EA Parameter Optimization: Genetic Algorithms vs Grid Search in MT4

Summary: Technical deep dive into EA parameter optimization methods. Genetic algorithm reduces runtime by 80% vs grid search. Includes overfitting detection and walk-forward validation code for MT4.




Parameter optimization is the most deceptive step in EA development. Grid search on 10 parameters with 100 steps each requires 10^200 backtests - impossible. Genetic Algorithms (GA) reduce this to practical iterations.

1. Grid Search Limitations
Standard MT4 Strategy Tester uses brute force. For 3 parameters (TakeProfit, StopLoss, TrailingStop) with 50 values each:
```cpp
// Pseudo: 125,000 backtests
for(int tp = 10; tp <= 500; tp += 10)
for(int sl = 10; sl <= 500; sl += 10)
for(int ts = 0; ts <= 100; ts += 10)
RunBacktest(tp, sl, ts);
```
Runtime: ~43 hours on typical hardware. GA achieves same in 2 hours.

2. Genetic Algorithm Implementation Logic
GA mimics natural selection: population size N=50, generations G=30, crossover rate 0.7, mutation rate 0.05.
```cpp
// MT4 GA fitness function example
double Fitness(double ¶ms[]) {
double tp = params[0], sl = params[1];
int ticket = OrderSend(Symbol(), OP_BUY, 0.1, Ask, 3, sl, tp, "GA_EA", 0, 0, Green);
// Return Sharpe ratio or profit factor
return GetProfitFactor();
}
```
Selection: tournament selection (size 3). Crossover: single-point at random index. Mutation: add Gaussian noise with sigma = 0.05 * range.

3. Overfitting Detection Formula
Forward Walk-Forward Validation (WFV): split data into in-sample (80%) and out-of-sample (20%). Accept parameters only if OOS performance degradation < 15%.
```
Degradation = (PF_is - PF_oos) / PF_is
Accept if Degradation < 0.15
```
where PF = Profit Factor.

4. Practical Optimization Workflow
Step 1: Run GA with 30 generations, population 50. Step 2: Select top 5 parameter sets. Step 3: Run walk-forward on each. Step 4: Retain only sets with degradation < 15%.

5. Code: Export GA-Optimized Parameters
```cpp
// Save best parameters to file
void SaveOptimalParams(double &best[]) {
int handle = FileOpen("OptimizedParams.dat", FILE_WRITE|FILE_BIN);
if(handle != INVALID_HANDLE) {
FileWriteArray(handle, best);
FileClose(handle);
}
}
```

Reference: Kaufman, P. J. (2019). "Trading Systems and Methods." Wiley. Chapter 12 - Optimization Techniques.