Can You Trust an EA You Can't Read?

Most commercial EAs are distributed as a compiled .ex5 (MT5) or .ex4 (MT4). These are binaries, so you cannot read the logic inside. Is it averaging down? Is there a hidden martingale? What conditions does it enter on? As long as the source is closed, the code can't tell you.

So how are you supposed to trust an EA you can't see inside?

Why closed source is normal

First, a premise: keeping the source closed is not, in itself, shady.

• An EA's logic is the developer's intellectual property; publish it and it is trivially copied.
• If you sell it, protecting the internals is simple economic sense.

So we too distribute our paid products as compiled .ex5 / .ex4. "Closed source = suspicious" is not the reality. The real issue is that since you can't read the code, trust has to be established by something other than the code.

If you can't read it, judge by verification

If you can't read the source, judge the EA not by what it does internally but by what it did — that is, by external verification data.

1. Real-tick backtest: is the model real ticks (the real-ticks article)? Hidden thin-margin logic is exposed on real ticks.
2. Trade count: 500+ across multiple regimes (the sample-size article).
3. Equity drawdown: equity, not balance. Averaging and grid reveal themselves in equity DD (the drawdown article).
4. Year-by-year P&L: not concentrated in one year.
5. Third-party-verified live record: a reinforcing signal if present (the track-record article).

Even without reading the code, if this external data is all present the EA's true nature is fairly visible. Conversely, an EA that discloses none of it and just says "the internals are secret, but it's profitable" offers nothing to back the trust.

The role an open-source free EA plays

Here an open-source (source-published) free EA plays a distinct role.

Our free EA (a closed-bar Donchian breakout) is fully open-source. That is the opposite of our paid policy, and it is deliberate.

• You can confirm the developer's discipline directly in the code: whether a hard stop is really on every trade, whether any averaging or martingale is hidden — reading the code makes it obvious at a glance.
• It gives you a basis to decide whether to trust the same developer's paid (closed) products: the free EA lets you check "is this developer honest at the code level?"

In other words, an open-source free EA works as a transparency sample for judging whether to trust the closed paid products. Show the discipline in a free EA you can read; back the paid products with verification data — that is our transparency design.

Check it yourself

The free EA is open-source, so you can open the code and directly confirm "is a hard stop on every trade?" and "is there any averaging, grid or martingale?" Use it as one of the few EAs where you can read the logic itself, not just the backtest numbers.

Donchian Trend Engine (free, open-source)

Our paid products are closed, but in exchange we publish their real ticks, trade count, equity DD and year-by-year P&L in full on the verification page.

Summary

• Most commercial EAs ship as .ex5 / .ex4; the logic can't be read.
• Closed source is legitimate IP protection (our paid products are closed too).
• Since you can't read it, judge by external verification (real ticks, trade count, equity DD, year-by-year, live record).
• An EA that discloses no verification data has nothing backing the trust.
• An open-source free EA is a "transparency sample" that shows the developer's discipline in code.

Related: Backtest profitable but losing live · How to tell if a live track record is real · How to spot scam EAs and hype