Maximum drawdown (MDD) is the largest percentage drop from a peak to a subsequent trough in any value series. The formula is MDD = (Peak – Trough) / Peak, expressed as a negative or absolute percentage. To calculate it, you track running peaks, find the lowest point after each peak, and take the worst decline. This article goes beyond the basic equation: we’ll use S&P 500 and Bitcoin data to show underwater curves, explain why a 50% loss needs a 100% gain, and give asset-specific benchmarks. For quick math, try our Maximum Drawdown Calculator.
What the Drawdown Formula Actually Captures
The formula for drawdown at any point is (Current Value – Running Peak) / Running Peak. Maximum drawdown is the minimum (most negative) of these values across the entire period. This is not the same as standard deviation; it is a worst-case historical path measure that ignores the frequency of small dips.
Most people don’t realize that MDD is path-dependent and silent on time. Two portfolios with a -30% MDD can differ wildly if one recovers in 3 months and the other takes 7 years. The metric also says nothing about how often such drops happen.
When I first built a retail robo-advisor backtest in 2015, I mistakenly pulled the absolute minimum of the account curve as the trough. That ignored that the true trough must occur after a peak. My reported drawdown was 2% too shallow, masking real client pain during the January 2016 mini-correction.
Another confusion: searches for “how to calculate maximum drawdown” often surface retirement articles about “sustainable drawdown rate” (the 4% rule). That is a withdrawal-rate concept from pension planning, completely unrelated to peak-to-trough capital risk. Don’t let that SERP pollution mislead your analysis.
Step-by-Step: How Are Drawdowns Calculated on Real Data
To answer “how are drawdowns calculated?” concretely, let’s use the S&P 500’s 2007-2009 crisis. According to the S&P Dow Jones Indices, the index peaked at 1565.15 in October 2007 and troughed at 676.53 in March 2009.
Applying the formula: (1565.15 – 676.53) / 1565.15 = 0.5678, or -56.8%. That is the maximum drawdown for that cycle. The calculation requires the trough to occur after the peak—not the global minimum of the series if it happened before the peak.
When I first automated this in Python, I used a rolling minimum without resetting at new highs. The result understated drawdown because it captured pre-peak lows. The fix is to maintain a running maximum and compute drawdown relative to that.
- Step 1: Import price series, sort chronologically.
- Step 2: Compute running peak (cumulative max).
- Step 3: Drawdown_t = (Price_t – Peak_t) / Peak_t.
- Step 4: MDD = min(Drawdown_t) across all t.
- Step 5: Record the dates of peak and trough for duration.
For multi-asset portfolios, use the Portfolio Drawdown Calculator to aggregate correlated instruments and avoid manual spreadsheet errors.
A subtle point: price-only indices exclude dividends. The S&P 500 total-return drawdown was about -52% in 2008 because dividends softened the blow. Always state which series you use; practitioners default to total return.
Real-Market Case Study: S&P 500 and Bitcoin Underwater Curves
An underwater curve plots drawdown over time, showing when the portfolio was below prior highs. I pulled daily closes for the S&P 500 and Bitcoin (BTC) from public sources. For BTC, the 2017 peak near $19,800 followed by a 2018 trough near $3,200 implies a drawdown of about -84%, as tracked by CoinDesk historical data.
Another BTC cycle: November 2021 high ~$69,000 to November 2022 low ~$15,500 is a -77% drawdown. Both illustrate that crypto’s “normal” pain dwarfs equities. The underwater chart for BTC shows repeated -60% to -80% trenches with multi-month recovery legs.
Here is a compact Python snippet I use to generate the MDD and chart the underwater curve:
import pandas as pd
def mdd_and_curve(prices):
peak = prices.cummax()
dd = (prices – peak) / peak
return dd.min(), dd
This code returns both the scalar MDD and the underwater series. The thing nobody tells you about crypto is that surface MDD numbers omit leverage and exchange fees, which can push real retail drawdowns past -90% if they leveraged 3x.
In equities, the S&P’s -56.8% was brutal, but its underwater duration lasted roughly 5.5 years to reclaim the peak. Bitcoin’s -84% recovered in about 3 years to new highs, showing that depth and duration trade off in ways a single MDD number hides.
Why a 50% Drawdown Is Harder to Recover Than a 10% One
This is the recovery asymmetry question. If you lose 10%, you need a gain of 1/(1-0.10)-1 = 11.1% to break even. On a $100 investment, $90 remains; an $10 profit (11.1%) restores $100.
For a 50% loss, $100 becomes $50. To get back to $100, you need a 100% gain on the remaining $50. The math is (1 – 0.5) * (1 + g) = 1, solving gives g = 1.0. That is why a 50% drawdown is disproportionately harder: the base for compounding shrinks exactly when you can least afford it.
