Orometra analyzes your MT5 optimization and tells you which configuration of your EA is stable and how much evidence backs it.
All you need is the XML file MT5 exports · How to export it →
Export your MT5 optimization results to XML — and the forward tab, if you used it.
They are analyzed in your browser, in seconds.
The recommended one with its reasons, how reliable it is, a .set ready for MT5 and the range to re-optimize around it.
What MT5 gives you (example)
MT5’s #1: Pass 1842The highest Result among thousands of passes
MT5 sorts configurations by the criterion you chose. With thousands tested, the ones at the top are often the least reliable: change one parameter a single step and they can fall apart.
What Orometra does
Recommended: Pass 4283#1 rejected: isolated peak
It looks for zones where neighboring values also work — a plateau. That is harder to get by chance, so it tends to hold up better.
This is the first thing the sample report shows: which configuration to pick, how reliable it is and what to do next.
Chosen within a plateau of 5,018 similar configurations.
The most the optimized period and the forward can support: a validated, stable plateau with no warnings about it. An unseen period can keep it there or lower it.
And when there is nothing? See an example with no real edge →
We ran the engine on 800 simulated optimizations where we knew in advance whether there was a real edge or not. Orometra rates the evidence on five levels, from “insufficient” to “strong”.
0.3%
Out of 300 optimizations with no real edge at all, only 1 got “moderate” or higher, and none got “strong”.
89%
Out of 38 optimizations with a clear real edge, 34 got “moderate” or higher.
What would take you hours in a spreadsheet and a lot of judgment — done in seconds, the same way every time.
Out go the passes that miss your minimums for profit, drawdown and trades.
It measures quality with objective metrics, not with what you optimized for.
Inside the plateau it picks what holds in both periods alongside its neighbors, not the peak.
What holds up in the forward moves up; what falls apart moves down.
It warns you if your result doesn't beat what pure chance would produce.
It checks that the recommendation doesn't depend on its own settings.
Sometimes its most valuable answer is “there is nothing here”. You do not have to take our word for it:
Every calculation the engine makes can be read on GitHub. No black box.
See the code →Step by step and with its limits: what it does and what it does not do.
Read the method →Open your browser’s developer tools, Network tab, while you analyze: you will only see the site’s own files and an anonymous usage count. Your files are not sent anywhere.
Privacy →Before going live, each version passes hundreds of automated checks, including runs on data where the right answer is known in advance.
An AI always has an answer. Orometra only gives you the one your data can defend.
| An AI chat | Orometra | |
|---|---|---|
| Same table, another day | It can change its answer depending on how you ask | Always the same answer |
| The method | Improvised in each conversation | Fixed, published and tested |
| The numbers | It can make up a figure without warning you | Every figure comes from your data and can be checked |
| When the data can't support a conclusion | It answers anyway, with full confidence | It tells you: insufficient evidence |
| Your strategy | Uploaded to another company's servers | Never leaves your computer |
| What you get | Text | A configuration, its evidence level and a .set for MT5 |
Use AI to understand the report. To decide which configuration you risk your money on, use a criterion that doesn't change. Read the full explanation →
Something else? Write to us at hello@orometra.com.
The XML that MT5 exports with your optimization results: right-click the table and export to XML. If you enabled forward, export that table too. The .opt file does not work. Step-by-step guide.
You can still analyze it, but the evidence will not go beyond “weak”: without data the optimization never saw, a robust plateau cannot be told apart from a lucky one. Next time, enable it in the tester settings (a quarter or a third of the period).
Yes. Since the genetic algorithm does not test every combination, plateaus are measured conservatively: the result may fall short, but it never invents support. The report tells you.
With any MetaTrader 5 EA, on any symbol and timeframe: Orometra does not read your code, only the results table. MetaTrader 4 is not supported for now.
Nothing: in this phase Orometra is free, with no sign-up and no card. If paid features ever come, they will be announced in advance.
Nothing leaves your computer: the analysis runs in your browser. We only count, anonymously, visits and how often each feature is used. Details in the privacy page.
No. It tells you how much your data supports the chosen configuration; no level means “trade it”. It does not predict the market or turn a weak EA into a good one, and if your backtest is optimistic (spread, history quality), so is the analysis.
It will give you a convincing explanation, different every time. Orometra is a fixed calculation: same files, same result; it knows how to say “insufficient”, and your parameters never leave your computer. Full comparison.
Drop your optimization here and see it in a minute. Nothing is uploaded.