered on 23 February 2017 | [Coin Metrics community BTC data · pinned revision](https://github.com/coinmetrics-io/data/blob/f1a36afb962731c387bb03982758ab0103063da5/csv/btc.csv) |
| 2017–2020 drawdown | 16 December 2017 · $19,640.513883 | 15 December 2018 · $3,185.074044 | −83.7831%; prior peak first recovered on 30 November 2020 | [Coin Metrics community BTC data · pinned revision](https://github.com/coinmetrics-io/data/blob/f1a36afb962731c387bb03982758ab0103063da5/csv/btc.csv) |
| 2021–2024 drawdown | 8 November 2021 · $67,541.755508 | 9 November 2022 · $15,758.291282 | −76.6688%; prior peak first recovered on 4 March 2024 | [Coin Metrics community BTC data · pinned revision](https://github.com/coinmetrics-io/data/blob/f1a36afb962731c387bb03982758ab0103063da5/csv/btc.csv) |
| Full-sample power-law fit | 18 July 2010–31 December 2025 | 5,646 observations | log10(price) = −16.540574 + 5.699593 × log10(days since genesis); log-space R² 0.960985 | [Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json) |

## A price history is a market record, not a Bitcoin consensus record {#what-the-price-series-measures}

Bitcoin does not contain a dollar price. The protocol validates transactions, blocks, and issuance; prices arise on exchanges where bids and offers meet under particular custody, liquidity, and jurisdictional conditions. The series used here is Coin Metrics PriceUSD, sampled daily in UTC and reproduced in the site dataset. It is an index-style market observation rather than the closing auction of one canonical Bitcoin venue. That distinction matters most in the early years, when markets were thin, fragmented, and operationally fragile.

All prices in this dossier are nominal U.S. dollars. They are not adjusted for consumer-price inflation, and daily observations suppress intraday extremes. A date identified as a peak or trough therefore means the relevant daily observation in this dataset, not necessarily the highest or lowest trade printed anywhere in the world. Publishing the metric, time convention, retrieval date, and source checksum makes the analysis reproducible without pretending that one series exhausts every historical market.

Sources: [Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json); [Coin Metrics API documentation](https://docs.coinmetrics.io/api); [Coin Metrics prices methodology](https://gitbook-docs.coinmetrics.io/market-data/methodologies/coin-metrics-prices-methodology); [Coin Metrics community BTC data · pinned revision](https://github.com/coinmetrics-io/data/blob/f1a36afb962731c387bb03982758ab0103063da5/csv/btc.csv); [GetBlockSubsidy implementation](https://github.com/bitcoin/bitcoin/blob/128456b62d5e38abea031f97f823d5b28aef9357/src/validation.cpp#L1844-L1855)

## The ascent is real, but a linear chart conceals most of the history {#price-ascent-on-a-log-scale}

The dataset’s 31 December observations move from $0.30 in 2010 to $13.55 in 2012, $729.56 in 2013, $13,921.48 in 2017, $29,022.67 in 2020, and $93,389.73 in 2024. Expressed as a simple endpoint multiple, that is an exceptional long-run ascent. Expressed as a lived market path, it includes several periods in which a holder at the prior high saw more than four fifths of the quoted dollar value disappear before recovery. Both statements describe the same series.

A logarithmic price axis is normally the more honest way to show a market that spans many orders of magnitude because equal vertical distances represent equal proportional changes. It should not be used to make volatility look harmless. The companion drawdown view starts each new high at zero and measures the percentage distance below that running high. Together, log price and drawdown preserve the upward scale change while showing the severity and duration of the losses along the way.

Sources: [Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json); [Coin Metrics community BTC data · pinned revision](https://github.com/coinmetrics-io/data/blob/f1a36afb962731c387bb03982758ab0103063da5/csv/btc.csv)

## Four completed cycles contained losses between 76.67% and 92.75% {#four-completed-major-drawdowns}

For this dossier, a completed drawdown begins at a new daily closing all-time high, reaches the lowest later daily close, and ends on the first daily close that equals or exceeds the old peak. The 2011 decline fell 92.75% over 163 days and required 622 days from peak to first recovery. The decline from December 2013 reached 84.52%; its first recovery did not arrive until February 2017, 1,177 days after the peak.

The December 2017 peak was followed by an 83.78% loss and a 1,080-day peak-to-recovery interval. The November 2021 peak was followed by a 76.67% loss, with the old level first exceeded in March 2024 after 847 days. Calling these episodes ‘crypto winters’ is convenient market language, not an objective protocol classification. The explicit rule above makes inclusion reproducible and avoids declaring a current decline complete before a later recovery or lower trough can be observed.