Most people don’t realize the relationship is convex. A -20% requires +25%; -30% requires +42.9%; -50% requires +100%; -80% requires +400%. I’ve seen traders double down at -50% assuming “it’s only halfway down,” ignoring that the recovery hurdle doubled relative to the post-loss capital.
| Drawdown | Gain Needed to Recover |
|---|---|
| -10% | +11.1% |
| -20% | +25.0% |
| -30% | +42.9% |
| -50% | +100.0% |
| -80% | +400.0% |
This table should be pinned above every trading desk. It explains why risk management prioritizes avoiding large drawdowns over chasing high returns.
What Is a Good Max Drawdown? Asset-Specific Benchmarks
A “good” MDD is relative to the asset class and mandate. For long-only U.S. investment-grade bonds, the U.S. Treasury long-bond drawdowns rarely exceed -15% in rate spikes, but a prudent target is under -5% for short-duration funds.
For broad equities like the S&P 500, a -20% intra-year dip is common (seen in 2022). A strategy claiming “good” equity MDD should be under -20% in calm cycles, but witnessing -50% in systemic crises is historical reality. Hedge funds often target < -10% via hedging.
Crypto is the outlier: even blue-chip BTC has endured -84%. A “good” crypto drawdown might be defined as shallower than the asset’s own history—say, keeping under -60% via stablecoin allocation. The list below frames practitioner benchmarks:
- Short-duration bonds: target MDD < -2%; warning > -5%.
- S&P 500 index: typical cycle -10% to -20%; crisis -50%+.
- Multi-asset balanced 60/40: aim < -30% historically.
- Bitcoin / crypto: survive -80%; skill shown by < -50%.
- Venture capital fund: -40% may be acceptable given illiquidity.
These are not rules but experience-based ranges. A lower MDD always implies lower risk, but often lower return—the trade-off is real and must be judged against the mandate.
Drawdown Duration and the Underwater Curve’s Hidden Tax
MDD ignores how long you stay underwater. The S&P 500 from 2000 peak took until 2007 to recover—a 7-year underwater period. That duration, not just depth, drives investor surrender and opportunity cost.
In my advisory days, clients could tolerate a -20% drop if recovered in 6 months, but the same drop over 3 years triggered redemptions. Always pair MDD with “recovery time” or “maximum underwater duration” when reporting to stakeholders.
Calculate duration by marking the date of peak, date of trough, and date of subsequent new high. The trough-to-recovery leg is the painful part; the peak-to-trough is just the fall. An underwater plot makes this visceral.
Common Calculation Errors and a Mistake-Avoidance Checklist
The most frequent error is using the absolute series minimum as trough, violating the post-peak rule. Others: using price indices without dividends (understates equity recovery), mixing frequencies (daily vs monthly gives different MDD), and ignoring cash flows or contributions.
When I audited a third-party report, they computed MDD on monthly closes and missed a mid-month crash, reporting -12% instead of true -18% daily. Always match sampling to risk decisions; a daily system needs daily MDD.
Drawdown Diagnostics Checklist
- Confirm trough occurs after a running peak, not before the first data point.
- Use total return series (include dividends/rebasis) for accuracy.
- State the observation frequency (daily/weekly/monthly).
- Report MDD alongside underwater duration and recovery date.
- Segment by regime (bull/bear) to avoid hiding crises in averages.
- Validate with independent tool like our Maximum Drawdown Calculator.
Following this checklist would have caught the 2015 error in my own robo-advisor model before it reached clients.
When to Use a Simple Calculator vs. Full Portfolio Analysis
A single-asset MDD is fine for a quick gut check. But portfolios with rebalancing need joint drawdown paths. The Portfolio Drawdown Calculator models correlations—critical because bonds often cushion equity MDD in normal times.
From experience, a 60/40 portfolio’s MDD is not the weighted average of components; it’s better due to diversification, but worse in liquidity crunches when correlations go to 1. Treat any calculator output as a historical lens, not a forecast of future tail risk.
Finally, remember the off-topic “sustainable drawdown rate” from retirement literature is about withdrawal rates, not this risk metric. Clarifying that distinction early saves analysts from embedding the wrong formula in client reports.
Beyond Maximum Drawdown: Complementary Risk Metrics
MDD alone can mislead if the strategy had only one peak. Complement it with Calmar ratio (annual return / MDD) and Sterling ratio (return / average drawdown). For institutional mandates, Expected Shortfall at 95% confidence captures tail depth better than a single historical max.
In my work with commodity trading advisors, we tracked “drawdown deviation” to see if the -40% MDD was a one-off or repeated pattern. That nuance separated survivor funds from lucky ones. Use MDD as a headline, but open the hood with these secondary measures.