QUESTION:
In MT5 Strategy Tester I'm using 2/3 backtest and 1/3 forward test. There's an optimize results for both backtest and forward test. Where should I choose my final optimal parameters? on backtest results? or on forward test results? or both?

Short Answer

Choose your final parameters based on the forward test performance — not the backtest.

But:

Only choose among parameter sets that passed the forward test AND were reasonably strong in the backtest.

Why?

1. Backtest = Where the EA learns

• Backtests are used for optimization, not for choosing final parameters.
• They are prone to curve-fitting, even with good metrics.
• The purpose of the 2/3 backtest stage is to find promising parameter sets, not to validate them.

2. Forward Test = The real filter

• Forward test simulates unseen, out-of-sample market conditions.
• It tells you which parameters generalize to new data rather than overfitting the past.
• Parameters that fail here should be discarded even if they look perfect in the backtest.

3. The correct logic is:

Step 1: Optimize on backtest

→ Identify clusters of robust parameter sets (not just the single best one).

Step 2: Evaluate those clusters in the forward test

→ Select the set(s) that perform well out of sample.

Step 3: Final selection

→ The final optimal parameters must be chosen only from the forward test winners.

How to choose concretely in MT5:

In Strategy Tester you have two result tabs:

• Optimization Results (Backtest 2/3)
• Forward Optimization Results (1/3)

You should:

1. Sort Forward Optimization Results by your key metric (Profit Factor, Sharpe, Drawdown, etc.).
2. Choose the top parameter sets from the forward test.
3. Confirm that they also performed reasonably well in the backtest (not necessarily #1, but consistent).

Do NOT pick parameters that are #1 only in backtest but fail in forward.

Professional Tip (This is what prop firms do):

Use the Intersection Approach:

The final parameter set must meet:

✔ Forward test performance ≥ your minimum thresholds

AND

✔ Backtest performance shows stability across nearby parameter variations (robustness)

This avoids curve-fitting and ensures long-term performance.

Final Recommendation:

Your final parameters should be chosen based on forward test performance, not backtest performance.

But only among parameter sets that were already strong in backtesting.