Two systems return 20% a year. The first never sits more than 3% below its own high-water mark. The second gets there after eleven months with the account 40% underwater.
On a year-end statement they look identical. In real life almost nobody survives the second, because eleven months is long enough to decide the thing is broken, switch it off, and miss the recovery. The number that separates them is drawdown, the most useful figure on any performance report. It is more useful than the return, by a wide margin.
The definition, with nothing added
Drawdown is the drop from a peak in your account equity to the trough that follows it, before a new peak is made. That is the whole definition.
Say an account runs $50,000, then $53,000, then $51,000, then $49,500, then back to $54,000. Peak $53,000, trough $49,500, drawdown $3,500. Once equity passes $53,000 the clock resets. Maximum drawdown is the deepest of those episodes across the history.
Two things follow that people get wrong constantly. Drawdown is measured against your own equity high, not your starting balance, so an account up 40% on the year can still be in a painful one. And it is measured on equity, so a report built on closed trades understates what you watched on the screen.
The one thing to remember
Return tells you what a system paid. Drawdown tells you what it demanded before it paid. Almost every failure with an automated system is a person quitting inside a drawdown the system had already survived before, in the backtest they read and believed.
Dollars or percent: you need both
A drawdown in dollars is meaningless without the account size. A drawdown in percent is meaningless when a fixed dollar limit sits underneath it, which is exactly the situation on a funded account. Rentabilio's published backtest shows a maximum drawdown of $4,379 on a $50,000 account, which in percent is 8.8%. Both numbers do work:
- 8.8% tells you the shape of the ride: the worst stretch in more than seven years of simulated trading took less than a tenth of the account. A normal person can sit through that.
- $4,379 tells you whether the account survives at all. On a funded evaluation with a hard dollar threshold, the percentage is irrelevant and the dollar figure is the only thing the rulebook reads.
A third framing: the backtest averages roughly $2,900 net per month, so its worst drawdown is about a month and a half of average profit.
Max drawdown is not the number you will experience
Maximum drawdown is one observation: the worst thing that happened once, in a specific sequence of history. It says almost nothing about a normal bad week.
Average drawdown, the mean depth across every peak-to-trough episode in the record, describes your actual life with a system. It is almost always far smaller than the maximum, and it is the one to check your tolerance against, because you will meet it dozens of times a year and the maximum maybe once.
The relationship between them is informative too. If the maximum is three or four times the average, the system has a fat left tail: it mostly bleeds a little and occasionally opens a hole. If the maximum is barely above the average, either the system is well behaved or the sample is too short to have found its bad day. Seven years and 4,557 trades is long by retail standards. It is still not proof; see why beautiful backtests die in live trading.
The maximum drawdown in any backtest is simply the worst one that has happened so far.
Time underwater is the part that actually breaks people
Depth gets all the attention. Duration does the damage.
Time underwater is how long the account spends below a previous high before making a new one. A 6% drawdown that recovers in nine days is a non-event. The same 6% grinding sideways for five months is what gets systems switched off. That is long enough for doubt to become a decision.
Nobody quits on the day of the biggest loss. People quit on week fourteen of nothing happening, when the story in their head has moved from "this is a rough patch" to "this used to work and now it doesn't." So look for two duration figures alongside the drawdown: longest time underwater and losing months. A seller who publishes a max drawdown but no recovery time has shown you the easier half of the picture.
The arithmetic underneath is not mysterious. Rentabilio wins 46.2% of its trades, so it loses 53.8%. Six losses in a row is 0.538 to the sixth power, or 2.4%, which across 4,557 trades is roughly fifty occurrences. Ten in a row comes up about four times. Those runs are not malfunctions; they are what a 46.2% win rate looks like from the inside, and the full calculation is in risk-reward and win rate.
On a funded account, drawdown works differently
This is the section that costs people real money, and the reason a comfortable-looking figure can still be fatal. In your own account a drawdown is just a smaller balance: nothing happens, and you decide whether to keep going. On a prop-firm evaluation, drawdown is a rule with a switch attached, and hitting it does not reduce your account. It ends it.
Worse, most firms do not measure it the way you just learned. They use a trailing threshold: a floor starting a fixed distance below your balance that follows your equity up as you make money and never comes back down. Many stop trailing once the floor reaches your starting balance. Figures and mechanics vary by firm and account size and they change, so read the rulebook of the account you bought.
| Type of limit | Measured against | What happens when you hit it |
|---|---|---|
| Ordinary account drawdown | Your own equity peak | Nothing formal. You have less money and a decision to make |
| End-of-day trailing | Highest closing balance so far | Account fails. Evaluation fee is gone |
| Intraday trailing | Highest equity ever touched, unrealized included | Account fails, and profit you never banked counted against you |
| Daily loss limit | That day's opening balance | Account fails or locks for the day, however profitable you are overall |
The intraday version deserves a worked example, because it is genuinely counterintuitive. Take a $50,000 account with a $2,500 trailing threshold, so the floor starts at $47,500. You open a trade, it runs $800 in your favor, and the floor follows your equity up to $48,300. Then price comes back and you are stopped out for a $200 loss. Realized result: minus $200. Remaining buffer: $1,500, down from $2,500.
