A max drawdown calculator measures one specific thing: the largest fall from a running high to a subsequent low across a series, expressed as a percentage of that high. It is the arithmetic of the worst stretch a series actually experienced, computed after the fact from numbers you supply. It is not a forecast, not a limit on future losses, and not a statement that anything similar will or will not happen again.
Arb Digital publishes this alongside the Sharpe ratio calculator, which measures excess return per unit of total volatility, and the Sortino ratio calculator, which penalises only downside deviation. Those two summarise the shape of the whole return distribution. This page ignores the distribution entirely and reports a single worst path. Where the value at risk calculator asks a forward-looking question about a threshold loss over one horizon, drawdown asks a backward-looking question about a path. All three are views of downside and none substitutes for another.
What This Max Drawdown Calculator Does
It walks the series once, tracking the highest value seen so far. At every observation it computes the decline from that running high. The deepest such decline is the maximum drawdown. Around that number it reports the peak that preceded it, the trough that ended it, the number of periods from peak to trough, the number of periods from trough back to a new high, and the gain required from the trough to restore the old peak.
It also reports the longest underwater stretch, which is a different quantity from the deepest one. A portfolio can have its worst percentage fall in one quarter and its longest period below a prior high somewhere else entirely.
The three bars rank the deepest separate drawdown episodes in the series. A single maximum can be a one-off event or the worst of several similar ones, and those two situations read very differently. If the second and third bars are close to the first, the series produces declines of that size routinely.
How to Use It
- Choose the input type first. Equity curve mode reads your numbers as levels. Returns mode reads them as percentages and compounds them from a base of 100, which is convenient when your data is a monthly return column rather than a balance column.
- Keep the series in time order and at a constant frequency. Mixing monthly and quarterly observations in one column produces duration figures that mean nothing, because a "period" no longer has a fixed length.
- Use the same frequency you plan to talk about. Drawdown measured on month-end values will almost always be shallower than drawdown measured on daily closes, because the daily series contains lows the month-end series never sampled. This is the single most common reason two people quote different maximum drawdowns for the same holding.
- Enter your own annualised return if you want the Calmar ratio. The page does not supply one. Calmar divides annualised return by maximum drawdown, and both halves have to come from the same period of history to mean anything.
- Read duration and depth together. The hero number is depth. The fourth grid item is duration. They answer different questions and either one alone is an incomplete description.
The Formula / How It's Calculated
Let Vt be the value at observation t. The running peak is the highest value observed up to and including that point:
Peakt = max(V0, V1, …, Vt)
The drawdown at each point is the shortfall against that peak, as a fraction of the peak:
DDt = (Vt − Peakt) ÷ Peakt
Maximum drawdown is the most negative of those values, conventionally reported as a positive percentage:
MDD = −min(DD0, DD1, …, DDn)
The gain required to recover from a trough back to the prior peak is not the same number as the drawdown, because the base has changed:
Recovery gain = Peak ÷ Trough − 1 = MDD ÷ (1 − MDD)
Worked example, matching the values the page loads with. The series is 100, 120, 90, 130, 110, 150, 120, 96, 140, 165. The running peak reads 100, 120, 120, 130, 130, 150, 150, 150, 150, 165. Drawdowns are therefore 0%, 0%, −25%, 0%, −15.38%, 0%, −20%, −36%, −6.67% and 0%. The most negative is −36%, at the value of 96, measured against the prior peak of 150. Recovery gain is 150 ÷ 96 − 1 = 56.25%, which also equals 0.36 ÷ 0.64. The peak sat at position 6 in the series and the trough at position 8, so the fall took 2 periods. A new high above 150 arrived at position 10 with the value 165, so recovery took a further 2 periods and the underwater stretch ran 4 periods in total. With a 12% annualised return entered, the Calmar ratio is 12 ÷ 36 = 0.33.
Why the Recovery Gain Is Always Larger Than the Fall
This is the piece of drawdown arithmetic people get wrong most often, and it is worth sitting with because the gap widens fast.
A 10% fall needs an 11.1% rise to recover. A 25% fall needs 33.3%. A 50% fall needs 100%. An 80% fall needs 400%. The reason is that the loss is measured against the old, larger base and the recovery has to be earned on the new, smaller one. Losing half your capital means the remaining half has to double, and doubling is a much rarer event than halving.
The practical consequence is that drawdown and return do not net out symmetrically. A series that gains 50% and then loses 50% ends at 75% of where it started, not level. The CAGR calculator captures this by compounding rather than averaging, which is why the compound annual figure for a volatile series is always below its arithmetic mean return.
This asymmetry is also why two portfolios with identical average returns can end up in very different places: volatility drag is the arithmetic gap between the arithmetic and geometric means, and it grows with dispersion.
Depth, Duration and the Difference Between Them
Maximum drawdown is a depth measure. Two more measures describe the same episode differently and both are reported here.
Time to trough is how long the decline took. A fall that unfolds over eighteen months feels different from one that happens in three weeks, even when the percentage is identical, and the slow one is the harder of the two to sit through because there is no obvious point at which it is over.
Time to recovery is how long it took from the trough to reach a new high. This is the figure that determines how long capital was unavailable in any practical sense. Historically, deep equity drawdowns have taken years rather than quarters to recover, and a plan that assumed a quick bounce would have failed on timing even when it was right about direction.
Underwater duration is the sum of the two — the whole stretch spent below a prior high. It is the honest answer to "how long was I behind", and it is usually much longer than people remember.
