What Is Maximum Drawdown? How to Measure It and Why Backtests Understate It

Most people look at a strategy and ask, "How much did it make?"

The better first question is, "How far did it fall before it got there?"

That second number is called maximum drawdown. It's one of the simplest risk numbers in trading, and one of the most ignored. This post covers what it is, how to work it out, and why the number in your backtest is usually better than what you'll actually live through.

What drawdown means

A drawdown is a drop from a high point in your account (or a strategy's results) to a later low point, before a new high is made.

Say your test account goes like this:

1. Starts at $1,000.
2. Climbs to $1,300. That's the peak.
3. Falls to $910.
4. Climbs back to $1,400. New peak.

The drop from $1,300 to $910 is a drawdown of $390. As a percent of the peak, that's 390 / 1,300 = 30%.

Maximum drawdown is simply the biggest of those drops over the whole period. If the 30% fall was the worst one, the maximum drawdown is 30%.

How to calculate it

You don't need fancy math. For each day (or candle):

1. Track the highest value so far. That's the running peak.
2. Work out how far today's value is below that peak, as a percent.
3. Keep the worst number you see.

In a spreadsheet, that's three columns: value, running peak, and (value − peak) / peak. The smallest number in the last column is your maximum drawdown.

Two things to watch:

• Use the same timeframe you trade on. Daily closes can hide a much deeper drop that happened during the day.
• Include fees. Fees make every losing streak a little deeper.

Why it matters more than the return

There are two reasons.

First, losses are harder to undo than gains are to make. The math isn't fair:

• Lose 10%, and you need about 11% to get back to even.
• Lose 20%, and you need 25%.
• Lose 30%, and you need about 43%.
• Lose 50%, and you need 100%.

The deeper the hole, the steeper the climb out.

Second, drawdown is where people quit. A rule you'll abandon halfway through a bad stretch isn't really your rule. If you'd panic and stop at a 25% drop, a strategy with a 40% drawdown is not a fit for you, no matter what the return says. We touched on this in Position Sizing for Beginners: the size you trade decides how deep the drop feels.

Why your backtest number is probably too kind

Here's the part most people miss. The maximum drawdown in a backtest is one path. It's the worst drop that happened to show up in that exact stretch of history. The future will be a different path, and over time the worst drop tends to get worse, not better.

It gets worse again when you pick strategies because they had a small drawdown. Some of that "smooth" look is just luck.

We ran a quick simulation to show it. We made 100 random strategies with no real edge: daily results were pure noise with no average gain, swinging about 2% a day. Each one got a 365-day "backtest" and then 365 new days it had never seen.

• Backtest: 44 of 100 net positive (simulated).
• Next 365 days: 41 of 100 net positive (simulated).
• The worst drop in the new period was deeper than in the backtest for 54 of 100 (simulated).

So far, about a coin flip. Now the important part. We picked the 10 strategies with the smallest backtest drawdown, the ones that would look the safest on a results page.

• Their median backtest maximum drawdown: 20.7% (simulated).
• The new period was deeper for 10 of 10 (simulated).
• Their median maximum drawdown in the new period: 39.7% (simulated).

None of them were actually safer. They were the luckiest paths. Picking the smoothest curve picked the luck, the same way picking the best return does. We go deeper on that trap in The Deflated Sharpe Ratio and Backtest Overfitting: Signs.

How to use drawdown honestly

1. Plan for worse than the backtest. A common habit is to assume the real drawdown could be one and a half to two times the backtest number. Size so you could live with that.

2. Look at how long it lasted, not only how deep. A 20% drop that takes two years to recover can be harder to sit through than a 30% drop that recovers in a month.

3. Shuffle it. Take the backtest's trades, shuffle their order many times, and look at the range of drawdowns you get. The original order is just one of many. See Monte Carlo Simulation for Trading Strategies.

4. Don't pick on smoothness alone. If you tried many versions and kept the one with the smallest drawdown, count how many you tried and expect it to get bumpier.

5. Watch it forward. Run the locked rule on new data in paper trading and compare the drawdown to the backtest's. If it blows past what you planned for, that's real information. See Backtest vs Forward Observation.

Where AI fits

AI tools like ChatGPT, Claude, Grok, or Cursor are great at writing strategy rules and running quick tests. Ask them to report maximum drawdown and how long it lasted every time, not only the return. And if you ask for "the version with the smoothest curve," remember the simulation above.

How Pickled Lizard tests ideas like this

Pickled Lizard is an AI trading and research platform. You bring the idea, or your own AI does, and we test it the same way we test everything:

1. Backtest on the full data set we have, with fees and next-candle timing.
2. Paper forward testing on new market data, with real fees and timestamps. The idea shows as "still collecting" while it builds a record.
3. Verified only when its forward results reach 80% of its own backtest trade rate and 80% of its backtest win rate.

Results show on the public paper trading leaderboard. Pickled Lizard isn't an exchange and doesn't hold your money. It's a non-custodial aggregator that connects to your own exchange accounts. Our servers are in Europe. The beta is free, and live trading is off during the beta.

Want to see how your idea holds up on new data?

Test it free at pickledlizardz.com/join

Quick recap

• Maximum drawdown is the biggest drop from a peak before a new high.
• Losses are hard to undo: a 50% drop needs a 100% gain to get back to even.
• In our simulation, the 10 smoothest-looking backtests all had deeper drops on new data: median 20.7% became 39.7% (simulated).
• Plan for worse than the backtest, check how long drops last, and watch it forward.

New here? Start with How to Backtest a Trading Strategy Without Fooling Yourself.