AI across the whole build — from idea to live bot
How to lean on your copilot at every stage of the rest of this course — research, backtesting, debugging, MQL5, deployment — with ready prompts you can adapt.
From here on, every part of this course pairs naturally with your AI copilot. This lesson is a map: where AI helps most in each stage, and a prompt you can lift and adapt. Keep it open as you go.
Research & indicators (Parts 3–4)
Implement the RSI indicator from scratch in pandas (no TA library),
input a 'close' Series, period 14. Explain the maths in comments.
Then show me how to VERIFY it matches a reference implementation on
one example so I trust it.Strategy design (Part 4)
Use AI to *pressure-test* ideas, not just generate them. It is a great sparring partner for "why might this edge not be real?"
I'm considering a mean-reversion strategy on gold that fades 2-sigma
moves. Before I code it: what is the economic reason this might work,
what market regime would destroy it, and what is the most likely way
I'd fool myself in the backtest? Be blunt.Backtesting & debugging (Parts 5–6)
Here is my backtest code and the FULL traceback:
[paste code]
[paste traceback]
Also here's df.head() and df.dtypes:
[paste]
Find the bug, explain the cause in one sentence, and give the minimal fix.Python → MQL5 translation (Part 9)
One of AI's best trading tricks: you research and validate a strategy in Python, then have the copilot draft the MQL5 Expert Advisor — which you then verify handler-by-handler against Part 9.
Translate this validated Python SMA-cross signal into an MQL5 Expert
Advisor for MT5: use iMA handles, act once per new bar, and place
orders with CTrade including a stop-loss. Add comments mapping each
MQL5 block back to the Python logic so I can check they match.Deployment & monitoring (Part 10)
Write a systemd unit file to run my Python trading bot (bot.py) as a
service on Ubuntu with automatic restart on failure and log rotation.
Explain each directive so I understand what I'm running.You now have the copilot mindset, the loop, the prompting recipe, and the verification checklist. Carry them into Part 3, where we set up Python and get real market data into your hands — with your AI copilot beside you.