None of these tells you what happens next. A long underwater stretch is not evidence that recovery is due, and a quick recovery is not evidence that the next drawdown will also be short. They are descriptions of the past series you pasted in.
Why Data Frequency Changes the Answer
Maximum drawdown is path-dependent, and a series only shows the path at the points you sampled it. A daily series contains every intraday close; a monthly series contains twelve numbers a year. Any low that happened between two month-ends is invisible to the monthly calculation.
The direction of the bias is always the same: coarser sampling understates drawdown. It never overstates it. So a fund reporting maximum drawdown on monthly data and a trader reporting it on daily bars are not disagreeing — they are measuring different objects, and the daily figure will be larger. When comparing two track records, check the frequency before comparing the numbers.
Two further sampling issues matter. Drawdown computed on closing prices ignores intraday extremes, which can be substantial in volatile instruments. And a survivorship-filtered series understates drawdown by construction, because the series that failed outright are not in the sample.
What Maximum Drawdown Does Not Tell You
It is a single observation from one realised path, which makes it statistically fragile in ways a percentage sign hides.
It is a sample maximum, not a parameter. Extend the history and the maximum can only stay the same or get worse; it can never improve. A five-year track record with a small maximum drawdown may simply not have been alive during a bad stretch. Longer histories mechanically produce deeper maxima, so a like-for-like comparison needs equal-length windows.
It says nothing about frequency. One 30% fall and six 30% falls produce the same maximum drawdown. The three ranked bars on this page exist specifically to expose that difference.
It ignores everything above the peak. Two series with identical drawdown profiles can have wildly different returns. Drawdown is a risk statistic and must be read next to a return statistic, which is what the annualized return calculator provides.
It does not model leverage or forced selling. A leveraged position can be liquidated partway into a drawdown, which converts a paper decline into a realised loss and removes any chance of participating in the recovery. The mechanics of that are covered by the margin trading calculator. FINRA's guide to margin accounts states plainly that a firm can sell securities without contacting the holder first. FINRA's overview of investment risk covers the broader categories a single statistic cannot capture.
Revenue has peaks and troughs too. Arb Digital builds the acquisition side that shortens them — search visibility and content that keeps demand arriving when one channel stalls.
See Web Growth Services Talk to Arb DigitalCommon Mistakes to Avoid
- Comparing drawdowns measured at different frequencies — monthly data always produces a smaller maximum than daily data for the same holding, so the comparison is between sampling methods rather than between investments.
- Assuming the recovery gain equals the drawdown — a 36% fall requires a 56.25% rise, and the gap grows sharply beyond 50%.
- Comparing track records of different lengths — maximum drawdown can only worsen as history extends, so a shorter record is flattered automatically.
- Treating it as a limit on future losses — it is the worst outcome that happened in one sample, not a boundary on what can happen next.
- Ignoring duration — depth is what gets quoted, but the length of the underwater stretch is usually what determines whether a plan survives contact with reality.
Related Free Tools From Arb Digital
Pair this with the Sharpe ratio calculator for return per unit of volatility, the Sortino ratio calculator for downside-only deviation, and the value at risk calculator for a threshold-loss view over a single horizon. Use the standard deviation calculator on the same return column to see dispersion and the beta calculator for sensitivity to a benchmark. Everything else sits in the free online tools hub.
Frequently Asked Questions
It is the largest percentage decline from a running high to a subsequent low across a series of values. At every observation the calculator compares the value with the highest value seen up to that point, and the deepest of those shortfalls is the maximum drawdown. It is computed from history you supply and describes what already happened.
Because the two percentages use different bases. The fall is measured against the larger prior peak and the recovery has to be earned on the smaller trough value. A 36% fall needs 150 divided by 96, which is a 56.25% gain. The general relationship is drawdown divided by one minus drawdown.
Yes, and always in one direction. Coarser sampling understates drawdown because any low occurring between two sampled points is invisible. Monthly data will produce a smaller maximum than weekly, and weekly smaller than daily, for the same underlying holding. Compare frequencies before comparing figures.
Volatility measures dispersion of returns around their mean and treats gains and losses symmetrically. Maximum drawdown ignores the distribution and reports the worst realised path from peak to trough. Two series can share a standard deviation and have very different drawdowns, because drawdown depends on the order in which returns arrived.
It is an annualised return divided by maximum drawdown, using the annualised figure you enter. The calculator does not supply a return, because it has to come from the same history as the drawdown to be meaningful. A higher figure means more return per unit of worst-case decline over that specific period.
Yes. Switch the input type to returns and enter period percentages. The calculator compounds them from a base of 100 to build an equity curve and then measures drawdown on that curve. The resulting percentages are identical to what you would get from the underlying balances.
No. Maximum drawdown is a sample maximum from one realised path, and it can only get worse as history lengthens. A short track record with a shallow drawdown may simply not have included a difficult period. It describes the past and places no boundary on future outcomes.
It is the greatest number of consecutive periods the series spent below a prior high, counting from the peak until a new high is reached. It can belong to a different episode than the deepest drawdown. It is often the more useful figure, because duration rather than depth is what tests patience.
This tool performs a historical calculation on data you supply. It is not investment advice, not a recommendation to buy, sell or hold any security, and its output does not predict future losses. Past performance does not indicate future results, and decisions about your money should involve a licensed financial adviser regulated in your jurisdiction.