Sources: [Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json); [Coin Metrics community BTC data · pinned revision](https://github.com/coinmetrics-io/data/blob/f1a36afb962731c387bb03982758ab0103063da5/csv/btc.csv)

## Prices rose after each observed halving, but four windows cannot identify an effect {#halvings-and-market-windows}

The local dataset compares the daily price on each subsidy-halving date with observations 365 calendar days before and after it. The following-year returns were approximately 8,069.11% after the 2012 halving, 284.42% after 2016, 558.92% after 2020, and 31.11% after 2024. The prior-year returns were also positive in all four windows: approximately 385.26%, 141.75%, 17.29%, and 129.75%. These calculations establish the historical paths around the selected dates, not what would have happened without a halving.

A halving changes the maximum subsidy a valid block may create. It does not command buyers, set a dollar price, or create a mechanical timetable for demand. The four episodes are not independent randomized trials: expectations can develop before the event, the same global asset appears in every observation, and liquidity, leverage, monetary conditions, regulation, custody, and investor composition changed at the same time. Event windows belong in the evidence archive, but a causal claim needs an identification strategy that a recurring annotated chart does not supply.

Sources: [Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json); [GetBlockSubsidy implementation](https://github.com/bitcoin/bitcoin/blob/128456b62d5e38abea031f97f823d5b28aef9357/src/validation.cpp#L1844-L1855)

## The power law is a descriptive log-log regression, not a protocol law {#what-the-power-law-fit-is}

The published fit defines network age as days since 3 January 2009 and estimates ordinary least squares on positive daily observations: `log10(price_usd) = alpha + beta × log10(days_since_genesis)`. For the 18 July 2010 through 31 December 2025 sample, alpha is −16.5405739393, beta is 5.6995933722, and log-space R² is 0.9609848770. The residual standard deviation is 0.303042 log10 units, equivalent to a descriptive multiplicative factor of about 2.009 around the fitted line.

The modern public argument is closely associated with Giovanni Santostasi’s time-based Bitcoin power-law posts. A 2026 peer-reviewed article by Şenel and Yılmaz applied a related price-versus-time model and reported evidence of long-horizon mean reversion and shorter-horizon momentum. A separate 2026 research review by Carlos Baquero emphasized that the Bitcoin price power law had not yet received the kind of formal distributional and walk-forward evaluation needed to turn an attractive historical fit into a settled forecasting result. Those records make ‘debate’ the accurate label: there is a reproducible pattern, an affirmative research case, and unresolved model-risk criticism.

The high R² says the selected curve summarizes much of the variation in these transformed historical data. It does not establish a physical scaling law, a causal mechanism, or future accuracy. Price and network age both trend; daily errors are serially correlated; the earliest observations have unusual leverage; and the origin, index, sampling interval, start date, and loss function are researcher choices. Calling the relationship a ‘law’ can obscure those modeling decisions and make an in-sample summary sound inevitable.

Sources: [Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json); [Coin Metrics prices methodology](https://gitbook-docs.coinmetrics.io/market-data/methodologies/coin-metrics-prices-methodology); [Bitcoin power law over a ten-year period](https://www.reddit.com/r/Bitcoin/comments/9cqi0k/bitcoin_power_law_over_10_year_period_all_the_way/); [Bitcoin Valuation Through Power Law Analysis: Evidence for Long-Term Mean Reversion and Short-Term Momentum](https://dergipark.org.tr/en/pub/jbef/article/1982320); [Bitcoin Price Prediction: Peer-Reviewed Evidence and Social Media Discourse](https://arxiv.org/abs/2606.00071)

## Changing the sample changes the curve {#fit-sensitivity-and-forecast-restraint}

The site artifact reruns the same regression after progressively excluding early data. Starting in 2011 produces beta 5.55596 and R² 0.95352; starting in 2012 produces beta 5.65425 and R² 0.94244; starting in 2013 produces beta 5.32894 and R² 0.91784. Those are not catastrophic changes, but they are large enough to produce increasingly different extrapolations. A residual band calculated from the fitted sample is likewise not a confidence interval or prediction interval when the model’s errors violate ordinary independence assumptions.

The responsible use is descriptive: show the observations, publish the equation and sensitivity table, and let readers see where market cycles sit above or below the curve. This dossier deliberately supplies no future price target. It also treats Mt. Gox, regulated futures, FTX, interest-rate changes, and other events as contemporaneous context rather than arrows that ‘explain’ a drawdown by themselves. The result is less dramatic than a deterministic cycle theory and more useful to researchers who need a transparent record.

