Bitcoin and Ethereum Drawdowns: A 10-Year Statistical Deep Dive

A Decade of Crypto, Compressed Into Numbers

Bitcoin's price history is the closest thing financial markets have to a controlled experiment in volatility. From the cents-per-coin era through six-figure prints, BTC has experienced drawdowns whose magnitude would have ended most equity portfolios — and recovered from each of them. Ethereum, shorter-lived but only by a few years, has compiled a similar archive.

This post is a quantitative tour of those drawdowns: the major bears, what they looked like, how they compare to the S&P 500's volatility, and what a couple of simple rules-based strategies would have done across the period. All figures here are approximate, based on publicly available daily closing prices from 2014 (BTC) and 2015 (ETH) through 2024.

Table of Contents

• BTC drawdowns by cycle
• ETH drawdowns by cycle
• Side-by-side: the largest crypto bears
• Comparing crypto to equities: the volatility ratio
• Cycle theory and peak-to-peak intervals
• "Buy every -50% drop" — a backtest of intuition
• Why sample size matters more in crypto
• Key takeaways

BTC Drawdowns by Cycle

Bitcoin's history is conventionally divided into roughly four major price cycles, each marked by a peak followed by a multi-year drawdown.

• 2013-2015 cycle. Peak in late 2013 around $1,100. Trough in early 2015 near $200. Approximate peak-to-trough drawdown: -85%.
• 2017-2018 cycle. Peak in December 2017 near $20,000. Trough in December 2018 around $3,200. Approximate drawdown: -84%.
• 2021-2022 cycle. Two peaks (April 2021 near $64k and November 2021 near $69k). Trough in late 2022 around $15,500. Approximate drawdown from the November peak: -78%.
• 2024 cycle (in progress at the time of writing). New all-time highs in 2024 above the prior cycle's peak. Drawdown structure within this cycle is still being written.

A pattern is visible: each cycle has produced a peak-to-trough drawdown in the 75-85% range. None has been smaller. Whatever else BTC is, it has been an asset that periodically loses three-quarters of its value before resuming its long-term trajectory.

Within each cycle, smaller drawdowns of 20-40% are routine — BTC sees several of those per year, on average. The cycle-defining bears (75%+) are the ones that get the headlines.

ETH Drawdowns by Cycle

Ethereum's history starts in mid-2015. It has lived through fewer complete cycles than BTC, but the broad pattern echoes:

• 2017-2018 cycle. Peak in January 2018 around $1,400. Trough in late 2018 around $85. Approximate drawdown: -94%.
• 2021-2022 cycle. Peak in November 2021 near $4,800. Trough in mid-2022 around $880. Approximate drawdown: -82%.
• 2024 cycle. New highs in 2024; the cycle continues to develop.

ETH's first major bear (2018) was actually deeper than BTC's same-era decline, which is consistent with its higher beta to BTC. As a general rule, ETH amplifies BTC's moves in both directions, with deeper drawdowns and sharper recoveries.

Side-by-Side: The Largest Crypto Bears

| Asset | Cycle | Approx. peak | Approx. trough | Drawdown | Duration | |---|---|---|---|---|---| | BTC | 2013-2015 | $1,100 | $200 | -85% | ~14 months | | BTC | 2017-2018 | $20,000 | $3,200 | -84% | ~12 months | | BTC | 2021-2022 | $69,000 | $15,500 | -78% | ~12 months | | ETH | 2018 | $1,400 | $85 | -94% | ~11 months | | ETH | 2022 | $4,800 | $880 | -82% | ~7 months |

A point worth pausing on: the duration of crypto bears is much shorter than the equity equivalent. The S&P 500's 2000-2002 bear took roughly 30 months to bottom; its 2007-2009 bear took 17 months. BTC's worst bears reach their lows in 12-14 months. The collapse is sharper but more compressed in time.

That has practical consequences. Crypto bears are over before most investors finish working through their psychological reaction to them.

Comparing Crypto to Equities: The Volatility Ratio

A useful single-number summary is the volatility ratio: the standard deviation of daily returns of a crypto asset divided by that of a reference equity index.

Using rough decade-long averages of daily log returns:

• BTC daily volatility: approximately 3-4% (annualized: roughly 60-75%).
• ETH daily volatility: approximately 4-5% (annualized: roughly 75-95%).
• S&P 500 daily volatility: approximately 1% (annualized: ~16-18%).

So BTC's annualized volatility is roughly 4x the S&P 500's, and ETH's is roughly 5x. That ratio is remarkably stable across the decade — it does not vary much from one cycle to the next.

This means a crude rule of thumb works for translating equity intuition to crypto: if a -10% S&P drop feels notable, a -40% BTC drop is roughly equivalent in standardized terms. A -20% S&P drop (a classic bear-market line) corresponds, very roughly, to a -80% BTC drop. The cycle-defining BTC bears are, in volatility units, comparable to a major equity bear market.

