Definition you can verify first
Maximum Drawdown (MDD) is the largest observed decline from a historical peak to a subsequent trough over a specified measurement period. To verify information about MDD, you start by confirming what “peak,” “trough,” and “largest” mean in that source.
A practical way to make the definition checkable is to require three explicit items:
- Input series: the time series used (for example, account equity, net asset value, or a computed value from returns).
- Measurement window: the start and end dates (or number of bars).
- Drawdown rule: whether drawdown is computed as a percentage or an absolute amount, and how the algorithm finds the peak before each trough.
If a source omits any of these, the claim cannot be independently reproduced.
Mechanics: what to compute, with stated assumptions
To verify MDD, use the same step-by-step method on the same underlying series. A common reproducible approach is:
- Choose a measurement window and a start index. Assumption: the chosen window matches the source’s reported period.
- Construct the series values V(t) (e.g., equity at time t). Assumption: V(t) already includes the relevant costs and is computed consistently.
- Compute the running peak: P(t) = max{V(τ) for τ ≤ t}.
- Compute drawdown at each time: DD(t) = V(t)/P(t) − 1 (percentage form). For absolute form, use V(t) − P(t).
- The Maximum Drawdown over the window is the minimum drawdown: MDD = min{DD(t) over t in window}.
Verification is then not about trusting a published number, but about checking whether the same algorithm applied to the same series gives the same result (within rounding).
Evidence and reproducible example (with failure checks)
A verification-friendly example uses an invented, small series so you can reproduce the logic:
- Assume V(t) for five points is: 100, 110, 105, 120, 90.
- Running peaks: 100, 110, 110, 120, 120.
- Drawdowns DD(t) = V/P − 1: 0, −0.00, −0.0455…, 0, −0.25.
- MDD is the minimum: −0.25 (a 25% peak-to-trough decline in that window).
If a real source reports a different magnitude, the most likely explanations to check are:
- Different input series: returns-derived “equity curve” vs reported equity.
- Different window: including/excluding a warm-up period changes peaks.
- Different rounding: percentage formatting can hide small differences.
Material limitation: even if MDD is computed correctly, different resampling choices (e.g., daily vs intraday, end-of-day vs intraday minima) can change the largest trough and therefore the reported MDD.
Limitations and risks to understand before trusting numbers
- Method mismatch: sources may compute drawdown using different series, start dates, or drawdown definitions (absolute vs percentage). This prevents direct verification.
- Incomplete accounting of costs: if the series used by the source excludes fees, financing, or spreads, the drawdown figure may not match a net-of-cost view.
- Data granularity risk: if intraday lows are not included, the true trough may be missed, understating MDD.
- Future-meaning limitation: Maximum Drawdown is descriptive of observed history within a window. Historical relationships do not guarantee future outcomes.
Verification checklist and next question
Use a short checklist to verify MDD information independently:
- Identify the exact input series and whether it is equity/net value or something derived.
- Confirm the measurement period (start/end) matches the source.
- Confirm the drawdown formula (percentage vs absolute) and rounding conventions.
- Recompute using a running peak method and check if the peak-to-trough path matches.
Next question to resolve when verifying: Does the source specify the series construction, window, and drawdown rule clearly enough to reproduce the same peak and trough positions?