How to Choose an EA

Choosing an expert advisor is mostly an exercise in scepticism. The sales page is built to make the decision for you; your job is to slow down and check the few things that actually predict whether it will still be making money in six months. Here is a practical checklist, and the red flags that should end the conversation before you reach for your card.

What to look for

A verified live track record — real money, real trades, independently verified. Not a backtest, not a demo. This is the single most important filter, and most EAs fail it.
Length of live history — a few months proves little. Look for a year or more that trades through different market conditions, not one lucky run.
Maximum drawdown and worst month — how far down did it go, and how long did it stay there? If the vendor will not show you the worst period, assume it was bad.
Realistic returns — 3–8% a month with visible drawdown is plausible. 20% a month on a smooth curve is a warning, not an opportunity.
Strategy type — know what it does. Grid and martingale systems show lovely curves right up until they hand it all back. If you cannot tell how it makes money, treat that as a no.

The rest of the checklist

Vendor transparency — a named seller who explains the strategy, shows live results and answers hard questions beats an anonymous page every time.
Support and updates — an EA needs maintenance as markets change. Check the seller is active and reachable, not vanished after the sale.
Refund terms — a clear, honoured refund policy signals some confidence. Read the conditions; "no refunds on digital goods" is common.
Prop-firm compatibility — if you plan to run it on a funded account, check it respects the firm's drawdown and news rules. Many EAs breach them instantly.
Cost — weigh the price against the account you will run it on. A £500 EA on a £1,000 account has to work miracles just to cover itself.

A quick worked example

Two EAs, both claiming 10% a month:

One shows a two-year verified live record, a worst drawdown of 18%, a named vendor and a clear grid strategy with stated risks.
The other shows a flawless backtest, no live results, an anonymous seller and "guaranteed" returns.

The first might suit you if you can stomach an 18% drawdown. The second is not a difficult decision: it is missing every piece of evidence that matters.

Red flags

Guaranteed returns — markets do not guarantee anything. The phrase alone is enough to walk away.
No drawdown, or a drawdown figure that seems impossibly small.
Backtests only, with no live account to back them up.
No verification — screenshots and a spreadsheet are not proof.
Pressure and secrecy — countdown timers, "limited copies", and a strategy the vendor will not describe.

Why it matters

An EA is one of the few purchases where the seller's interest and yours are opposed until you have paid. The checklist exists to move the decision off the sales copy and onto evidence. If a system passes it, forward test before you scale. If it fails the first line — no verified live record — nothing further down the page can rescue it.

Before you commit

Even a system that passes every line above has not yet proved itself to you. Run it on demo, then on a small live account, before it manages real size. A vendor confident in their EA will not object to you testing it for a month; one who pushes you to buy and deploy today is telling you something. The checklist filters out the obvious failures. Your own forward test is what turns a plausible EA into one you are willing to fund.

The Karnek note

Once you have chosen an EA, the work becomes watching it. Karnek keeps a verified live record from your terminal and alerts you when performance drifts from what you were sold, or a drawdown nears your limit. It reads read-only and can never place a trade or touch the account.

See it in Karnek: see how Karnek works, set up read-only monitoring in minutes.