Stress the story
Spread/slippage assumptions, overfitting to one regime, and look-ahead bias are common. Ask what would break the curve before you ask how pretty it is.
Forward samples and prop-rule constraints matter more than a perfect historical line.