How to Buy an EA

What you are actually buying

When you buy an EA you buy a licence to run a program. You do not buy results. The program is a set of rules, often compiled so that you cannot read them. What you can check before paying is the evidence the seller offers and the terms of the sale. What you cannot check is what the rules will do next month.

That shapes the whole process. Treat the purchase like any due diligence: assume the claims are unproven until you have seen the evidence, and keep the cost of being wrong small. The cost of being wrong is the price, the running costs, and any trading losses while you find out.

The checks, in order

• Ask for a live track record in a format you can check. That means a statement or report from a real account that lists every trade with its open and close time, size, price and costs. A screenshot of a balance, or a curve, is not that. Ask which account type and which broker conditions it ran under, because a different spread or commission changes the result. If the seller shows only a back test, say so to yourself: it is a picture of the past with the settings fitted to it. See overfitting.
• Ask how long it ran and how many trades it made. A short record or a small count of trades tells you little, as the figure below shows.
• Ask for the drawdown and the largest lot. You need the worst dip and the biggest position it ever held. See max drawdown. If the lot grows after losses, read the martingale page first.
• Ask to test it on a demo account. A trial version, a demo licence or a time-limited key. A seller who will not allow a demo test is asking you to trust a picture. Even so, a demo is not a live account. See demo account.
• Read the refund terms. Is there a refund, for how long, under what conditions, and who decides. A refund that is only for "technical faults" does not cover an EA that works and loses. Get the terms in writing and keep a copy of the page as it stood on the day you paid.
• Read the licence terms. How many accounts can it run on, does it tie to one account number, does it expire, and what happens if the seller closes. Ask whether updates are included and for how long.
• Check the seller's claims for words that cannot be true. Promises of sure results, "no losing months", or no risk at all. If you see them, stop.
• Ask how it works in plain words, even if the code is closed: what triggers a trade, where the stop is, what the lot is. If the seller cannot say, you will not know when it is behaving wrongly.

Worked example: the cost you start behind by

A purchase has costs that arrive before the EA has traded once. This is a made-up case: a licence for $300 paid once and a hosted server at $25 a month, so $600 in the first year. See forex VPS for why a server is often used. It is a fixed cost whatever the EA does.

On $1,000 the EA must first earn 60.0 percent to cover what it cost to own. On $10,000 the same costs are 6.0 percent. Calculated by us on 29 Oct 2026, exact. These numbers are made up and ignore spread, commission and any trading loss, all of which come on top. The point is to know the hurdle before you buy. Prices and server costs vary widely.

What proof can and cannot show

What a good record can show

• That the EA traded on a real account over a stated period, and what it did there.
• The size of its losing runs in that period, if the full trade list is shown.
• Whether its claims, such as a stop on every trade, match its behaviour.

What it cannot show

• The future. A market that suited the rules for six months can stop suiting them.
• Your results. Your broker's spread, your commission, your server delay and your lot size differ. Costs are covered on the forex EA page.
• That the record is honest. An account that is shown may be chosen from many, and the ones that failed are not shown. This is survivorship bias. An independent, third-party verified record is stronger than one the seller supplies, but it still cannot predict.
• Anything about the rules you cannot see. A closed program can change its own behaviour only if the seller issues an update, but you cannot read why it trades.

If you decide to go ahead

Run it first on a demo account for long enough to see several of its trades match its claims. Start with the smallest size you can. Set a loss limit in advance and write down what result would make you stop. This is a way to limit the cost of finding out, not a promise it will work.