How far have you verified the costs of backtesting?

When looking at backtest results, total profit and drawdown catch the eye. But have you also checked how the results incorporate trading costs?

While spreads and commissions may be small per trade, they accumulate with a high number of trades and can significantly influence performance for EAs with many transactions. This time, I will整理 the approach for checking backtest costs.

Spreads in MT5 Tester

In MT5 Strategy Tester, spreads are obtained from historical data and are treated as floating, i.e., always variable. In other words, test results are affected by what spreads were recorded in the historical data used.

Therefore, it is reassuring to check the following:

• What data was tested: Spreads recorded differ by brokerage and server
• Modeling method: The level of reproduction detail varies depending on whether actual ticks are used or something else
• Spread conditions during the test period: whether periods or times with wide spreads are included

Even the same EA can yield different results when tested with data from a different broker.

Fees

Depending on the account type, there are per-trade fees in addition to spreads. In MT5 Tester, there are features to use the current account settings or to configure fees independently.

It is important to confirm that the test conditions match the actual account you plan to use. If you test under no-fee conditions on an account that incurs fees, the results may look better than in reality.

Other test conditions

• Test period: Is it not too short, and does it include different market regimes
• Initial capital and lot size: Are they close to the actual planned conditions
• Order slippage: The tester has a setting for delay. It is worth testing under conditions close to the real environment
• Swap: For EAs that hold positions longer, you should consider the impact of swaps

Why verify costs

The more trades you have and the smaller the profit per trade, the more sensitive the results are to costs. For example, an EA with an average profit per trade of only a few pips can see its performance change significantly with a 1-pip spread Change.

One reason why backtest profit is positive but live results are negative is this drift in costs.

Cautions

Even with aligned tester conditions, real trading will not be exactly the same. Slippage from sudden price moves and moments when spreads widen briefly may not be fully reproducible. It is advisable to approach backtest results with some margin for error.

From the Institute

In EA research and validation, we check cost conditions before looking at results. When publishing, we intend to release only after thoroughly verifying these conditions.