Your account at a glance.
Paste the numeric Order Number from your mql5.com Payments tab. Admin reviews submissions on the console and approves your account directly — no license key needed. Your dashboard auto-upgrades to EA Owner once approved, usually within 24 hours.
Direct sales come with a 32-character license key sent by the seller. Paste it below to unlock EA Owner access on this account. MQL5 buyers do not need this step — your approval on the console grants access automatically.
Quant AI Agents is the Expert Advisor itself — five AI quants run a structured debate on every signal before a single order is placed: the Analyst gathers evidence, Bull and Bear argue both sides, the Judge weighs the case, the Decision sizes the order. No agent acts alone; the trade only goes through when the team agrees. You bring your own API key, a four-layer risk engine runs locally to defend the account even if the AI provider goes down, and you can rewrite the entire strategy by editing one text file. The two panels below are illustrative; the sections under them are exactly what ships in the box.
Quant AI Agents is a fully automated framework — it opens, manages and closes every trade on MT5 with no clicks from you. What's different is what decides the trade: your own plain-English instructions, not a single strategy the developer locked in. Think of it as your own trading firm — five trained specialists you brief in your own words, powered by an API key you bring. One EA, hundreds of possible strategies.
Locked to one strategy the developer hard-coded. You can't change how a standard EA thinks without writing complex code — so you're a passenger, not the driver.
Your instructions are the strategy. Brief the desk in plain English and the same five-agent engine trades it — automatically, while strictly protecting your capital.
Not just a tool, but a completely new way to interact with the markets: you finally have the power to command a team of AI experts using just your own words — and you decide how the account trades.
Every signal walks the same five steps. No single agent gets to act on its own — that's the whole point.
Pulls candles from the chart and a higher timeframe, the indicators on it, swing highs/lows, and the MT5 native news calendar. No AI yet — just evidence.
Picks the 3-8 indicators that matter, names the regime — trend, range or volatile — and writes one invalidation condition.
Bull and Bear each present evidence, then reply to the other's objection and rate their own confidence. Both sides on the record.
Weighs the case against memory of past trades, then returns a verdict: BUY · OVERWEIGHT · HOLD · UNDERWEIGHT · SELL.
Sizes the order, sets the stop, runs a final regime check. Low confidence → it downgrades to HOLD and the trade doesn't go.
A small quant desk argues each trade in plain English — and you can read every word on the chart panel.
Nothing hides behind a black-box score. Open the on-chart panel's Debate tab and read each agent's argument, objection and confidence for the exact trade you're looking at.
The agents pick the trade; this engine keeps the account alive. Every layer runs inside the EA on your terminal, with no AI call in the loop.
Daily, weekly and monthly loss caps plus a hard equity floor. Hit the floor and every position closes and auto-trade switches off — no "are you sure" dialog.
The AI sets the initial stop; local code runs every tick: break-even, vol-aware trailing, partial close at an R-multiple, time-based BE, and emergency flat on a volatility spike.
Reads the MT5 native calendar — no Forex Factory plugin. High-impact events on the traded currencies auto-pause new entries inside a window you set.
Every closed trade is written to a memory file. Next run the Judge reads it — your losing patterns become Bear arguments in next week's debate. No fine-tuning.
Also under the hood: loss-streak damper (5 losses in a row halves risk), daily profit lock (+2% locks the day green), loss cooldown, session filter, multi-chart pipeline lock, randomized startup delay, broker-spec validator, and symbol normalization across 47+ broker variants.
This isn't a black-box bot — it's a programmable engine. Want breakouts today and mean-reversion tomorrow? Don't touch the code. If you can type, you can build a quant EA.
Open eatrading_prompts.txt and write a [PRESET] block — six plain fields: NAME, SUMMARY, TIMEFRAME, WHITELIST, STRATEGY, RULES.
One click of the ↻ button on the panel — no recompile, no restart, no license reset, no support ticket.
The same five agents start trading your new rules at once. Three archetypes ship built-in: Trend Breakout H4, Mean Reversion D1, Volatility Compression H4.
Or skip writing your own: the Prompt Library ships 99 ready-to-load strategies — filter by indicator, timeframe or symbol, copy, reload.
Free and Pro share the same five-agent engine and the same prompt library. Free shows you the verdict; Pro lets the EA act on it.
The indicator-only edition for newcomers. The five agents run and show their verdict on the panel — you click the button. $0 forever, no card.
Full auto-execution: when the agents agree, the EA opens, manages and closes the trade and runs the four-layer risk engine end to end.
