Why am I getting different backtest results in optimization and backtest modes?

The underlying backtest engine is the same for optimization and backtest modes.

The only difference is in how the period of backtests (by that I mean the candle data) is used differently.

In backtest mode, you select starting date, and finishing date, knowing that your strategy will be backtested since exactly that date.

But in the optimization mode, you select the starting date, and the ending date, but behind the scenes, Jesse will divide them into a "training period" (85%) and a "testing period" (15%).