Backtests

Backtests explain where the predictions come from. Each model is replayed against historical windows after modeled fees and slippage, then sampled from different entry conditions: buying near a local peak, buying after a dip, buying from a middle-of-the-road point, and rolling starts across broader history.

This is the bridge from backtest to prediction. A strategy that only works from dip entries but fails from peak entries may still look attractive in a simple window return, but it is less reliable as a model for a user buying today.