One AI risk layer sits above your entire MT5 terminal — your own Quant AI robots and any third-party EA alike. Every cycle it reads the live market regime, scores how well each robot fits it, then runs, scales, pauses or partially closes them before drawdown bites — and you can ask it, in plain language, what's going on. The two panels below are illustrative; the sections under them are exactly what ships in the box.
Every hour the Supervisor runs the same four-step loop — cheap, auditable, boring on purpose.
Read every open trade across every robot — account size, position size, profit so far, room left before your cap. No AI yet, just inventory.
Trending up? Range-bound? News-driven chop? An AI labels the live regime with a confidence number.
Per robot, 0.0–1.0: how well does this robot's playbook fit today's regime? Anything below your floor is on the chopping block.
Five commands: RUN · SCALE 0.5× · PAUSE · CLOSE 50% · TIGHTEN STOP. Each carries a written reason and is rate-limited so it can't oscillate.
How the Supervisor stops a robot depends on whether the robot can listen. Both modes ship in the box.
The Quant AI family checks a shared message every second. The Supervisor writes "pause" / "scale to 0.5×" / "go full" and the robot obeys on its next tick. No order placed.
Third-party robots can't be asked nicely — we don't have their source. So the Supervisor uses MT5 itself to close part of the position and pull the stop in tighter. The robot wakes up to a smaller trade.
The Supervisor installs in Watch mode. You see what it would do before any move is real — and you decide when it earns the trigger.
Every cycle: full decision + written reason logged. Nothing executes. Read its mind without paying for its mistakes.
The Supervisor proposes; moves wait in a queue; you click ✓ to fire. Train your gut on its instincts before trusting it solo.
Moves fire automatically, each with a logged reason and one-click rollback. Big actions still need two AI calls to agree.
Drop back to Watch any time, promote one robot at a time — the Supervisor never auto-escalates above your current setting.
A small team — analyst, bull, bear, risk officer, portfolio manager. Each does one job; every output is written down.
Every agent's full input and output sits in supervisor_log.txt in your MT5 data folder — nothing hidden, nothing redacted, nothing sent to anyone but your own LLM provider.
An AI risk guardian that goes rogue is worse than none. These guardrails ship on by default — each tested under broken data, broken network, broken AI.
New installs start read-only — watching and logging without touching your broker. Going Live is one deliberate toggle; dropping back is just as easy.
Close-all, or any single move above 20%, needs two AI calls to agree. One agent saying "do it" never fires.
Repeated API errors or junk responses lock new entries, keep existing trades on their own stops, and alert you. Quiet beats wrong.
At most one intervention per robot per 30 minutes (you set the number). No flip-flopping a robot on and off.
Pings itself every 60s. Two missed pings and robots revert to their own defaults — never stuck on "Supervisor said pause".
One Supervisor per terminal, per account, per chart — each namespaces its state and logs by its own magic ID. No two AIs fighting over one robot.
Plain English → structured [PRESET]. The local heuristic answers instantly; anything ambiguous falls through to a DeepSeek backstop. Owning the EA unlocks the generator automatically.
← Describe your idea and hit Generate — a structured prompt appears here.
You own the LLM bill. Dial the knobs — pipeline mode, debate rounds, update interval, provider — and see the real monthly cost. Math derived from the EA's spec; provider prices are Jan 2026 published rates.
Defaults match the EA's recommended Full pipeline at 60-minute cadence.
Bull and Bear each speak this many times. Only applies in Full mode.
Covers all 15 models exposed in the EA's ENUM_AI_MODEL input panel — GPT-5.5 / 5.4 / 5.4-mini, legacy GPT-5 + GPT-4o, Claude Sonnet & Opus 4.6, Gemini 3 Pro & Flash, DeepSeek Chat & Reasoner, Grok 4 & Grok 4 Fast. The EA logs token usage so you can verify against your provider's bill.
The same formulas the Supervisor runs at every cycle — wired to sliders you can poke. See if your setup survives a bad day, and watch a regime switch play out across a basket of robots in real time. Estimating AI cost? See How much API?
Bad day hits a basket of robots. Does Supervisor save the account?
Pick the day's weather. Watch the basket react.
The Supervisor runs on your own AI key (BYOK) — billed by your provider, never through us. Pick a model, set how many robots it watches and how often it checks. We do the token math so you see the real monthly bill before you start.
Pick model + check interval. We do the token math.
Billed by your provider directly. We never see your traffic or take a cut. Built-in budget cap auto-downgrades the model if daily spend exceeds your limit.
The Prompt Generator is included with the EA. Sign in or own the EA to unlock — if your purchase is already activated, the generator will unlock automatically.
Already paid but still locked? Ask @Thangforex_Bot for manual unlock.
Loading…