Crypto drawdown recovery time is the elapsed time from a prior equity peak to the first observation that regains that level. Record the trough-to-recovery interval separately. An unfinished drawdown has a known age but no observed recovery time. Build an episode ledger with peak, trough, recovery, depth and status so a long unresolved loss stays visible beside shorter completed recoveries.
There is no fixed number of days in which a crypto portfolio must recover. Historical recovery times depend on the measured portfolio, valuation currency, observation schedule and rules that remained active. Some episodes never recover within the available record. A useful report describes that uncertainty instead of turning a completed-episode average into a deadline.
Define the episode before starting the clock
A drawdown episode connects one reference peak to the first later observation that reaches or exceeds it. Until then, rebounds below the reference remain part of the same episode. A new local high inside a loss does not reset the original recovery target.
For positive equity, calculate depth at each observation as 100 × (1 − equity / running peak). The episode's maximum depth uses its lowest observed equity. Keep that maximum separate from the current depth: an account can be close to recovery after spending much of the episode further underwater.
The PerformanceAnalytics drawdown-table documentation distinguishes the starting high-water mark, trough, ending recovery period, depth and several duration fields. Its package documentation reports periods. Here we use elapsed calendar days between stated UTC timestamps and explicitly define the counting rules.
Use these conventions for the ledger below:
- Equality with the reference peak closes an episode; exceeding it also closes the episode.
- Repeated observations at a peak update the origin to the latest equal peak before the next decline. Flat time at the peak therefore does not extend the next episode.
- If the lowest equity repeats, retain its earliest timestamp as the trough date.
- Zero-depth stretches are not episodes. An episode opens only when an observation falls below its reference peak.
- At the endpoint, keep an unresolved episode open. Its lowest mark is the trough observed so far and can change later.
Other conventions can be legitimate. Store the timestamps so another reader can reconcile a different period-count display. The maximum-drawdown guide explains the loss denominator; this ledger adds the boundaries needed to identify every separate episode.
Follow one portfolio through four episodes
This original hypothetical fixture uses a performance-equity index with one mark at 00:00 UTC each day. Day 0 is 1 September 2026; Day 24 is 25 September. There are no external contributions or withdrawals. Assume open positions and all applicable costs are included in each invented equity mark. These are not historical market observations or KLINEO results.
| Day | Equity index | Running peak | Depth below peak |
|---|---|---|---|
| 0 | 100.00 | 100.00 | 0% |
| 1 | 110.00 | 110.00 | 0% |
| 2 | 88.00 | 110.00 | 20% |
| 3 | 104.00 | 110.00 | 5.45%, rounded |
| 4 | 110.00 | 110.00 | 0% |
| 5 | 112.00 | 112.00 | 0% |
| 6 | 106.40 | 112.00 | 5% |
| 7 | 112.00 | 112.00 | 0% |
| 8 | 112.00 | 112.00 | 0% |
| 9 | 108.64 | 112.00 | 3% |
| 10 | 112.00 | 112.00 | 0% |
| 11 | 114.00 | 114.00 | 0% |
| 12 | 108.30 | 114.00 | 5% |
| 13 | 100.32 | 114.00 | 12% |
| 14 | 104.88 | 114.00 | 8% |
| 15 | 107.16 | 114.00 | 6% |
| 16 | 111.72 | 114.00 | 2% |
| 17 | 109.44 | 114.00 | 4% |
| 18 | 112.86 | 114.00 | 1% |
| 19 | 112.86 | 114.00 | 1% |
| 20 | 111.72 | 114.00 | 2% |
| 21 | 112.86 | 114.00 | 1% |
| 22 | 112.86 | 114.00 | 1% |
| 23 | 112.86 | 114.00 | 1% |
| 24 | 112.86 | 114.00 | 1% |
Episode A uses Day 1's peak of 110 as its origin and opens when Day 2 falls below it. Its trough is 88 on Day 2, so depth is 100 × (1 − 88 / 110) = 20%. Day 3's rebound remains underwater. Day 4 reaches 110 and closes the episode.
Episode B opens after Day 5 and recovers on Day 7. Day 8 repeats the 112 peak. Under our latest-equal-peak convention, Day 8 becomes the origin for Episode C, which closes on Day 10. Counting from Day 7 instead would add one day of flat equity to C's duration.
Episode D starts from Day 11's 114 peak. Its observed trough is 100.32, giving 100 × (1 − 100.32 / 114) = 12%. None of the later observations reaches 114. The bounce on Day 16 and smaller subsequent declines remain within D.
