# Max drawdown: how deep can a good strategy fall?

> Max drawdown is the deepest fall of an equity curve from a peak. Why a backtest's figure is one draw of many, and how far a good strategy can really fall.

Published October 13, 2026 · updated October 18, 2026 · Backtesting · by the EdgeLuma team · 8 min read

Source: https://edgeluma.com/blog/max-drawdown

Maximum drawdown is the deepest fall of an equity curve from a peak to the low that follows it, before a new peak is made. A backtest shows one maximum drawdown, the one its trades happened to produce in that order, and the same strategy can fall a good deal further.

The backtest says the worst stretch was 9R. You size the account so that 9R wouldn't hurt, and three months in you're down 13R on a strategy that is doing exactly what the backtest said it would. Nothing has broken. The 9R was the worst fall of one particular order of trades, and the market never replays trades in the same order.

This guide explains what maximum drawdown measures, then shows with simulations how much of it is the luck of the order and how it grows the longer a strategy trades. It covers how to estimate the drawdown you should plan for, what a drawdown in R means for an account in money, and how EdgeLuma reports it.

## What is maximum drawdown?

**Maximum drawdown** is the largest drop of an equity curve from a peak to a later low, before the curve makes a new peak. It's counted in the units of the curve. In R (R is the amount risked on one trade) it says how many risk units the strategy gave back, and in money how much of the account.

To find it, follow the curve trade by trade, keep track of the highest point so far, and measure how far the curve sits below it after every trade. Plotted on its own, that distance is the underwater chart, and its deepest point is the maximum drawdown. Next to the depth, two durations are worth knowing: how long the fall lasted, and how long the curve took to get back to its peak.

*Figure: An equity curve of 100 trades in R with its running peak, and under it the drawdown after every trade. The deepest point of the underwater chart is the maximum drawdown.*
## Why is a backtest's drawdown only one draw?

Because a drawdown depends on when the losses come, and the order of trades in a backtest is an accident of history. The same set of wins and losses, in a different order, produces a different drawdown.

We took one set of 100 trades, 40 wins of +2R and 60 losses of 1R, worth +20R in total, and shuffled their order 20,000 times. The maximum drawdown ranged from 4R to 25R. The middle of the range was 9R, and 90 shuffles in 100 fell between 6R and 15R. Every shuffle ended at the same +20R.

*Figure: The same 100 trades in three orders. The total is +20R every time; the maximum drawdowns are 6R, 9R and 18R.*
Your backtest is one of those orders. If it happened to be a kind one, its drawdown understates what the same edge can put you through.

## Does drawdown get worse the longer you trade?

Yes. The more trades a strategy takes, the more chances it gets to run into a bad streak, so its deepest drawdown keeps growing, more slowly the bigger the edge. We simulated 20,000 runs of a strategy that wins four trades in ten at +2R and loses 1R on the rest, an edge of +0.2R a trade:

| Trades | Typical max drawdown | Deepest in 1 run in 10 | Deepest in 1 run in 20 |
| --- | --- | --- | --- |
| 100 | 10R | 16R | 19R |
| 200 | 12R | 20R | 23R |
| 500 | 16R | 25R | 28R |
| 1,000 | 20R | 29R | 32R |

The mathematics agrees. For a random walk with a positive drift, Magdon-Ismail, Atiya, Pratap and Abu-Mostafa showed that the expected maximum drawdown keeps growing with time, roughly with its logarithm, in [On the Maximum Drawdown of a Brownian Motion](https://www.cambridge.org/core/journals/journal-of-applied-probability/article/abs/on-the-maximum-drawdown-of-a-brownian-motion/F9E3B8A454B020DDEBF0AC3390EF7807) (Journal of Applied Probability, 2004). A backtest of 200 trades tells you about the drawdowns of 200 trades. Trade the strategy for 1,000 and you should expect deeper ones. How many trades a backtest needs before its numbers mean anything is a question of its own, worked out in our guide to [how many trades a backtest needs](https://edgeluma.com/blog/how-many-trades-backtest).

## How do you estimate the drawdown to expect?

Reshuffle the backtest's own trades and look at the whole range instead of the one number. It takes four steps:

1. Take the backtest's trades in R, after costs.
2. Shuffle their order many times, a [Monte Carlo simulation](https://en.wikipedia.org/wiki/Monte_Carlo_method), and record the maximum drawdown of each shuffle.
3. Read a high value from the results, such as the drawdown that one shuffle in twenty exceeds. That's a planning number, not a worst case.
4. Size each trade so that a fall of that size, in money, is one you would actually sit through.