Importantly, the direction of returns is also amplified: BTC's bull cycles have produced annualized returns over multi-year periods that are multiples of equity returns. The volatility ratio cuts both ways.

Cycle Theory and Peak-to-Peak Intervals

A widely discussed pattern in BTC history is the apparent four-year cycle, frequently linked to Bitcoin's halving schedule (the protocol-level event that reduces miner block rewards every roughly four years).

Peak-to-peak intervals in BTC's price history:

• 2013 peak to 2017 peak: approximately 4 years.
• 2017 peak to 2021 peak: approximately 4 years.
• 2021 peak to 2024-25 new highs: approximately 3-3.5 years.

Whether this pattern is causal (driven by halving-induced supply shocks), behavioral (driven by narrative cycles), or coincidence (three data points is a small sample) is a debate that has more heat than light. The honest statistical position is that three observations do not establish a cyclic law. They establish a hypothesis. Each future cycle either confirms it or weakens it.

ETH's price history is even shorter and does not have a clean halving analog. Its cycles broadly track BTC's, with amplification.

"Buy Every -50% Drop" — A Backtest of Intuition

One of the most common retail intuitions about crypto is: "Just buy every time it drops 50%." Let's see what that would have looked like, roughly, for BTC from 2014 through 2024.

A simple rule: every time BTC's drawdown from its all-time high reaches -50% (a fresh signal — re-triggers only after a recovery and a new high), allocate one unit of capital. Hold each unit forward indefinitely.

Approximate trigger points and one-year forward outcomes for each:

| Trigger date (approx.) | BTC price at trigger | Approx. 1y forward return | |---|---|---| | Aug 2014 (post 2013 peak) | ~$500 | -25% | | Jan 2018 (post 2017 peak) | ~$10,000 | -65% | | Mar 2020 (COVID crash) | ~$6,000 | +500% | | Jun 2022 (post 2021 peak) | ~$20,000 | +50% | | Jul 2024 (within cycle) | varied | varied |

A few observations.

First, the strategy's short-term outcomes are mixed. In two of the four episodes shown, the one-year forward return was negative (sometimes substantially so). Buying -50% drops is not a guaranteed near-term win; in fact, it can be painful for a year or more before recovering.

Second, the long-term outcomes are heavily skewed by the highly positive episodes (especially March 2020). Over a multi-year horizon, the strategy looks excellent in aggregate — but most of the aggregate return comes from a small number of episodes, which is exactly the fat-tail property that statistical intuition consistently underweights.

Third, this is a tiny sample. Four to five real signals across a decade. No serious statistician would call that a strategy validation. It is more like an existence proof: the strategy has not historically been a disaster, and at least once it has worked spectacularly. That is interesting but not definitive.

Why Sample Size Matters More in Crypto

Equity indices have decades — in some cases over a century — of price history. The S&P 500's 100+ year archive contains hundreds of distinct 5%+ drawdowns and dozens of 20%+ bears. Statistical inferences about the index have a large-N foundation.

BTC has about ten years of meaningful price history. ETH has about nine. Within each, the number of deep drawdowns (50%+) is in the single digits. The number of complete cycles is three or four.

This means base-rate analysis on crypto is structurally noisier than on equities. A median forward return from a particular drawdown bucket might be based on dozens (or fewer) of observations rather than hundreds. The headline number is informative but should always be read alongside the bucket's sample size and prediction error.

A practical implication: in crypto, prediction error is typically much larger relative to the median forecast than in equities. A forecast of "+20% over 90 days" from a deep BTC drawdown might come with a ±40% historical error — meaning the realized outcome could easily be -20% or +60%. The signal is still useful; it is just not a target.

Conservative position sizing matters more in crypto for this reason. It is not that the median returns are bad — they are, on average, quite good from deep drawdowns. It is that the dispersion around those means is wider, and the sample size that informs them is smaller.

Key Takeaways

• BTC has produced three cycle-defining bears in a decade, each in the -78% to -85% range. ETH has matched or exceeded those magnitudes.
• Crypto bears are deeper but faster than equity bears: 7-14 months from peak to trough vs. 12-30 for major equity indices.
• BTC's annualized volatility is roughly 4x the S&P 500's; ETH's is roughly 5x. Mental rule of thumb: a -10% S&P drop is comparable, in standardized terms, to a -40% BTC drop.
• The four-year cycle pattern is suggestive but rests on only three observations. It is a hypothesis, not a law.
• A "buy every -50% drop" rule has had mixed short-term outcomes but solid long-term results — driven heavily by a small number of episodes, which is the hallmark of fat-tail assets.
• Sample size is the binding constraint on statistical inference in crypto. Always read forward-return forecasts alongside the bucket's prediction error.

Try It Yourself

Pull up BTCUSD and ETHUSD side by side. Compare the bucket sample sizes, the median forward returns, and especially the prediction errors. You will see immediately how much wider the error bands are than on a typical large-cap equity — which is the data telling you to respect the noise floor.

Educational content. Past performance does not guarantee future results. This is not financial advice.