Keep three different durations in the ledger
Peak-to-trough time describes the descent to the observed low. Trough-to-recovery time describes the subsequent climb to the reference. Peak-to-recovery time covers the whole completed episode. Those two components add to the total when they use the same timestamps.
| Episode | Peak day | Trough day | Recovery day | Maximum depth | Peak to trough | Trough to recovery | Total elapsed or open age |
|---|---|---|---|---|---|---|---|
| A | 1 | 2 | 4 | 20% | 1 day | 2 days | 3 days, recovered |
| B | 5 | 6 | 7 | 5% | 1 day | 1 day | 2 days, recovered |
| C | 8 | 9 | 10 | 3% | 1 day | 1 day | 2 days, recovered |
| D | 11 | 13 | Unobserved | 12% so far | 2 days so far | Unobserved | 13 days so far, open |
For A, 4 − 1 = 3 elapsed days. Its two strictly below-peak observations are Days 2 and 3. That count is two, which illustrates why “three days to recovery” and “two underwater observations” can both be correct.
D has already lasted longer than A, despite its smaller observed maximum depth. At Day 24 its current depth is only 100 × (1 − 112.86 / 114) = 1%. Reporting just the 1% current figure would omit its 12% low and its persistent failure to regain the old peak.
The Ulcer Index guide summarizes squared drawdown across observations. Pair that compressed measure with episode dates when the operational question concerns how long capital has remained below a specific peak.
An open episode is not a completed recovery
The completed peak-to-recovery durations here are 3, 2 and 2 days. Their mean is (3 + 2 + 2) / 3 = 2.33 days, rounded, and their median is two days. Label both as completed-episode statistics and report the accompanying count: three complete, one open.
D is not a zero-day recovery, a 13-day completed recovery or a missing row to discard. Its age is 13 days at the endpoint, and its eventual recovery time is unobserved. Averaging that age with completed durations would mix a partial observation with three finished events.
NIST's explanation of censoring describes event times that lie beyond a fixed observation endpoint as right censored. Applying that observation concept to recovery makes the missing event explicit. NIST's source concerns reliability testing; it does not supply a crypto recovery model or imply that an open episode will eventually close.
The practical reporting consequence is visible without fitting a model: a two-day median of finished episodes coexists with an open episode already 13 days old. Publish the open row beside the summary. A few episodes from one portfolio path cannot establish a reliable probability that today's loss will recover by a particular date.
Avoid false recoveries and false resets
A contribution can lift an account balance over its old high without repairing investment performance. Select the measured series first: raw capital balance and performance equity answer different questions. Use a declared flow adjustment such as the cash-flow-adjusted drawdown ledger before identifying performance episodes.
Changing valuation currency can also change the reference. A portfolio may regain a peak measured in one currency while remaining underwater in another. Changing from marked equity to realized profit halfway through a report similarly breaks comparability. Retain the same object and valuation basis across every row.
Observation gaps limit what the clock can claim. Daily marks identify the first observed recovery, not the exact intraday crossing. A crossing and renewed fall between two daily observations may remain invisible. Missing marks should be flagged; they do not establish that equity stayed below, or above, the peak throughout the gap.
At the beginning of a reporting slice, carry in the earlier peak and episode start if those records exist. Resetting the peak to the first mark in a monthly export can erase an ongoing loss. If earlier history is unavailable, label the episode boundary as unknown rather than inventing a fresh start.
Daily risk controls have a separate clock. A daily loss limit may reset at a specified session boundary while a portfolio drawdown remains open. Passing a daily gate does not close the historical episode. Likewise, calendar and threshold rebalancing rules can change holdings during an episode without resetting its performance reference.
Build a report another reader can reproduce
Use this worksheet before comparing strategies or deciding whether a report supports your existing risk rules.
| Field | What to retain |
|---|---|
| Measured series | Account or strategy boundary, currency, marks, costs and external-flow treatment |
| Observation window | Start and end timestamps, frequency, timezone and missing intervals |
| Episode boundaries | Peak, first below-peak mark, trough and first recovered mark; tie rules |
| Completed episode | Maximum depth and all three elapsed durations |
| Unfinished episode | Observed maximum depth, current depth, age and explicit open status |
| Summary context | Complete and open counts, completed-only mean/median, and longest observed open age |
Replay the rows chronologically. Verify that a bounce below the peak does not create a new episode, equality closes one, and repeated peak marks follow the stated tie rule. Check every displayed duration against timestamp subtraction, then confirm the last row's status against the final equity mark.
For this fixture, the useful report is concrete: four episodes, three recovered, a 20% deepest completed decline, and a 12% decline still open after 13 days. Keep that unfinished row visible before interpreting the two-day completed median as anything about the next recovery.