The Truth About EA Automated Trading: Why Most Traders End Up Returning to Manual

EA?

Traders' attitudes toward EA (Expert Advisor) are usually polarized: either they think EA is the holy grail, or they think it's a scam. The reality lies between the two.

Four legitimate motivations for EA:

1. Eliminating emotional interference: EA doesn't revenge trade, doesn't FOMO, and doesn't get overconfident
2. 24-hour operation: No need to watch the screen, ideal for part-time traders
3. Easy backtesting: Quickly verify a strategy's historical performance
4. Execution discipline: Won't abandon a plan because "something feels off"

But each motivation has its trap, and this article breaks them down one by one.

Three Main Types of EA

*Robots are the visual representation of EA automation—the opposite of emotional trading.*

1. Trend-Following

Follows the trend direction to enter, with wider stops, lower win rate but higher RR. The biggest weakness is consecutive losses in range-bound markets.

2. Grid

Places orders at fixed intervals, profiting from market back-and-forth swings. Performs well in ranging markets, but constantly adds positions in one-sided trends, which can ultimately blow up the account.

3. Martingale

Doubles the bet after a loss, hoping to recover in one go. This is the most dangerous type of EA, mathematically destined to go bankrupt (see the next article).

Curve Fitting: The Fatal Trap of EA

Overfitting (Curve Fitting) is the most common fatal mistake in EA development.

What is overfitting? You repeatedly tweak parameters on historical data until the backtest results look perfect. But this "perfection" is merely over-memorization of past data, with no predictive power for future data.

How to Identify Overfitting?

• Backtest win rate exceeds 80% (extremely rare in real markets)
• Slight parameter changes cause dramatic shifts in results
• Only performs well in specific time periods or specific market environments
• No Walk-Forward testing was done

Walk-Forward Testing

The correct approach is: use 70% of historical data to optimize parameters, then verify on the remaining 30%. If performance drops significantly during the validation period, your EA is overfitted.

Why Is Manual Trading Still the Mainstream?

1. Market Environments Are Constantly Changing

Trending, ranging, news-driven, holiday markets—no single EA can be effective in all environments. Manual traders can switch strategies based on environment; EA cannot.

2. Structural Breakouts Require Judgment

Identifying HH/HL/LH/LL, recognizing BOS/MSS, confirming wick rejections—these all require human visual judgment and experience, which current EAs cannot accurately simulate.

3. Risk Management Requires Flexibility

"VIX is spiking today, I should reduce position size" or "NFP data is coming out, I'll close my positions first"—these flexible risk management decisions cannot be automatically executed by EA.

4. The Value of Trading Psychology Is Underestimated

Many traders think EA can solve their psychological problems, but in reality, EA only delays the psychological challenge—when EA suffers consecutive losses, you'll face the test of "should I turn off the EA?", which is fundamentally still a discipline issue.

MQL5 Introduction

MQL5 is MT5's programming language, with syntax similar to C++. For traders with a programming background, the learning curve is not too steep.

There's now a growing trend: using AI (such as ChatGPT) to assist in writing MQL5 code. You can describe your strategy logic in natural language, let AI generate the initial code, and then manually debug it.

"Writing EA with AI is fast, but debugging takes forever."

AI-generated code usually requires extensive modification to run stably, especially the stop-loss logic and exception handling.

Conclusion on Automated Trading

Automation is neither a cure-all nor a scam. It is a tool, suitable for specific types of traders and specific types of strategies:

• Suitable for automation: Pure rule-based strategies (e.g., moving average crossovers, breakout strategies), strategies that need to run 24 hours
• Not suitable for automation: Strategies requiring subjective judgment (e.g., market structure analysis, candlestick pattern confirmation), strategies requiring flexible risk control

The most ideal model is semi-automation: use EA to execute entries and stops, but let humans decide when to turn the EA on or off.

Chapter Summary

1. EA is neither a holy grail nor a scam—it is a tool that needs to be used correctly
2. Overfitting is the number one killer of EA development
3. The flexibility and judgment of manual trading cannot be replaced by EA at present
4. Semi-automation (human decision-making + EA execution) is the most pragmatic approach