Sources: [Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json); [Bitcoin Price Prediction: Peer-Reviewed Evidence and Social Media Discourse](https://arxiv.org/abs/2606.00071); [Mt. Gox civil rehabilitation announcement](https://www.mtgox.com/img/pdf/20140228-announcement_eng.pdf); [CME Bitcoin futures launch notice](https://www.cmegroup.com/media-room/press-releases/2017/12/01/cme_group_self-certifiesbitcoinfuturestolaunchdec18.html); [FTX founder sentencing record](https://www.justice.gov/archives/opa/pr/samuel-bankman-fried-sentenced-25-years-his-orchestration-multiple-fraudulent-schemes)

## Power-law coefficient sensitivity

The same log-log specification changes as early observations are removed. These are descriptive in-sample fits, not forecasts.

| Sample start | Observations | Alpha | Beta | Log-space R² | Interpretation |
| --- | --- | --- | --- | --- | --- |
| 2010-07-18 | 5,646 | −16.540574 | 5.699593 | 0.960985 | Published full-sample fit |
| 2011-01-01 | 5,479 | −16.028095 | 5.555957 | 0.953520 | Excludes the earliest 167 observations |
| 2012-01-01 | 5,114 | −16.381321 | 5.654249 | 0.942439 | Later start changes coefficients and lowers fit |
| 2013-01-01 | 4,748 | −15.203931 | 5.328942 | 0.917844 | A materially flatter exponent under the same equation |

## Claims and their limits

- The published analysis documents a fit using 5,646 positive daily PriceUSD observations from 18 July 2010 through 31 December 2025; its downloadable artifact contains selected annual observations, halving windows, fit statistics, diagnostics, and caveats rather than all daily rows. ([Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json); [Coin Metrics community BTC data · pinned revision](https://github.com/coinmetrics-io/data/blob/f1a36afb962731c387bb03982758ab0103063da5/csv/btc.csv))
- The four completed daily-close drawdowns measured here were approximately 92.75%, 84.52%, 83.78%, and 76.67%. ([Coin Metrics community BTC data · pinned revision](https://github.com/coinmetrics-io/data/blob/f1a36afb962731c387bb03982758ab0103063da5/csv/btc.csv); [Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json))
- In the selected 365-day windows, price was higher one year after each of the four observed halvings. ([Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json); [GetBlockSubsidy implementation](https://github.com/bitcoin/bitcoin/blob/128456b62d5e38abea031f97f823d5b28aef9357/src/validation.cpp#L1844-L1855))
- The full-sample regression has beta 5.699593 and log-space R² 0.960985 under the published equation and sample definition. ([Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json))
- The regression is best treated as a contested descriptive fit rather than evidence of a causal market law or dependable forecasting model. ([Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json); [Coin Metrics prices methodology](https://gitbook-docs.coinmetrics.io/market-data/methodologies/coin-metrics-prices-methodology); [Bitcoin Price Prediction: Peer-Reviewed Evidence and Social Media Discourse](https://arxiv.org/abs/2606.00071))
- Recent research includes both an affirmative empirical application of a Bitcoin price-time power law and a methodological review calling for stronger out-of-sample and distributional testing. ([Bitcoin Valuation Through Power Law Analysis: Evidence for Long-Term Mean Reversion and Short-Term Momentum](https://dergipark.org.tr/en/pub/jbef/article/1982320); [Bitcoin Price Prediction: Peer-Reviewed Evidence and Social Media Discourse](https://arxiv.org/abs/2606.00071))

## Caveats

- PriceUSD is a market-data metric, not a value emitted or endorsed by the Bitcoin protocol.
- Daily observations omit intraday extremes, and early Bitcoin markets were thinner and less standardized than later markets.
- Prices are nominal U.S. dollars and are not adjusted for inflation.
- A completed-drawdown rule is an analytical convention; other cycle definitions can produce different dates and counts.
- Four halving windows are not independent controlled experiments and cannot identify a causal price effect.
- Power-law coefficients depend on the origin, data source, start date, transformation, and fitting method.
- The residual factor is historical dispersion, not a confidence interval, prediction interval, or trading boundary.
- Event annotations document chronology; they should not be read as proof that one event caused a rally or decline.

