Drawdown from All-Time High

For every Bitcoin daily close since 18 July 2010, this plots the percent gap between that day’s close and the highest close the network had ever recorded up to that point. The series is non-positive by construction. It sits at zero on every day a new all-time high prints, and grows more negative the further price retreats from the running maximum — so the chart reads as a sequence of basins, each one a bear market, separated by brief flat stretches at the top edge where a fresh high reset the count.

Today the line sits at −47.2% below the running high of $124,773.51, set on 07 Oct 2025 — 251 days ago — which places the network in the Bear territory regime. Drawdown is the rare cycle read with no parameters to tune and no fit to argue about: it is arithmetic on the price series, which is exactly why it is a clean place to test whether Bitcoin’s bear markets are genuinely getting milder.

The input is a single series: every Bitcoin daily close in USD (provenance documented on the data sources page). For every day t we compute the running maximum and the distance to it:

ath_t = max(price_0, price_1, …, price_t)
dd_t = price_t / ath_t − 1

Both quantities are monotone: ath can only rise, and dd only ever returns to zero by setting a new ath. The series carries one extra column, days_since_ath, counting calendar days since the running maximum was last set — the number that climbs into the hundreds through deep bear phases and resets on every fresh peak.

The one judgement call is using daily closes rather than intraday lows. There is no smoothing and no model fit — drawdown is a pure pass over the close series — but a daily close rounds off the wicks, which understates the depth of the violent 2011 and 2013 flash crashes. The full derivation, and why we hold to closes for site-wide consistency, is on the methodology page.

Find today’s mark on the right edge and trace the basin to the left: how far below the top edge it sits is the current drawdown; how far back it stretches before touching the zero line is the time since the last all-time high. The six regime bands below carry the historical context for each depth. The thresholds are conventional and community-set, not original to this site — they exist to label the basin, not to predict the next print.

Walking the daily series and emitting one row for every multi-month underwater stretch that hit at least a 20% drawdown surfaces the cycle anatomy directly — peak, trough, depth reached, and days underwater, with nothing hard-coded. Every cycle’s deep bear is here, plus the 2021 mid-cycle correction (the one that took spot from the April $63,577 high to roughly $30k that summer) and any post-2024 dips that cleared the filter. This is the exhibit to read for the “getting shallower” claim — and for the one print that contradicts it.

Modern-era cycle bottoms step shallower on close: 2013–2015 bottomed near −85%, 2017–2018 at −84%, 2021–2022 at −77%. Each is a narrower basin than the last, and time-underwater compressed alongside it — roughly 1,130 days from the November 2013 peak to a fresh high, 1,100 from December 2017, and 850 from November 2021. That is the headline stat, and it reads exactly like a maturing asset class should.

The honest qualifier is that the line is not monotone. The deepest daily-close drawdown on the whole record is −92.8%, dated 18 Nov 2011 — the 2011 Mt. Gox-era crash, deeper than 2015’s −85%. The clean “each cycle shallower than the last” story only holds if you start the clock in 2013 and quietly drop the 2011 episode as pre-liquidity noise. That is a defensible editorial choice, but it is a choice, and it is the first thing a sceptic will reach for.

A monotone sequence of four declining numbers is the kind of pattern that looks like a law and is almost always a small sample. With four full bear markets there is no honest way to attach a meaningful confidence interval to “troughs are getting shallower” — a single −65% bottom next cycle would still fit the maturation story, and a single −82% bottom would break it, and nothing in four observations tells you which is more likely. The pattern is real in the data we have; it is not statistically significant in any rigorous sense, and anyone selling it as a guarantee that “the next bottom won’t go below −77%” is overreading the sample.

There is a mechanical story that makes the shallowing plausible rather than coincidental: Bitcoin’s market cap and holder base have grown by orders of magnitude, so an identical dollar outflow buys a smaller percentage move, and a deeper pool of long-horizon holders absorbs more selling before price caves. That is a reason to take the pattern seriously. It is not the same as proof, and the maturation it implies could just as easily lengthen the grind as shorten it (see the four-year-cycle debate below).

It says where the floor has been, never where this floor will be. Drawdown carries no information about the next print. A −55% reading deepened to −84% across 2018 and to −77% across 2022 before either bottom held, and the path down each time ran through bull-trap rallies — the mid-2019 run to roughly $13k that then retraced back toward the lows, the May–June 2022 cascade from Terra/Luna into Three Arrows into Celsius. A reading in “Bear territory” has never, on this record, marked the bottom on its own.

The 2011 print already broke the monotone. The shallowing story depends on starting in 2013. The −92.8% drawdown of 18 Nov 2011 is deeper than 2015’s, so even within the existing record the bottoms are not strictly descending. A fifth cycle that bottomed deeper than 2022 would not be unprecedented — it would just be a return to the pre-2013 regime.

The four-year cycle that produced these troughs may itself be ending. Bitwise’s Matt Hougan argued in late 2025 that the halving-driven pattern under which prior drawdowns sat may be over: “The bitcoin halving is by definition half as important as it was four years ago” (Bitwise CIO memo, 23 Dec 2025), with ETF flows, regulation, and corporate treasuries substituting for the older halving dynamics. Lyn Alden frames the same point as a liquidity-cycle reframe. If either is right, future drawdowns may not cluster in the −75-to-−85% range at all — they may bottom shallower, drag on far longer, or both, and the historical bands stop being a useful map.

Daily closes round off the intraday lows. The 2010 and 2011 Mt. Gox-era flash crashes produced intraday drawdowns deeper than the close-on-close numbers here — the 19 June 2011 incident, when a single compromised account briefly drove the Mt. Gox price to a cent, is the canonical example. We hold to daily closes for consistency with the rest of the site, but the true depth at the early troughs was worse than the line shows.

If you accumulate on a schedule, the deep-bear bands are the discount Bitcoin has historically offered patient buyers — a steady weekly buy across the 2018 and 2022 deep-bear stretches dollar-cost-averaged in well below the subsequent recovery. There is nothing to time: depth plus a multi-year horizon does the work, and trying to call the exact basin floor from this chart is precisely what the “signal or noise” section warns against.

If you are timing the cycle, treat the depth bands as regime confirmation, never as the trigger. They are slow and backward-looking; pair them with a faster floor read like the 200-week MA and a valuation read like MVRV‑Z. Given the shallowing pattern — and its weak statistical footing — do not assume the next bear stops at any particular depth in either direction.