Drawdowns, recoveries and what they measure

A drawdown is the decline from a running peak to a subsequent trough, expressed as a percentage of that peak. It is a path-dependent measure: it depends not only on where the price ended up but on the highest level it had reached before falling. An asset that triples and then halves has suffered a fifty per cent drawdown even though it is still up on where it started, which is why drawdown and return answer different questions.

Maximum drawdown is the largest such decline over a chosen period. It is the standard summary of downside risk because it describes the worst experience an investor could have had by buying at the worst moment and selling at the worst moment. For Bitcoin, the maximum drawdown over the full record is the 2011 episode, when the price fell from roughly thirty-two dollars to two — a decline of more than ninety-three per cent. No later episode has come close, and that is the single most important fact on this page.

Recovery time is the interval from the peak to the point at which the price first exceeds that peak again. It is measured from the peak, not the trough, because the investor who bought at the top is underwater for the whole interval. The 2013 peak took 1,132 days to recover; the 2017 peak took 1,095; the 2021 peak took 846. Those are long periods to hold a losing position, and they are the reason drawdown analysis matters more than the headline return figure.

The three measures are usually read together. A shallow drawdown that takes years to recover is a different risk from a deep one that recovers quickly, and the table above lets a reader see both dimensions at once. What none of them capture is the behaviour of the holder: a drawdown is only realised as a loss if the position is sold into it, and the recovery figures assume the holder stayed.

The most tempting reading of the table is that every drawdown so far has been recovered, so the next one will be too. That reading is not supported by the evidence, and it is worth being precise about why. The sample is five episodes, and four of them recovered. Five observations cannot establish a rule about an asset whose market structure has changed materially between each one. The recoveries happened, but they happened under conditions — a growing user base, an expanding set of venues, a widening pool of buyers — that are not guaranteed to repeat.

There is also a survivorship problem in the framing itself. The record only includes declines that were eventually recovered, because a decline that never recovers would not appear as a completed episode. The 2025 correction is still open in this dataset, which is a useful reminder that the table is a snapshot of an unfinished process rather than a closed set of outcomes. An asset that goes to zero would produce a final row with no recovery date, and no amount of historical recovery data would have warned the holder.

What the record does support is narrower and still valuable. It shows that Bitcoin has repeatedly fallen by more than half and remained a functioning market, that the deepest declines occurred when the asset was smallest and least liquid, and that recovery has historically taken between two and three years from the peak. Those are facts about the past. They are a reasonable basis for expecting that a future decline will be severe and slow to heal, and no basis at all for expecting that it will heal.

The honest position for a reader is to treat the drawdown record as a description of what has happened rather than a promise about what will. Position sizing, time horizon and the willingness to hold through a multi-year underwater period are the variables that actually determine the outcome, and none of them can be read off a chart of the past.