Reshuffling assumes every trade is independent of the one before. Real losing trades tend to cluster in the same bad stretch of market, so a real drawdown can go deeper than the shuffles suggest. Resampling with replacement, which draws trades at random and can pick the same one twice, also varies the mix of wins and losses, and usually gives a wider range again.

## What does a drawdown in R mean in money?

In R, a drawdown is a count of risk units, and in money it depends on how much each unit is worth. Risking a hundredth of the account per trade, a 20R drawdown takes away about a fifth of it. At two hundredths per trade, the same 20R takes about two fifths.

Getting back takes more than the fall did. An account that drops by a fifth needs to gain a quarter to return to its peak, and one that halves needs to double. That asymmetry is why the risk per trade matters as much as the strategy's rules: it decides how deep the same drawdown in R cuts into the money.

## How are drawdowns and losing streaks related?

A drawdown is mostly a losing streak with a few wins scattered through it, so strategies with long streaks have deep drawdowns. That's why a low win rate costs you in drawdown even when the expectancy is good: a strategy that wins three trades in ten will usually lose about ten in a row somewhere in a hundred trades. Our guide to [win rate and expectancy](https://edgeluma.com/blog/win-rate-vs-expectancy) measures those streaks, and the guide to [fees and slippage](https://edgeluma.com/blog/backtest-fees-and-slippage) explains why costs make every one of them deeper.

## How EdgeLuma reports drawdown

[EdgeLuma](https://edgeluma.com/) reports the maximum drawdown at the top of every backtest:

- In R, next to the expectancy and the win rate, with the longest losing streak under it.
- In money, in the same run read as an account, with the drawdown in currency and as a share of the peak.
- Two readings of the money: fixed, where every trade risks the same amount, which is what the engine runs, and compounded, the same trades re-sized as the account grows and shrinks. The two give different drawdowns in money from the same trades.

The drawdown is measured on closed trades, and how far an open trade went against you is in the trades list, as each trade's MAE (maximum adverse excursion). EdgeLuma doesn't reshuffle the trades for you: the drawdown in the report is the one the trades produced in the order they happened, so treat it as one draw and plan for a deeper one. A backtest replays the past: it doesn't tell you what to trade, and nothing in a report is financial advice.

## Key takeaways

- Maximum drawdown is the deepest fall of an equity curve from a peak to the low that follows it. In R, it's the stretch of losses you would have had to sit through.
- A backtest shows one drawdown: the one its trades produced in the order they happened. The same 100 trades, reshuffled, fell anywhere from 4R to 25R.
- Drawdowns grow with time. A strategy with an edge of +0.2R a trade typically falls about 10R at some point in 100 trades, and about 20R in 1,000.
- Plan for a deeper fall than the backtest shows: reshuffle its trades, read the fall you get in one shuffle in twenty, and size each trade so you could sit through it.
- In money, the same drawdown in R hurts more the more you risk per trade, and climbing back takes a bigger gain than the fall itself.

## Frequently asked questions

### What is a good max drawdown?

There's no number that suits every strategy. Judge the drawdown in R against the edge and the number of trades, and in money against what you would really sit through without abandoning the strategy. A drawdown that would make you quit is too deep for you, whatever the rest of the report says.

### How is maximum drawdown calculated?

Follow the equity curve trade by trade, keep track of its highest point so far, and after every trade measure how far the curve sits below that peak. The largest of those distances is the maximum drawdown.

### Is max drawdown the same as the largest loss?

No. The largest loss is a single trade. A drawdown adds up a whole stretch of trades, wins included, from a peak down to the lowest point before the next peak, so it's usually many times bigger.

### Why is my live drawdown deeper than my backtest's?

Usually because the backtest showed one order of trades and live trading dealt a different one, and because drawdowns grow the longer you trade. Costs the backtest left out, and a market that has changed, can make it deeper still.

### How can I reduce drawdown?

Risking less per trade shrinks the drawdown in money, though not in R. Rules that cut the drawdown in R are possible too, but test them on data you didn't tune on: rules fitted to dodge the past's bad stretches often walk straight into the next ones.