Move beyond “ask AI for a bot, paste the code and get stuck.” Learn a structured workflow for researching, engineering, testing and deploying an automated trading system—with enough understanding to evaluate the code and its behaviour.
We’ll combine MQL5, MetaTrader 5, VS Code and OpenAI Codex to turn a trading idea into a clear specification, build a modular experimental EA, investigate it in Strategy Tester and deploy a selected version to an MT5 demo environment on a VPS.
This four-week lab welcomes beginners, forex traders and existing EA developers. The goal is a repeatable engineering and research workflow, a working experimental system and a documented testing/deployment process. This is not a signals programme or a promise of trading profits.
Your four-week research-to-deployment plan
AI-Assisted Trading System Development
Set up MT5, MetaEditor, VS Code, suitable MQL5 tooling and a versioned project workspace. Learn effective Codex use, modular prompts, code review and the limits of one-shot generation. Turn the shared research idea into a technical specification: symbols, timeframes, ticks/candles, entry and exit rules, risk logic, filters and EA architecture. Practical work: install the tools, document assumptions and complete the system specification. Outcome: a functioning development environment and clear build plan.
Engineer the System Layer by Layer
Build the shared EA incrementally: market data, signal logic, filters, buy/sell execution, position management, risk controls, inputs and logging. Cover entry/exit rules, indicators, stop loss, take profit, position sizing, Magic Numbers, duplicate-trade prevention and state management. Use Codex on individual modules, review its output and debug behaviour instead of blindly accepting code. Practical work: compile, inspect logs and exercise order-management safeguards. Outcome: the first complete experimental EA version and its source code.
Backtest, Investigate and Iterate
Use MT5 Strategy Tester to investigate behaviour, trade histories and logs. Read drawdown, profit factor, expectancy and win rate in context; examine trading costs and why headline profitability can mislead. Find logical bugs, execution issues and edge cases with AI-assisted debugging. Compare a baseline with evidence-driven changes to entries, exits, risk management or filters; study parameter sensitivity and avoid overfitting. Practical work: record assumptions, tests and experimental versions. Outcome: a version comparison and a reasoned candidate for forward testing—not a profitability claim.
Deploy, Forward Test and Monitor
Select a candidate based on test evidence and prepare a controlled demo deployment. Cover VPS selection/setup, MT5 installation and configuration, demo-account connection, EA installation, autostart/restart considerations, monitoring, logs, updates and risk safeguards. Compare backtests with forward behaviour, maintain source versions and discuss evidence needed before any future live use. Practical work: build a deployment checklist and configure or follow the shared VPS/demo deployment. Outcome: the selected experimental EA running in a demo environment, with an approximately 30-day observation plan where practical.
Frequently asked questions
Is this a forex trading course?
It is a trading-system engineering and research lab. We use trading ideas to teach specification, EA development, testing and deployment. It is not a replacement for learning trading fundamentals.
Is this a signals course?
No. We build and investigate automated systems; we do not provide trade signals or tell you which trades to take.
Do I need programming experience?
No prior programming experience is mandatory. You should be comfortable installing software, following a practical workflow and learning between sessions.
Do I need MQL5 experience?
No. We introduce the MQL5 concepts needed for the shared system and explain how the parts fit together.
I already build EAs. Is this still useful?
Yes, if you want a more structured AI-assisted workflow: clear specifications, modular changes, code review, logs, version comparisons, testing and controlled deployment.
I already use ChatGPT to generate bots. What is different?
We move beyond one large prompt and copied code. You will define rules first, build in modules, inspect AI output, diagnose behaviour and use experiment evidence to guide each version.
Will we use Codex?
Yes. OpenAI Codex is part of the planning, modular development, review and debugging workflow. We will clarify the tool access needed for the practical setup.
Will we use VS Code?
Yes. VS Code is part of the working environment, alongside MT5 and MetaEditor where needed.
Will we build a real EA?
Yes. We work progressively toward a functioning experimental MQL5 Expert Advisor. A bot that runs is not automatically a strategy with a proven edge.
Can I bring my own strategy?
Yes, bring it for discussion and specification practice. Live development follows a shared project so everyone can follow the same architecture and experiments; individual custom bots are not guaranteed.
What strategy will the cohort work on?
The shared strategy or research idea will be selected and clearly specified in Week 1. Its rules and assumptions will be documented before development; it is not advertised as a secret or proven profitable strategy.
Will the strategy be profitable?
We do not know in advance. We investigate results and limitations, including trading costs, drawdown and sensitivity. An experiment can reject an idea, and that is a useful research outcome.
Are profits guaranteed?
No. There is no promise of profits, returns, passive income or a winning bot. The programme teaches engineering, testing, deployment and iterative improvement.
Will we backtest?
Yes. Week 3 uses MT5 Strategy Tester, logs, trade histories and version comparisons to investigate behaviour and avoid misleading conclusions or overfitting.
Will we forward test?
Yes. The selected candidate moves to a demo environment. Where practical, observation continues for approximately 30 days after the teaching period.
Will we use a VPS?
Yes. Week 4 covers VPS setup, MT5 configuration, demo deployment, restarts, monitoring and updates. Any optional personal VPS costs will be clarified before setup.
Do I need a live trading account?
No. A live-funded account is not required, and paying for this cohort does not require funding a trading account.
Can we use demo accounts?
Yes. Demo environments are used for deployment and forward observation. Demo results do not guarantee equivalent live results.
Can I participate using only a phone?
A phone can help with messages, but the practical development, backtesting and deployment require a laptop or desktop. Phone-only participation is not suitable for this lab.
What kind of computer do I need?
A laptop or desktop capable of running MT5 and VS Code, with reliable internet. Windows is the straightforward setup. If you use another operating system, ask about installation options before paying.
Will I get the source code?
Yes. The shared cohort project source and experimental versions are part of the practical build. You will learn how to maintain and compare those versions.
What happens between live sessions?
Complete research and development tasks, refine the specification, build or debug modules, run tests and record findings. Bring your blockers and questions to the next approximately two-hour weekly session.
What happens after Week 4?
Where practical, the selected demo system continues for approximately one month of observation. We study behaviour, bugs, drawdowns and differences from backtests. This is an observation stage, not a promise of extra weekly teaching or profitable results.