The Algomyth engine backtest has a worst drawdown of $5,273 across the whole period. It also publishes an average annual drawdown, $4,101, which is a good deal smaller. Both numbers are true, both come from the same equity curve, and if you read only one you will misjudge the system in one of two ways: blindsided by a fall you were not braced for, or quitting during an ordinary dip you thought was a disaster.
Max drawdown and average drawdown answer two different questions. Here is what each one tells you, why they diverge, and why the number you size your account around is always the bigger of the two.
A drawdown is the fall from an equity high to the low that follows it, before a new high is made. If that idea is new, start with what drawdown is. Any track record has many drawdowns, most small, a few large, and the two headline statistics summarize that collection differently.
The max is about survival. The average is about expectations. You need both, because a system that is survivable but constantly uncomfortable gets switched off just as surely as one that blows up.
You experience the average drawdown most of the time, but you have to survive the maximum one. So you set your expectations with the average and you size your account with the max. Confuse the two and you will either be surprised by the worst day or panicked by an ordinary one.
The worst drawdown of the period is the worst it got, full stop. For the published backtest that number is $5,273: the deepest the simulated equity curve fell from a high before recovering, on a mid-size account. It is a single event in four and three-quarter years of data, and the most important single figure for deciding whether you could have lived with the system.
A drawdown the strategy recovers from is still a drawdown your account has to survive. If your account, or a prop firm's rule, cannot withstand a repeat of $5,273, the fact that the strategy later recovered does not help you: the account was already closed. The max is the line the strategy came closest to failing at, so it is the line you plan against.
Average drawdown describes the everyday texture of the equity curve: the routine dips below a recent high that happen constantly and resolve quickly. In this backtest, and in most records, it is far shallower than the maximum. It is the number that tells you what a normal week or month feels like when nothing dramatic is happening.
This is the more honest guide to your day to day experience, and it is the one people ignore. They read the max, brace for it, then get rattled by a perfectly ordinary dip because they never formed a picture of normal. A system whose target sits at three times its risk produces small dips constantly as pure arithmetic. That is not the system breaking; it is the system working, and the tie between reward-to-risk, win rate and the shape of the equity curve is in risk, reward and win rate.
| Worst drawdown of the period | Average annual drawdown | |
|---|---|---|
| What it measures | Deepest peak-to-trough fall, once | Typical depth of a year's dips |
| the Algomyth engine backtest, mid-size account | $5,273 | $4,101 |
| The question it answers | How bad can it get? | What does a normal stretch feel like? |
| Use it to set | Account size and survival | Expectations and patience |
Change the account size and the contract count, and see how the modeled drawdown scales before you commit to anything.
Imagine the system with the same small average drawdown. One has a maximum of $4,000, the other $20,000. On a normal day they feel identical. But the second will, at some point, hand you a fall five times deeper than anything you were used to, and if your account was sized for the average, that is where it ends.
This is why the average, on its own, is dangerous: it is comfortable precisely because it excludes the event that matters most. Planning around the maximum is not pessimism; it is how you stay in the game when the average shows up again on the far side of the worst stretch. Most people who fail with a working system fail exactly here, which is the theme of risk and a recurring point in how much a trading bot makes.
The average tells you what most of your days look like. The maximum tells you what one of them will look like, and whether you will still have an account afterward.
Hypothetical performance. The $5,273 worst drawdown and every other figure here come from a backtest of the Algomyth engine over historical data on a mid-size account, not from a live account. Simulated results are prepared with hindsight, carry no financial risk, and cannot fully reflect real execution. Past performance, real or simulated, does not guarantee future results, and a future drawdown can exceed the largest one in the record.
Both numbers move with the sizing level, and they move together. Futures results scale almost linearly with size, and the drawdown scales alongside the profit. Move from Medium to High and you double the annual average, from $18,814 to $37,629, but you double both drawdown figures with it: the worst of the period goes from $5,273 to $10,546, and the average annual dip from $4,101 to $8,202.
That symmetry is the argument for starting small. Each published row belongs to one sizing level, so the figures that apply to you unchanged are the ones on the row you actually run. Move up a level and you have a deeper drawdown profile you never tested against your own nerves. Sizing is covered in position sizing for automated futures, and the profit side of the trade-off in profit factor explained.
If you trade funded capital, the maximum drawdown collides with a hard rule. A prop firm's trailing limit is often tighter than a strategy's own worst historical dip. If that limit is, say, $2,500, it is already below the $5,273 the backtest reached, which means a repeat of the deepest stretch would close the account before it recovered, even though the strategy went on to make money.
This is not a defect; it is the arithmetic of the funded model, and it is why accounts get consumed even by a working system. The trailing limit does not care about your average drawdown at all, only how far you fall from your peak, so the max is the only number that speaks its language. How that rule behaves, and how to size under it, is the subject of prop firm trailing drawdown explained.
When you look at any published record, ask for both numbers and be suspicious when you only get one. A record that quotes a small average and hides the maximum is showing you the comfortable half. A record that quotes a scary maximum with no sense of the average leaves you unable to tell normal from disaster.
Two more things belong beside the drawdown figures. First, recovery time: how long the equity curve stayed underwater before a new high, because a shallow drawdown that lasts a year is its own kind of hard. Second, the reward-to-risk geometry, because a target set at three times the risk tells you upfront that losing runs are normal and the drawdowns are features of the design, not signs it has stopped working. Together those describe the experience; the profit figure alone never does.
They matter for different jobs, so you cannot skip either. The maximum decides whether your account is sized to survive the worst historical stretch, and it is the number a prop firm's rule effectively enforces. The average decides whether your expectations match a normal week, so you do not quit during an ordinary dip. Plan your account size around the maximum and set your expectations with the average.
It is $5,273, measured on a mid-size account from January 2022 to September 2026. That is the deepest the simulated equity curve fell from a high before recovering. It is a hypothetical figure from a backtest, not a live result, and a future drawdown could be deeper. It is the single number to size against, because your account has to survive a repeat of it.
Roughly, yes, because futures results scale almost linearly with size and the drawdown scales with them. Moving from Medium to High takes the worst drawdown from $5,273 to $10,546 and the average annual one from $4,101 to $8,202. The profit doubles as well, but the pain arrives just as fast as the reward, which is why the report publishes all three levels rather than leaving you to guess.
Yes, and you should assume it can. A backtest records the worst fall that happened over the tested period; it cannot record a worse one that has not occurred yet. Live execution, slippage and market conditions the data never contained can all push a real drawdown past the historical maximum. That is why you size with headroom rather than sizing exactly to the backtested figure, and why past performance never guarantees future results.
Everything we write comes back to one idea: verify, then decide. The live terminal exists for exactly that.
Unlock Algomyth Live