QuantForge
en

AI EA builder that shows its work

Describe your strategy in plain words, paste an article or YouTube link, or attach a PDF or a chart screenshot. The AI turns it into blocks on a canvas, so you can see every rule, backtest it, and export a real MetaTrader EA.

Start freeRead the docsFree plan, no card needed.

From any source to blocks you can check

  1. Give it the strategy. Type it, dictate it, paste a link (article, strategy page, RSS feed, YouTube video), or attach a PDF, a screenshot or a chart template.
  2. Read what it understood. The AI lists the rules it found and how it made vague points precise: “buy when the trend is up” becomes a concrete moving-average condition you can change.
  3. Apply it to the canvas. The proposal becomes blocks and wires in one step, which Ctrl+Z undoes. Unknown blocks, broken wires and loops are refused before anything changes.
  4. Backtest and export. Run it on your own M1 history, then export MQL4 or MQL5, or the compiled .ex4 / .ex5.

An assistant that knows your strategy

The assistant sees the blocks, inputs and the latest backtest summary of the strategy that’s open. Ask it to:

  • explain what the strategy does, step by step,
  • find weak points and suggest how to test them,
  • add a filter (“only trade during the London session”),
  • change risk (“risk 1% per trade with a 30-pip stop”),
  • fix an issue the canvas reports.

Why blocks instead of generated code

Code written by a language model looks right and often isn’t: invented functions, wrong order handling, silent off-by-one errors on bar indexes. In QuantForge the model proposes blocks, and the code comes from the same generator as every other QuantForge EA, which is compiled in MetaEditor 4 and 5 in our tests. What you backtest is exactly what you export.

The AI can be wrong. Backtest every change, and test on a demo account before trading live. New to the editor? Start with the no-code EA builder or a tested template.

Frequently asked questions

How is this different from asking a chatbot to write MQL code?

A chatbot writes MQL you have to trust, debug and compile yourself. Here the AI proposes blocks: every rule is visible on the canvas, wires and unknown blocks are checked before the proposal can be applied, and the code is generated by QuantForge's own tested code generator, so it compiles in MetaEditor. You backtest before you export.

What can I give the AI?

A description in your own words, a link to a web page, an RSS feed or a YouTube video, text you paste, or files: images such as chart screenshots, PDFs, Word (.docx) and HTML documents, and MetaTrader chart templates (.tpl). It lists the rules it found and how it made vague points precise.

Which plan includes the AI?

The full AI assistant (questions, explanations, changes, builds from videos and files) is part of Studio, with 2,000 credits a month and credit packs that never expire. Pro includes 3 strategy builds a month from a link or pasted text. See pricing.

Can the AI promise a profitable strategy?

No, and you should be wary of any tool that does. The AI turns rules into blocks and can suggest how to test weak points. Whether the strategy works is something only a backtest, a forward test and a demo account can tell you.

Is my strategy kept private?

Your projects are stored for your account only. When you use the assistant, your strategy and question are sent to the AI provider QuantForge uses to produce the answer. For links, only the link is kept in the conversation, not the page or the video. See the privacy policy.

Build your first EA today

The free plan includes three projects, backtests on H1 to D1 over a year, MQL4/MQL5 export and the converter.

Start free

Keep exploring