2026-09-22
How Deep Is Too Deep? Drawdown Depth vs Recovery Time Across 763 Securities
By FinScope Research · Data as of 2026-09-18
There is a well-known piece of arithmetic about drawdowns: a 50% loss needs a 100% gain to get back to even, and a 90% loss needs 900%. It is true, it is easy to put on a slide, and it tells you nothing about time. The question an actual holder has is not “what gain do I need,” it is “how many years am I going to sit here.”
We already measured recovery time by asset type — single stocks about 3.7 years, ETFs about 2.1. This article cuts the same universe a different way: by how far the security fell. We excluded the 52 leveraged and inverse products, whose lifetime figures are distorted by daily reset and reverse splits (we measured those separately), leaving 763 ordinary US stocks and ETFs. For each one we took its worst drawdown episode, bucketed it by depth, and measured the calendar days from trough back to the old high.
The table
| Max drawdown | Securities | Ever recovered | Median recovery | In years |
|---|---|---|---|---|
| −20% to −30% | 9 | 4 (44%) | 424 days | 1.2 |
| −30% to −40% | 37 | 31 (84%) | 238 days | 0.7 |
| −40% to −50% | 42 | 38 (90%) | 557 days | 1.5 |
| −50% to −60% | 115 | 99 (86%) | 728 days | 2.0 |
| −60% to −70% | 154 | 133 (86%) | 1,150 days | 3.1 |
| −70% to −80% | 127 | 103 (81%) | 1,321 days | 3.6 |
| −80% to −90% | 133 | 102 (77%) | 1,935 days | 5.3 |
| −90% or worse | 136 | 81 (60%) | 3,157 days | 8.6 |
Read the last two columns together. They are telling different parts of one story, and the second one is the part that is usually left out.
Recovery time bends, it does not tilt
From −30% to −60%, the median recovery climbs from roughly eight months to two years. That is bad but proportionate; the curve looks close to linear and most people’s intuition survives it.
Then it bends. Going from the −50% bucket to the −80% bucket roughly triples the wait, from 2.0 to 5.3 years, while the loss itself only grows by thirty percentage points. Past −90% the median is 8.6 years — more than four times the −50% figure.
The mechanism is compounding, and it is the same arithmetic as the break-even table, just expressed in time. To climb out of −50% you need to double. To climb out of −90% you need to grow tenfold. At any plausible rate of business growth, tenfold takes most of a decade even when everything goes right. The required gain scales geometrically, so the required time scales with it.
| Drawdown | Gain needed to break even |
|---|---|
| −30% | +42.9% |
| −50% | +100% |
| −70% | +233% |
| −90% | +900% |
The column people skip
The “ever recovered” column moves in the opposite direction, and it is arguably more important than the median.
In the −30% to −60% range, 84% to 90% of securities eventually got back to their old high. Past −90%, only 60% did. Put differently: four in ten securities that fell 90% or more in this universe have still not recovered, and many of them fell more than twenty years ago.
This matters because a median recovery time is computed only from the securities that recovered. The −90% bucket’s “8.6 years” is the median of the 81 survivors. The other 55 have no recovery time at all, and they are not in the number. The deeper the bucket, the more the median flatters the outcome, because more of the genuine disasters have been quietly excluded. Read the two columns as a pair or you will systematically underestimate the risk in the deep buckets.
The whole universe carries the same caveat at a larger scale: of 815 securities, 197 (24%) have never reclaimed the peak they fell from.
Why the shallowest bucket looks strange
The −20% to −30% row breaks the pattern: only 44% recovered, and the median is longer than the −30% bucket’s. That is a small-sample artifact, not a finding. Only nine securities in 763 have a lifetime maximum drawdown that shallow, because almost everything with a long history has been through 2008, 2020, or 2022.
The nine that qualify are mostly short-history listings and low-volatility bond funds, and several of them are in an ongoing drawdown that started recently — which counts as “not recovered” regardless of how mild it is. With n=9, one or two names move the percentage by more than ten points. We are reporting the row for completeness; do not build anything on it.
What this changes in practice
- Depth is a time commitment, not just a number. Before you average down into something that has fallen 80%, note that the median comparable took 5.3 years to repair, and that one in four never did.
- The break-even table is the same fact in a different unit. +900% is not a percentage, it is roughly a decade of flawless execution. Treat the two as one idea.
- Compare a median against its own base rate. “Typical recovery: 8.6 years” and “40% never recovered” describe the same bucket. The first number alone is marketing.
- Depth alone does not identify the cause. A −70% index fund and a −70% single stock are different risks: the fund cannot go to zero from a single business failure. That split is exactly what the asset-type breakdown measures, and it is the natural companion to this one.
To see which bucket something you own falls in, open its FinScope page and read the maximum drawdown and recovery row, or sort the whole universe in the drawdown rankings. For how the metric is defined and where it misleads, see the MDD guide, the recovery guide, and the limitations of drawdown.
FAQ
How long does it take to recover from a 50% drawdown? In this universe, securities whose worst drawdown was between −50% and −60% took a median of 728 days, about 2.0 years, from trough back to the old high. 86% of them eventually recovered.
Is recovery time proportional to drawdown depth? No. It bends upward. Moving from the −50% bucket to the −80% bucket roughly triples the median wait, from 2.0 to 5.3 years, and the −90% bucket’s median is 8.6 years. The gain needed to break even grows geometrically, and the time grows with it.
What share of deep drawdowns never recover? Among securities that fell 90% or more, only 60% ever reclaimed the old high, so roughly four in ten have not. Across all 815 securities in the universe, 197 (24%) are still below the peak they fell from.
Methodology: measured on split- and dividend-adjusted daily closes across 763 non-leveraged US stocks and ETFs as of 2026-09-18. Maximum drawdown is the largest peak-to-trough decline in the security’s full available history; recovery time is calendar days from that trough to the first close that reclaims the prior peak. Securities that have not reclaimed it are counted as “not recovered” and are excluded from median recovery times. Leveraged and inverse products are excluded here and covered in their own census. See the methodology page.