A $200 loss consumed $1,000 of your allowance. Do that four times and the account is gone despite being down $800 in real money. That is the mechanic behind most "the system was fine but my account blew up" stories, and it has nothing to do with the system.
What Rentabilio's $4,379 means on a funded account
Now put the two numbers next to each other, because this is the part a sales page skips. The backtested maximum drawdown is $4,379. A typical 50k evaluation carries a trailing threshold around $2,500. The system's worst historical stretch does not fit inside the account. If that drawdown repeats on a 50k evaluation, the account fails. Not maybe.
That is not a hidden flaw. It is the economics of trading funded capital, and it is why the published figures include a line most vendors would never print. In the last 7 months of the backtest the system produced $43,322 gross, roughly $41,000 net and about double its historical monthly average. Getting through that stretch consumed about seven funded accounts. At roughly $100 per 50k evaluation, around $700 of fees against about $41,000 of hypothetical net profit.
That is the trade you are making: you accept that accounts will die on drawdown rules and treat the evaluation fee as a cost of doing business rather than a failure. It works because the fee is small and capped and the capital at risk is not yours. If you would rather not lose $100 several times a year, funded capital is the wrong vehicle, and better to know now. The mechanics are in how funded accounts actually work and on the funded capital page.
Hypothetical performance. The $4,379 drawdown, the $43,322 stretch and every figure here come from a backtest, not a live account. Simulated results are prepared with hindsight, carry no financial risk, and cannot fully reflect real execution or slippage. Past performance is not indicative of future results.
A backtested drawdown is the smallest one you will ever see
Assume any historical simulation is optimistic, this one included. A backtest fills your stop at your stop price; a live market at 8:30 AM ET on a data release does not always agree. A maximum drawdown is the worst episode in this sample, and adding data has never made one smaller. And the person is not modeled: the backtest never skipped a day or doubled size to catch up. So budget as though a real bad stretch could be meaningfully deeper than $4,379, which is the reasoning behind the risk page.
The drawdown is on the same page as the profit
Depth, losing months, the trade list and the steps to reproduce it in your own NinjaTrader 8. Then put your account size against it in the calculator.
Why publishing a drawdown is a credibility signal
Drawdown is the number a dishonest seller has every incentive to leave out: the only figure on a performance report that makes the product look worse, and the one a buyer needs most. So a specific, unrounded figure ($4,379, not "under 10%") tells you something before you have checked anything. It says the seller ran the full report rather than cropping a screenshot, and is prepared to be held to it.
It is still a claim, not a proof. The point of a reproducible backtest is that you can run it yourself in the Strategy Analyzer inside NinjaTrader 8 and check that the drawdown you get is the one you were shown. That is the difference between a number you verify and a number you trust, and the whole argument of what a trading bot actually is.
How to use the number before you buy
- Compare it to the account rule, not the profit. Not "is this drawdown small?" but "does it fit inside the threshold on the account I intend to trade?" Often it does not, and the answer is a bigger account or fewer contracts.
- Scale it with contracts. Two contracts means roughly double the dollar drawdown with the same rulebook underneath. Sizing up multiplies the pain as fast as the profit.
- Decide your quit rule now. Write down the drawdown at which you would conclude the system has stopped behaving like its history. A number chosen in advance is a decision; one chosen during a bad week is a reaction.
Frequently asked questions
What is a good maximum drawdown?
There is no universal threshold, because it only means something relative to the return, the account size and the account rules. A more useful test is the ratio of net profit to maximum drawdown, sometimes called the recovery factor, plus whether the drawdown fits inside the limits of the account you plan to trade. A modest drawdown you can sit through beats a better return you abandon in month three.
What is the difference between drawdown and a losing streak?
A losing streak counts consecutive losing trades. A drawdown measures money lost from a peak, and needs no consecutive losses at all: alternating small wins and larger losses produce a drawdown with no streak in it. Streaks explain part of a drawdown, but the drawdown determines whether an account survives.
Why is trailing drawdown harder than it sounds?
Because the threshold follows your equity upward and never comes back down, and on many accounts it follows unrealized profit too. An open trade that moves in your favor and then reverses can consume far more of your allowance than the eventual loss suggests. Read whether your account trails on closing balance or intraday equity; the two behave very differently on the same trades.
Should I turn a system off when it hits its maximum drawdown?
Exceeding a historical maximum is a reason to look closely, not an automatic stop. What matters is whether the system still behaves like itself (similar trade frequency, similar average win and loss, similar hit rate) or whether something has changed. Decide the rule in advance and in writing; the worst time to invent one is inside the drawdown that prompted the question.
In short: drawdown is the drop from an equity peak to the trough that follows, and it decides whether you are still trading a year from now. Read it in dollars and in percent, look for the average as well as the maximum, and insist on the duration figures, because time underwater is what makes people quit. On a funded account it stops being a statistic and becomes a switch: Rentabilio's backtested $4,379 on a $50,000 account is larger than a typical 50k evaluation threshold, which is why the record shows about seven accounts consumed in seven months.
