You want to start automated trading but don't know where to begin? This guide walks you through seven steps — from first principles to a running, tracked EA setup — including the mistakes that cost beginners the most money.
Step 1: Understand what algo trading is (and isn't)
In algo trading, a program — on MetaTrader an Expert Advisor (EA) — executes your trading strategy automatically: it analyses the market by fixed rules and opens, manages and closes positions without your involvement.
What an EA is: a disciplined, emotionless executor of a strategy that runs 24/5 and never breaks a rule.
What an EA is not: a money printer. An EA can only be as good as the strategy behind it. If the strategy has no statistical edge, the EA merely automates losing — faster and more reliably than any human.
Step 2: Set expectations and risk first
Before you invest a single dollar, decide on:
- Realistic returns: Credible EA portfolios deliver roughly 2–8% per month over the long run — with swings and losing months. Anything well above that almost always means hidden risk.
- Maximum drawdown: Decide upfront how much interim loss you can stomach (e.g. 20%). We explain what drawdown and profit factor actually mean separately.
- Risk per trade: 0.5–2% of the account is a common range. How to translate that into lot sizes is required knowledge.
- Risk capital only: Trade exclusively with money whose total loss would not change your life.
One red flag you should know from day one: grid and martingale EAs show smooth equity curves for months and then wipe accounts in days. These are exactly the EAs marketed most aggressively to beginners.
Step 3: Pick the right broker
EAs care about different things than manual trading: fast, stable execution, tight spreads plus fair commission, no EA restrictions and ideally a free VPS. For most algo strategies, an ECN/raw-spread account is the right account type — you'll find the candidates in our comparison of the best ECN brokers for algo trading.
All the providers we've tested — with scores for execution, cost and EA-readiness — are in our broker reviews. Pay attention to regulation (ASIC, FCA, CySEC): in a dispute, it decides whether you see your money again.
Step 4: Demo first, then small and live
Open a demo account first and run your setup for 4–8 weeks. You're testing the plumbing here, not profitability: Does the platform run stably? Does the EA open and close as expected? Does it survive news spikes?
Then move to a small live account. Demo execution is flattering — you only see real slippage, real spreads and requotes with real money. Only scale up once several live months match the backtest.
Step 5: Find and judge an EA
Whether it's a marketplace (MQL5.com), a direct vendor or your own development — the evaluation rules are identical:
- A verified live track record of at least 6–12 months (e.g. Myfxbook with a verified account) beats any backtest.
- Read backtests critically: A pretty backtest is easy to manufacture. What actually matters is in our guide on backtesting and forward testing.
- Understand the strategy: The vendor must be able to explain why the EA makes money. "Secret AI logic" is a red flag.
- Check the risk behaviour: Fixed stop-loss per trade? Or are losing positions scaled up (grid/martingale)?
The full checklist: What makes a good forex EA? And to understand the demo-to-live gap: Why EAs fail live.
Step 6: Set up a VPS
An EA only trades while the platform is running. On a home PC, updates, standby and Wi-Fi dropouts mean missed trades. A VPS (a virtual server in a data centre) keeps MT4/MT5 running 24/5 with low latency — for 5–30 USD/month, or free at many brokers above a certain volume.
The fundamentals are in our VPS for EAs guide; concrete providers in the VPS comparison.
Step 7: Track and optimise your portfolio
What you don't measure, you can't manage. Connect your accounts to our portfolio tracker — free for MT4/MT5. You'll see equity curves, drawdown and, above all, the correlation between your strategies: two EAs that lose in the same phases double your risk instead of diversifying it. Long term, a portfolio of several uncorrelated EAs beats any single star EA — how to build one is in our EA portfolio management guide.
The most common beginner mistakes
- Starting with too much capital — demo first, then small and live, then scale.
- Believing return promises — "guaranteed" 100% per year is marketing, not trading.
- Treating a backtest as the future — without forward testing and a live track record, a backtest is just a hypothesis.
- Underestimating grid/martingale — smooth curves right up to account death.
- Tearing everything up after three losing days — EAs need months to show their statistics. If you constantly intervene, you're trading manually with extra steps.
- Running without a VPS — missed trades distort any strategy.
- Betting everything on one EA — diversification across strategies and pairs is the only free lunch.
Verdict
Algo trading is a learnable craft, not a lottery ticket: set expectations, pick a broker with clean execution, test on demo, judge EAs by live data instead of promises, run them reliably on a VPS, and measure everything in the portfolio tracker. Follow these seven steps in this order and you're already ahead of most beginners in the one way that matters: you'll survive your first twelve months.