## Questions this dossier answers

### What counts as a Bitcoin or crypto winter in this dossier?

It is an analytical label, not a protocol state. Here a completed episode runs from a new daily closing all-time high through its lowest later close until the first daily close that regains the old peak; the highlighted major drawdowns exceed 70%. ([Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json); [Coin Metrics community BTC data · pinned revision](https://github.com/coinmetrics-io/data/blob/f1a36afb962731c387bb03982758ab0103063da5/csv/btc.csv))

### Does Bitcoin follow a power law?

A power curve fits the selected historical log-price and log-age sample closely, but that does not make it a physical or protocol law. Its coefficients change with the sample, origin, price series, and fitting method, and this dossier does not use it as a forecast. ([Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json); [Bitcoin Valuation Through Power Law Analysis: Evidence for Long-Term Mean Reversion and Short-Term Momentum](https://dergipark.org.tr/en/pub/jbef/article/1982320); [Bitcoin Price Prediction: Peer-Reviewed Evidence and Social Media Discourse](https://arxiv.org/abs/2606.00071))

### Did halvings cause Bitcoin’s past price increases?

The observed price was higher one year after each of the four completed halvings, but four overlapping historical episodes cannot isolate causation. Expectations, demand, liquidity, leverage, macroeconomic conditions, regulation, and market infrastructure changed at the same time. ([Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json); [GetBlockSubsidy implementation](https://github.com/bitcoin/bitcoin/blob/128456b62d5e38abea031f97f823d5b28aef9357/src/validation.cpp#L1844-L1855))

### Why use daily closes instead of intraday highs and lows?

Daily UTC observations provide a consistent, reproducible rule across a long and fragmented market history. They necessarily omit intraday extremes, so every peak, trough, drawdown, and recovery on this page is explicitly a daily-observation result. ([Bitcoin price, halving and power-law selected-observation set](https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json); [Coin Metrics prices methodology](https://gitbook-docs.coinmetrics.io/market-data/methodologies/coin-metrics-prices-methodology); [Coin Metrics community BTC data · pinned revision](https://github.com/coinmetrics-io/data/blob/f1a36afb962731c387bb03982758ab0103063da5/csv/btc.csv))

## Sources

1. Bitcoin price, halving and power-law selected-observation set, Degrees of Satoshi editorial project. https://degreesofsatoshi.com/assets/history/data/bitcoin-price-halving-model-2010-2025-v1.1.0.json
2. Coin Metrics API documentation, Coin Metrics. https://docs.coinmetrics.io/api
3. Coin Metrics prices methodology, Coin Metrics. https://gitbook-docs.coinmetrics.io/market-data/methodologies/coin-metrics-prices-methodology
4. Coin Metrics community BTC data · pinned revision, Coin Metrics. https://github.com/coinmetrics-io/data/blob/f1a36afb962731c387bb03982758ab0103063da5/csv/btc.csv
5. GetBlockSubsidy implementation, Bitcoin Core. https://github.com/bitcoin/bitcoin/blob/128456b62d5e38abea031f97f823d5b28aef9357/src/validation.cpp#L1844-L1855
6. Bitcoin power law over a ten-year period — Giovanni Santostasi, Reddit. https://www.reddit.com/r/Bitcoin/comments/9cqi0k/bitcoin_power_law_over_10_year_period_all_the_way/
7. Bitcoin Valuation Through Power Law Analysis: Evidence for Long-Term Mean Reversion and Short-Term Momentum — İlhan Kerem Şenel and Faruk Yılmaz (2026-06-30), Journal of Business Economics and Finance. https://dergipark.org.tr/en/pub/jbef/article/1982320 · doi:10.17261/Pressacademia.2026.2034
8. Bitcoin Price Prediction: Peer-Reviewed Evidence and Social Media Discourse — Carlos Baquero (2026-05-20), arXiv. https://arxiv.org/abs/2606.00071
9. Mt. Gox civil rehabilitation announcement, MtGox Co., Ltd.. https://www.mtgox.com/img/pdf/20140228-announcement_eng.pdf
10. CME Bitcoin futures launch notice, CME Group. https://www.cmegroup.com/media-room/press-releases/2017/12/01/cme_group_self-certifiesbitcoinfuturestolaunchdec18.html
11. FTX founder sentencing record, U.S. Department of Justice. https://www.justice.gov/archives/opa/pr/samuel-bankman-fried-sentenced-25-years-his-orchestration-multiple-fraudulent-schemes

## Cite

Degrees of Satoshi editorial project. “Bitcoin Market Cycles and the Power-Law Debate: Rallies, Drawdowns and Evidence.” Degrees of Satoshi, published 2026-08-09; last reviewed 2026-08-09. https://degreesofsatoshi.com/history/bitcoin-market-cycles-and-power-law/

- HTML: https://degreesofsatoshi.com/history/bitcoin-market-cycles-and-power-law/
- Markdown: https://degreesofsatoshi.com/history/bitcoin-market-cycles-and-power-law/index.md
- Guide for agents: https://degreesofsatoshi.com/llms.txt