Drawdown Recovery Time: Include Unfinished Episodes

Drawdown recovery time: Report unfinished underwater episodes separately instead of assigning them a completed recovery duration.

Drawdown recovery analysis should keep unfinished episodes visible. A strategy can have a tolerable deepest loss yet spend a long time below its previous equity high. Depth, elapsed time, and recovery status answer different questions. A percentile calculated only from episodes that recovered leaves the remaining waiting time unknown.

The Reddit discussion raised concerns about prolonged underwater periods. Author results are unverified, and some replies remained inaccessible. The exercise below is original and synthetic; it does not reproduce that strategy.

Drawdown recovery time: define drawdown recovery before counting

Start with account equity marked on a stated calendar. For each observation, retain the highest prior equity and compare current equity with that peak. Declare whether equality ends an episode or only a strictly greater value does. Use the same convention throughout. Identify the peak timestamp, first underwater observation, deepest observed trough, and first qualifying recovery. If the data ends first, label the episode unfinished and report its observed age.

Do not silently call every observation a day. Ten hourly observations and ten daily observations have different elapsed durations. Calendar gaps also matter. A three-day closure is elapsed time without three new marks. Missing marks are not known flat returns, and a closed market is different from a failed data capture.

A synthetic recovery ledger

Consider daily equity values of $10,000, $9,800, $9,700, $10,000, $9,900, and $9,850. With equality accepted as recovery, the first episode falls 3% and recovers at the fourth observation. The second episode is still open when the sample ends. Its deepest observed decline is 1.5%, and its recovery time is unknown.

An average based only on the first completed episode describes that completed subset. It does not describe both episodes. Publish completed durations alongside the unfinished count and current underwater age. State whether duration begins at the peak or at the next underwater mark. Otherwise two implementations can disagree while each follows its own definition correctly.

Keep dependence and sampling visible

For an offline resampling exercise, use portfolio equity returns with a declared mark basis and keep no-trade periods. Compare several block lengths selected before inspecting results. Longer and shorter blocks preserve different portions of the observed sequence; neither supplies missing regimes. Record the seed, horizon, and number of paths so another researcher can reproduce the calculation.

Report simulated completed recovery distributions separately from simulated episodes that remain open at the horizon. Extending the horizon changes that reporting question. It does not recover lost historical information. Avoid translating a conditional simulation percentile into a forecast of the next actual recovery date.

Freeze the response to a long drawdown

A new volatility filter or drawdown cut discovered while viewing losses becomes another research decision. Bailey and colleagues examine selection overfitting. Keep the original equity path, register proposed changes, and evaluate them on a later reserved interval. Reduced position size can reduce dollar exposure without ensuring shorter underwater time.

Use Tradewink Advisory Research to find the offline workflow and input guide. First build the six-row synthetic ledger above, then compare it with your own marked-equity file. Record cadence, gaps, equality convention, completed episodes, and unfinished ages before interpreting any chart. This is an educational diagnostic, not an instruction to change a live account.

Continue the research

Trend Following Screens: Test the Empty-Portfolio Policy and Copy Trading Wallet Rankings: Freeze Eligibility develop adjacent checks. Use a small synthetic fixture first; these educational diagnostics do not validate a trading edge or change account settings.