Skip to main content
Research

How Long Deep Drawdowns Take to Recover

Across every published replay, we isolated each historical window whose peak-to-trough drawdown reached at least 20% (1,716 windows) and measured how many days it took for the trough close to reclaim the prior peak.

1,716
Deep-drawdown windows (≥20%)
28
Median recovery days (recovered)
31.6
Mean recovery days (recovered)
90%
Windows end before recovery

Key findings

  • ① Most replay windows (90%) end before the pre-crash peak is reclaimed — because the windows are sliced around the crisis itself. Recovery, in other words, usually needs a longer lens than the event.
  • ② Among the 175 windows that did recover within their bounds, the median was 28 trading days and the mean 31.6 — sharp V-shaped recoveries are the minority.
  • ③ The deepest drawdown in the dataset is WTI crude in 2020, where settlement prices turned negative — a reminder that drawdown percentages behave differently in markets where prices can go below zero.

The 30 deepest drawdown windows

Download dataset CSV (60 deepest windows)
#MarketWindow fromReplayDrawdownRecovery (days)
1Commodities2020-04-08WTI Crude Oil Historical Replay: 2020-04 to 2020-07−250.0%16
2Commodities2020-03-17WTI Crude Oil Historical Replay: 2020-03 to 2020-05−232.8%not in window
3Commodities2020-03-24WTI Crude Oil Historical Replay: 2020-03 to 2020-06−232.8%19
4Commodities2020-01-29WTI Crude Oil Historical Replay: 2020-01 to 2020-06−170.0%not in window
5Commodities2020-02-05WTI Crude Oil Historical Replay: 2020-02 to 2020-05−170.0%not in window
6Commodities2019-11-14WTI Crude Oil Historical Replay: 2019-11 to 2020-05−159.5%not in window
7FX2013-04-25USD/VND Historical Replay: 2013-04 to 2013-08−99.9%9
8FX2013-11-15USD/VND Historical Replay: 2013-11 to 2014-05−99.9%not in window
9FX2013-06-18USD/COP Historical Replay: 2013-06 to 2013-10−99.8%34
10FX2013-04-25USD/COP Historical Replay: 2013-04 to 2013-08−99.8%not in window
11FX2014-10-14USD/LBP Historical Replay: 2014-10 to 2015-04−99.5%not in window
12FX2016-09-26USD/CLP Historical Replay: 2016-09 to 2017-03−99.3%not in window
13FX2016-11-16USD/CLP Historical Replay: 2016-11 to 2017-05−99.3%not in window
14FX2014-10-14USD/COP Historical Replay: 2014-10 to 2015-04−99.0%4
15FX2013-11-15USD/CLP Historical Replay: 2013-11 to 2014-05−99.0%not in window
16FX2014-08-18USD/COP Historical Replay: 2014-08 to 2015-01−99.0%14
17FX2015-11-06USD/COP Historical Replay: 2015-11 to 2016-04−99.0%not in window
18FX2015-11-06USD/RUB Historical Replay: 2015-11 to 2016-04−99.0%1
19FX2018-02-14USD/NGN Historical Replay: 2018-02 to 2018-08−99.0%1
20FX2018-07-09USD/NGN Historical Replay: 2018-07 to 2018-10−99.0%1
21FX2014-06-08USD/ISK Historical Replay: 2014-06 to 2014-09−98.3%2
22FX2014-07-23USD/ISK Historical Replay: 2014-07 to 2014-10−98.3%2
23FX2014-08-25USD/PKR Historical Replay: 2014-08 to 2015-01−98.1%not in window
24FX2014-10-14USD/PKR Historical Replay: 2014-10 to 2015-04−98.1%not in window
25FX2023-02-06USD/RUB Historical Replay: 2023-02 to 2023-05−93.4%1
26FX2018-08-22USD/ALL Historical Replay: 2018-08 to 2018-11−92.1%4
27FX2018-10-23USD/ALL Historical Replay: 2018-10 to 2019-01−92.1%4
28FX2018-09-06USD/ALL Historical Replay: 2018-09 to 2018-12−92.1%4
29FX2008-02-04USD/UGX Historical Replay: 2008-02 to 2008-07−90.0%not in window
30FX2022-03-22USD/ALL Historical Replay: 2022-03 to 2022-06−89.6%not in window

Scope and caveats

  • ① The universe is all published replays with peak-to-trough drawdowns of 20% or worse (1,716 windows); 175 recovered within their window and 1541 are censored — unrecovered, and never counted as zero-day recoveries.
  • ② Recovery days counts calendar days from the trough close to the first close at or above the prior peak. Because source timestamps are inconsistent across the archive, we publish only relative durations and window starts — no intra-window dates.
  • ③ Nothing here is investment advice; replays are simulated-capital educational scenarios (see the Risk Disclaimer).