Are Forex Robots Profitable? The Answer Depends on One Number Nobody Shows

Are forex robots profitable? Some are, verifiably, on public live records. But here is the answer that actually protects your money: a robot is never profitable or unprofitable on its own. The pair (robot + account size) is. The same verified, genuinely profitable EA makes one owner money and destroys another, and the difference between those two owners is a single number that almost no sales page will show you: how much account buffer the robot's worst stretch demands. This post gives you that number and how to compute it.

If you have bought a robot that "worked for everyone else" and still lost, this is probably what happened to you. Not a scam, not bad luck, not weakness: arithmetic that nobody put in front of you, because the vendor's incentive is to make the entry ticket look small.

First, the Honest Industry Math

Let us not pretend the skepticism is unfounded. Most retail forex robots are not profitable over meaningful periods, for reasons that are structural rather than accidental: strategies overfit to historical data, marketing selected for beautiful backtests rather than live edges, and recovery mechanisms (martingale and friends) that manufacture months of smooth profits before returning them all in one week. The graveyard is enormous and invisible, because failed robots get delisted and relaunched under new names.

And yet "most robots fail" does not answer your question, because you are not buying most robots; you are buying one. Verified live track records (public, third-party, months long, losses visible) do exist, and they prove the honest version of the claim: a minority of well-built systems trade a real edge, profitably, for years. The evaluation skill that separates the minority from the graveyard is covered in the full answer to whether expert advisors really work. Assume for the rest of this post that you found one of the real ones. Now comes the part that still goes wrong.

The Number: What the Worst Stretch Costs

Every strategy, including every genuinely profitable one, has a worst historical drawdown: the deepest peak-to-valley loss its live record has ever printed. That number is the strategy's bill. It has already been paid once by the vendor's own account, it will be presented again to yours, and the future version usually comes with a surcharge.

So the profitability question, correctly asked, is: when the bill arrives, can your account pay it and keep trading? Concretely:

1. Find the verified maximum drawdown. From the live track, never the backtest. If there is no live track, stop here; the answer to "is this robot profitable?" is "nobody knows, including the vendor."
2. Add a surcharge for the future. A stretch 1.5× the historical worst is a sober planning assumption. Records exist to be broken, and drawdown records break at the worst possible time.
3. Translate to money on YOUR account. At your intended lot sizing, what does that stretch cost in euros? Write the actual number down.
4. Apply the survival test. Could you watch your account sit that far underwater, for weeks, and leave the robot running? If the honest answer is no, your sizing (or the robot) fails, no matter how profitable the track is.

That fourth step is the one that decides real outcomes, because a profitable system only pays the owners who are still running it when the recovery comes. Switch it off at the bottom and you have converted a temporary drawdown into a permanent loss, and the robot's track record will go on being profitable without you.

Why High Win Rates Make This Worse, Not Better

Here is the cruel irony of robot shopping: the metric buyers use to feel safe (a high win rate) often marks the products where the hidden bill is largest. A 90%+ win rate EA feels unstoppable for months, precisely because its losses are rare, and rare losses are structurally larger when they arrive. Meanwhile a scruffy-looking 47% win rate system can be solidly profitable because its average win dwarfs its average loss; the win-rate math is short and worth internalizing before you compare anything.

The point is not that high win rates are scams (some of the most robust systems I run are high win rate, and I say so with their full drawdown history in public view). The point is that the win rate tells you how often you will feel good, while the drawdown tells you whether you will survive, and only one of those pays your bills. Vendors advertise the first number and bury the second; "low drawdown" claims deserve their own forensic reading.

What Profitable Actually Looks Like, in Public

So that this is not theory: my own AI system's live public record currently reads +36.59% over 185 days, profit factor 1.24, 264 trades, and a maximum drawdown of -16.25%. Run the method on it, right now, as practice. Historical worst: -16.25%. Sober planning assumption: call it -25%. On a 10,000 account at the published risk settings, that is 2,500 underwater at the worst point, possibly for weeks. Can you leave it running through that? Then the profitability is available to you. Cannot? Then this profitable robot is unprofitable for you at that size, and the honest moves are a smaller size or a pass.

That is the entire framework, applied in one paragraph, by a vendor, against his own product. Demand the same arithmetic from everyone who wants your money.

The number, always in view.

Alpha Pulse AI publishes its full live record (the growth and the -16.25% worst stretch) on a verified Myfxbook, because you cannot size an account around a number you are not shown. The curve you want, with the bill attached. That is the whole deal.

The Checklist Before You Believe Any "Profitable Robot"

• Live, verified, six-plus months. No live track, no conversation. The 10-minute audit covers the badges and traps.
• Worst drawdown located and translated into your money at your sizing, with the 1.5× surcharge.
• Survival test passed in writing. "I accept being X euros underwater without intervening." If writing it feels absurd, the size is wrong.
• Profit engine understood. If you cannot say whether it wins by frequency (high win rate, rare big losses) or by asymmetry (frequent small losses, bigger wins), you cannot anticipate its bad weeks.
• Recovery logic identified. If losses are being averaged, gridded, or doubled away, the drawdown you see is a loan, not a cost. Read the track's floating drawdown, not just the closed one.

The Honest Close

Are forex robots profitable? The real ones are, for the owners who sized for the worst stretch and held through it. The rest of the industry's tragedy is not fake robots taking money from fools; mostly it is real enough systems, bought at sizes that could never survive their own arithmetic, switched off at the exact bottom by owners who were never shown the bill in advance.

Now you have the bill format. Ask for the number. Any vendor who will not show it has answered your question, and any vendor who shows it proudly (drawdown first, returns second) is speaking the only language in this industry that deserves your attention.

Weekly live numbers from my own systems, worst stretches included, are in the DoItTrading newsletter. It is what "profitable" looks like without the airbrush.

Frequently Asked Questions

Are forex robots actually profitable in 2026?

A minority are, and public verified track records prove it: months of live trading, third-party tracked, with losses and drawdowns visible. The majority of retail robots show no verified live edge. But profitability is never the robot's property alone: it belongs to the pair of robot plus account size, because a profitable system on an undersized account still bankrupts its owner during a normal worst stretch.

How much money can a forex robot make per month?

Any specific number would be an invention, and vendors who promise one are telling you what you want to hear. Real systems produce lumpy returns: strong months, flat months, and losing stretches, which is why honest evaluation reads the verified drawdown before the average return. As a calibration point, a live system returning +36.59% in about six months also spent part of that period -16.25% underwater. Both numbers are the truth.

What is a good maximum drawdown for a forex robot?

There is no universally good number; there is only compatibility with your account and nerves. Verified live systems commonly print worst drawdowns between -10% and -35%. The working rule: take the verified historical worst, assume the future exceeds it by half again, translate that into money at your sizing, and only trade the robot if you can honestly commit to not intervening at that depth. A "good" drawdown is one you have pre-accepted in writing.

Why do profitable robots still lose money for their buyers?

Two mechanisms: accounts sized so the robot's normal worst stretch forces a margin problem or a panic stop, and owners switching the system off mid-drawdown, which converts a temporary loss into a permanent one right before the recovery the edge eventually delivers. Both trace to the same root: the buyer was sold the return curve without the drawdown bill, so they never consented to the real cost of earning it.

Backtest shows 500% per year. Is that real?

Treat it as advertising, not evidence. Backtests are simulations the seller controls: optimized against known history, often with unrealistic spread and slippage assumptions, and immune to the execution frictions that tax live trading. The conversion rate between spectacular backtests and live profitability is dismal. The only numbers that count are on a third-party-verified live account, and any vendor leading with backtest returns instead is making a choice.