We are very mindful of overfitting our models, and very skeptical of backtests on their own. Successful strategies often have some fundamental theory about why

We are very mindful of overfitting our models, and very skeptical of backtests on their own. Successful strategies often have some fundamental theory about why they should work. If not, the backtest needs to be very robust, and demonstrate that it is not overfit by cross-validation, and other methods.